This page intentionally left blank

Top Predators in Marine Ecosystems

Their Role in Monitoring and Management

The sustainable exploitation of the marine environment depends upon our capacity to
develop systems of management with predictable outcomes. Unfortunately, marine
ecosystems are highly dynamic and this property could conflict with the objective of
sustainable exploitation. This book investigates the theory that the population and
behavioural dynamics of predators at the upper end of marine food chains can be used
to assist with management. Since these species integrate the dynamics of marine
ecosystems across a wide range of spatial and temporal scales, they offer new sources
of information that can be formally used in setting management objectives. This book
examines the current advances in the understanding of the ecology of marine
predators and will investigate how information from these species could be used in
management.

I A N B O Y D is Director of the Sea Mammal Research Unit at the University of

St Andrews. He is a Fellow of the Royal Society of Edinburgh and a recipient of the
Bruce Medal and the Scientific Medal of the Zoological Society of London for his
scientific studies in Antarctica.

S A RA H W A N L E S S , of the NERC Centre for Ecology and Hydrology, works on

long-term studies of bird populations.

C . J . C A M P H U Y S E N ’ S current research interests include: foraging ecology,

mortality and distribution patterns of seabirds in the Atlantic Ocean and in the North
Sea; the impacts of fishing on marine birds, and the spatial distribution and temporal
trends in abundance of cetaceans in the North Sea.
 Top Predators in Marine
Ecosystems
Their Role in Monitoring and Management

Edited by
i . l . b o y d , s . w a n l e s s a n d c . j . ca m p h u y s e n
  
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press

The Edinburgh Building, Cambridge  , UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521847735

© The Zoological Society of London 2006

This publication is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2006

- ---- eBook (NetLibrary)

- --- eBook (NetLibrary)

- ---- hardback

- --- hardback

- ---- paperback

- --- paperback

Cambridge University Press has no responsibility for the persistence or accuracy of s
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
 Contents

List of contributors page [viii]

Preface [xiii]

1 Introduction [1]
i . l . b o y d , s . w a n l e s s a n d c . j . ca m p h u y s e n

2 Effects of fisheries on ecosystems: just another top predator? [11]

a . w. t r i t e s , v . c h r i s t e n s e n a n d d . p au l y

3 Physical forcing in the southwest Atlantic: ecosystem control [28]

p . n . t r a t h a n , e . j . m u r p h y , j . f o r ca d a ,
j. p. croxall, k. reid and s. e. thorpe

4 The use of biologically meaningful oceanographic indices to separate

the effects of climate and fisheries on seabird breeding success [46]
b. e. scott, j. sharples, s. wanless, o. n. ross,
m . f r e d e r i k s e n a n d f . d au n t

5 Linking predator foraging behaviour and diet with variability in

continental shelf ecosystems: grey seals of eastern Canada [63]
w. d . b o w e n , c . a . be c k , s . j . i v e r s o n ,
d . au s t i n a n d j . i . m c m i l l a n

6 Distribution and foraging interactions of seabirds and marine

mammals in the North Sea: multispecies foraging assemblages and
habitat-specific feeding strategies [82]
c . j . ca m p h u y s e n , b . e . s c o t t a n d s . w a n l e s s
vi Contents

7 Spatial and temporal variation in the diets of polar bears across the
Canadian Arctic: indicators of changes in prey populations and
environment [98]
s. j. iverson, i. stirling and s. l. c. lang

8 Biophysical influences on seabird trophic assessments [118]

w. a . m o n t e v e c c h i , s . g a r t h e a n d
g. k. davoren

9 Consequences of prey distribution for the foraging behaviour of top

predators [131]
i. j. staniland, p. trathan and a. r. martin

10 Identifying drivers of change: did fisheries play a role in the spread of

North Atlantic fulmars? [143]
p. m. thompson

11 Monitoring predator–prey interactions using multiple predator

species: the South Georgia experience [157]
j. p. croxall

12 Impacts of oceanography on the foraging dynamics of seabirds

in the North Sea [177]
f . d au n t , s . w a n l e s s , g . p e t e r s , s . be n v e n u t i ,
j . s h a r p l e s , d . g r ém i l l e t a n d b . s c o t t

13 Foraging energetics of North Sea birds confronted with fluctuating

prey availability [191]
m . r . e n s t i p p , f . d au n t , s . w a n l e s s , e . m . h u m p h r e y s ,
k . c . h a m e r , s . be n v e n u t i a n d d . g r ém i l l e t

14 How many fish should we leave in the sea for seabirds and marine
mammals? [211]
r . w. f u r n e s s

15 Does the prohibition of industrial fishing for sandeels have

any impact on local gadoid populations? [223]
s. p. r. greenstreet
 Contents vii

16 Use of gannets to monitor prey availability in the northeast Atlantic

Ocean: colony size, diet and foraging behaviour [236]
k. c. hamer, s. lewis, s. wanless, r. a. phillips,
t. n. sherratt, e. m. humphreys, j. hennicke and
s. garthe

17 Population dynamics of Antarctic krill Euphausia superba at South

Georgia: sampling with predators provides new insights [249]
k. reid, e. j. murphy, j. p. croxall and p. n. trathan

18 The functional response of generalist predators and its implications for

the monitoring of marine ecosystems [262]
c . a s s e bu r g , j . h a r w o o d , j . m a t t h i o p o u l o s a n d
s. smout

19 The method of multiple hypotheses and the decline of Steller sea lions
in western Alaska [275]
n. wolf, j. melbourne and m. mangel

20 Modelling the behaviour of individuals and groups of animals

foraging in heterogeneous environments [294]
j. g. ollason, j. m. yearsley, k. liu and n. ren

21 The Scenario Barents Sea study: a case of minimal realistic modelling

to compare management strategies for marine ecosystems [310]
t. schweder

22 Setting management goals using information from predators [324]

a. j. constable

23 Marine reserves and higher predators [347]

s. k. hooker

24 Marine management: can objectives be set for marine

top predators? [361]
m. l. tasker

Index [370]
 Contributors

C. Asseburg C. J. Camphuysen
Centre for Conservation Science Royal Netherlands Institute for Sea
University of St Andrews Research
St Andrews PO Box 59
Fife KY16 9LZ, UK 1790 AB Den Burg, Texel, the Netherlands

D. Austin V. Christensen
Dalhousie University Fisheries Centre
Halifax, Nova Scotia University of British Columbia
Canada B3H 4J1 Vancouver, British Columbia
Canada V6T 1Z4
C. A. Beck
A. J. Constable
Alaska Department of Fish and Game
Australian Antarctic Division
Division of Wildlife Conservation
Australian Department of Environment and
Marine Mammals Section
Heritage
Anchorage
203 Channel Highway, Kingston
Alaska 99518
Tasmania 7050, Australia
W. D. Bowen J. P. Croxall
Marine Fish Division British Antarctic Survey, Natural
Bedford Institute of Oceanography Environment Research Council
Department of Fisheries and Oceans High Cross, Madingley Road
Dartmouth, Nova Scotia Cambridge CB3 0ET, UK
Canada B2Y 1A
F. Daunt
Dalhousie University NERC Centre for Ecology and Hydrology
Halifax, Nova Scotia Banchory Research Station
Canada B3H 4J1 Hill of Brathens
I. L. Boyd Banchory AB31 4BW, UK
Sea Mammal Research Unit G. K. Davoren
Gatty Marine Laboratory Zoology Department
University of St Andrews University of Manitoba, Winnipeg
St Andrews Manitoba
Fife KY16 8LB, UK Canada R3T 2N2
S. Benvenuti M. R. Enstipp
Dipartimento di Etologia Centre d’Ecologie et Physiologie
Ecologia ed Evoluzione Energétiques
Università di Pisa, Via Volta 6 Centre National de la Recherche
I-56126 Pisa, Italy Scientifique
 Contributors ix

23 rue Becquerel S. K. Hooker

F-67087 Strasbourg Cedex 2, France Sea Mammal Research Unit
J. Forcada Gatty Marine Laboratory
British Antarctic Survey, Natural University of St Andrews
Environment Research Council St Andrews
High Cross, Madingley Road Fife KY16 8LB, UK
Cambridge CB3 0ET, UK E. M. Humphreys
M. Frederiksen Earth Biosphere Institute and School of
NERC Centre for Ecology and Hydrology Biology, Ecology and Evolution Group
Banchory Research Station University of Leeds
Hill of Brathens Leeds LS2 9JT, UK
Banchory AB31 4BW, UK
S. J. Iverson
R. W. Furness Dalhousie University
Institute of Biomedical and Life Sciences Halifax, Nova Scotia
Graham Kerr Building Canada B3H 4J1
University of Glasgow
Glasgow C12 8QQ, UK S. L. C. Lang
Dalhousie University
S. Garthe
Halifax, Nova Scotia
Centre for Research and Technology
Canada B3H 4J1
Westkuste
University of Kiel Hafentörn S. Lewis
D-25761 Büsum, Germany NERC Centre for Ecology and Hydrology
S. P. R. Greenstreet Banchory Research Station
Fisheries Research Services Hill of Brathens
Marine Laboratory, PO Box 101 Banchory AB31 4BW, UK
Victoria Road
K. Liu
Aberdeen AB11 9DB, UK
Culterty Field Station
D. Grémillet The School of Biological Sciences
Centre d’Ecologie et Physiologie The College of Life Sciences and Medicine
Energétiques University of Aberdeen
Centre National de la Recherche Newburgh
Scientifique Ellon AB41 6AA, UK
23 rue Becquerel
F-67087 Strasbourg Cedex 2, France J. I. McMillan
Marine Fish Division
K. C. Hamer Bedford Institute of Oceanography
Earth Biosphere Institute and School of Department of Fisheries and Oceans
Biology, Ecology and Evolution Group Dartmouth, Nova Scotia
University of Leeds Canada B2Y 1A
Leeds LS2 9JT, UK
J. Harwood M. Mangel
Centre for Conservation Science and Sea Center for Stock Assessment Research
Mammal Research Unit University of California Santa Cruz
University of St Andrews 1156 High Street
St Andrews California 95064, USA
Fife KY16 8LB, UK A. R. Martin
J. Hennicke British Antarctic Survey, Natural
Zoological Institute and Museum Environment Research Council
University of Hamburg High Cross, Madingley Road
Hamburg, D-20146 Germany Cambridge CB3 0ET, UK
x Contributors

J. Matthiopoulos High Cross, Madingley Road

Sea Mammal Research Unit and Centre for Cambridge CB3 0ET, UK
Research into Ecological and
K. Reid
Environmental Modelling
British Antarctic Survey, Natural
University of St Andrews
Environment Research Council
St Andrews
High Cross, Madingley Road
Fife KY16 8LB, UK
Cambridge CB3 0ET, UK
J. Melbourne
Department of Astronomy and N. Ren
Astrophysics Oceanlab
University of California Santa Cruz The School of Biological Sciences
1156 High Street The College of Life Sciences and Medicine
California 95064, USA University of Aberdeen
Newburgh
W. A. Montevecchi Ellon AB41 6AA, UK
Cognitive and Behavioural Ecology
Program O. N. Ross
Memorial University St John’s University of Southampton
Newfoundland School of Ocean and Earth Sciences
Canada A1B 3X9 Southampton Oceanography Centre
Empress Dock
E. J. Murphy
Southampton S14 3ZH, UK
British Antarctic Survey, Natural
Environment Research Council T. Schweder
High Cross, Madingley Road Norwegian Computing Center
Cambridge CB3 0ET, UK Box 114 Blindern
J. G. Ollason 0314 Oslo, Norway
Oceanlab Department of Economics
School of Biological Sciences University of Oslo
The College of Life Sciences and Medicine Box 1095 Blindern
University of Aberdeen 0317 Oslo, Norway
Newburgh
B. E. Scott
Ellon AB41 6AA, UK
Department of Zoology
D. Pauly School of Biological Sciences
Fisheries Centre University of Aberdeen
University of British Columbia Tillydrone Avenue
Vancouver, British Columbia Aberdeen AB24 2TZ, UK
Canada V6T 1Z4
J. Sharples
G. Peters
Proudman Oceanographic Laboratory
Centre d’Ecologie et Physiologie
Bidston Observatory
Energétiques
Birkenhead CH43 7RA, UK
Centre National de la Recherche
Scientifique T. N. Sherratt
23 rue Becquerel Department of Biology
67087 Strasbourg Cedex 2, France Carleton University
Earth and Ocean Technologies 1125 Colonel By Drive, Ottowa
Hasseer Str 75 Ontario K1S 5B6, Canada
24113 Kiel, Germany S. Smout
R. A. Phillips Centre for Conservation Science and Sea
British Antarctic Survey, Natural Mammal Research Unit
Environment Research Council University of St Andrews
 Contributors xi

St Andrews A. W. Trites
Fife KY16 8LB, UK Fisheries Centre
University of British Columbia
I. J. Staniland
Vancouver, British Columbia
British Antarctic Survey, Natural
Canada V6T 1Z4
Environment Research Council
High Cross, Madingley Road S. Wanless
Cambridge CB3 0ET, UK NERC Centre for Ecology and Hydrology
I. Stirling Banchory Research Station
Canadian Wildlife Service Hill of Brathens
Edmonton, Alberta Banchory AB31 4BW, UK
Canada T6H 3S5
N. Wolf
P. M. Thompson MRAG Americas
Lighthouse Field Station 110 South Hoover Boulevard, Suite 212
School of Biological Sciences Tampa, Florida 33609, USA
University of Aberdeen Center for Stock Assessment Research
Cromarty IV11 8YJ, UK University of California Santa Cruz
S. E. Thorpe 1156 High Street
British Antarctic Survey, Natural California 95064, USA
Environment Research Council
J. M. Yearsley
High Cross, Madingley Road
Culterty Field Station
Cambridge CB3 0ET, UK
The School of Biological Sciences
P. N. Trathan The College of Life Sciences and
British Antarctic Survey, Natural Medicine
Environment Research Council University of Aberdeen
High Cross, Madingley Road Newburgh
Cambridge CB3 0ET, UK Ellon AB41 6AA, UK
 Preface

This book began its evolution in 1999 when the British Antarctic Survey,
where I worked at the time, began a new research programme on the man-
agement of marine ecosystems. This programme concentrated upon the
krill-based ecosystem at South Georgia which has been the subject of almost
continuous study since the Discovery Expeditions in the 1920s. Latterly,
international efforts to understand the dynamics of this ecosystem and the
wider Southern Ocean have been coordinated by the Commission for the
Conservation of Antarctic Marine Living Resources (CCAMLR). The daunt-
ing task of describing ecosystem dynamics over such a large oceanic area
with relatively limited resources led to the establishment of the CCAMLR
Ecosystem Monitoring Programme, an internationally coordinated effort at
data collection. Among other things, this contained a major component of
monitoring the seal and seabird populations in the region. The logic for
their inclusion was that they foraged over most of the regions of interest
but returned to breed at very well defined locations. By undertaking a series
of measurements of these predators at these locations, it was then argued
that aspects of the ecosystem dynamics should be reflected by variability in
the measurements of the predators. It was hoped that appropriate choices
of the predators and measurement variables would provide indicators of the
dynamics of their prey at different spatial and temporal scales.
The same concept has been developed in parallel within other ecosys-
tems during the past 20 years. The North Sea, California Current, northwest
Atlantic, Bering Sea, Gulf of Alaska and Barents Sea are regions in which
long-term monitoring studies of seabirds and seals are recognized as pro-
viding insights into ecosystem processes that can then be fed into the pro-
cess of management. Even though the implementation and use of mea-
surements has differed between regions, there has been a strong recogni-
tion that the interpretation of data about predator dynamics in the context
of ecosystem dynamics can only be achieved on the back of basic research
into the ecology of the species concerned. This book is, therefore, an effort
xiv Preface

to synthesize across a range of studies that have examined the ecology of

predators within the context of ecosystem approaches to management.
It is well recognised that people cannot manage ecosystems but can
only manage their own activities within ecosystems. The concerns about
the impacts of human activities upon ecosystems made this an appropri-
ate subject for a symposium sponsored and hosted by the Zoological Soci-
ety of London, and this took place in April 2004. At the same time, there
was an opportunity to build upon two major programmes of research: one
involving the Southern Ocean predators, mainly of krill, and being led by
researchers at the British Antarctic Survey, and one on North Sea preda-
tors, mainly of sandeels, being undertaken by a consortium of researchers
under the IMPRESS programme. The content of the book therefore reflects
the interest in these two contrasting ecosystems but also includes represen-
tations from other ecosystems.
Production of this book would not have been possible without the inter-
est and willing participation of the authors of each of the chapters and I am
grateful to them for their efforts to share their research results and ideas and
for delivering their manuscripts within the time and word limits. Since my
background is in Antarctic research, it was essential also to include lead-
ership in the project from the North Sea research community and I was
fortunate to have the support of Sarah Wanless and Kees Camphuysen as
co-editors of the book. I am grateful to Georgina Mace, Director of the Zoo-
logical Society of London, for supporting the proposal that developed into
the symposium and this book, and to Deborah Body from the Zoological
Society of London for all the assistance she provided in organizing the sym-
posium and in the early stages of the production of the book. I am also
grateful to Alan Crowden and others at Cambridge University Press for their
encouragement and diligence during the production of the book.

I. L. Boyd
 1
Introduction
I. L. BOYD, S. WANLESS AND C. J. CAMPHUYSEN

Marine ecosystems represent a rich assemblage of co-evolved species that

have complex, non-linear dynamics. This has made them difficult to man-
age and the recent record of exploitation of marine ecosystems suggests
that the mechanisms currently in place for their management are inappro-
priate for sustained and intensive exploitation (Pauly et al. 2002). Fisheries
science has developed sophisticated single- and multispecies approaches
to modelling resource dynamics but these have shown mixed success when
used to advise about the regulation of exploitation levels. However, it is com-
monly acknowledged that attempting to model whole or partial ecosystems
also has limited utility because the demands this has for data and know-
ledge about the system far outweigh the financial, logisitical and intellec-
tual resources available (Yodzis 1998). Although some computer-intensive
approaches are currently being attempted1 , their ability to improve predic-
tions of the dynamics of marine ecosystems appears to be quite limited.
This whole- or partial-systems approach to modelling marine ecosys-
tems is driven by a belief in the connectivity of predator–prey processes
within ecosystems and the conviction that, with appropriate parameteri-
zation, the behaviour of these systems can be predicted within bounds of
confidence that are sufficiently narrow to convince us that the investment
in the modelling effort has been useful. However, to date the cost–benefit
analysis of these approaches has not been computed and the few simple sys-
tems in which the approach has been applied soon run into trouble. Whole-
system approaches to modelling have been largely discredited because there
is always insufficient information for adequate parameterization (Plaganyi

1 The most recent version of an ecosystem-level model to be tested is known as GADGET.

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
2 I. L. Boyd et al.

& Butterworth 2004). The move towards the partial-system (or ‘minimum
realistic’, e.g. Punt & Butterworth 1995) approach leads to a necessity to
define a ‘horizon of relevance’, meaning that components of the ecosystem
that lie beyond this horizon are deemed to be of sufficiently low relevance
to the focus of management that they will not have an important influence
on the outcome of the scenarios being modelled (Schweder (Chapter 21 in
this volume)). However, these partial-system models are challenged by the
problems of diffuse effects (Yodzis 2000) which mean that the horizon of
relevance often lies well beyond our data resources (Plaganyi et al. 2001).
The problems that dog the whole-system approach to modelling marine
ecosystems therefore also dog the partial-system approach.
Like the ‘event horizon’ in cosmology, we contend that the horizon
of relevance in ecosystem modelling is an insurmountable boundary that
severely limits the extent to which we will ever be able to model rationally
constrained management scenarios for biological resources in the oceans
(and perhaps in all complex ecosystems). This is a fairly gloomy outlook
but there may be some hope for the future. This hope comes from two
directions: one involves the potential/possibility that ecosystem dynamics
could be constrained to a narrow set of rules similar to those involved in,
or associated with, the allometry of individual organisms (Garlaschelli et al.
2003); the other direction, which is the one that is explored in this book, is
to reject the reductionist approaches to ecosystem modelling by establish-
ing ecosystem boundaries and only examining ecosystem dynamics at these
boundaries. This is like attempting to understand the crustal dynamics of
the Earth by only looking at surface features. It may be possible to mea-
sure some of the critical outputs of the ecosystem in a way that provides an
insight into the internal dynamics and that could lead to some broad pre-
dictions about the behaviour of the ecosystem, especially when correlated
with known inputs. In biogeochemical terms the inputs and outputs of an
ecosystem involve primary production and the products of respiration plus
the sequestration of organic carbon, in this case as sediment on the seabed.
However, in ecological terms, the outputs could be seen as the terminal
links in food chains, sometimes also known as the top of the food chain.
Moreover, it may also be possible to understand the outputs from the ter-
minal links in the food chains without the necessity of understanding the
intermediate linkages between them and the physical-forcing processes that
are the inputs driving the food-chain dynamics. Many who like to model the
internal dynamics of these systems will consider this to be a leap of faith but,
where the intermediate dynamics have complex properties, there may be no
choice.
 Introduction 3

In practical terms, this means using the species at the top of marine food
chains as our indicators of ecosystem status and performance. We refer to
these species as ‘top predators’ but this is synonymous with ‘upper-trophic-
level predators’. For most purposes here we refer to top predators as pin-
nipeds (true seals, sea lions, fur seals and walrus), seabirds, cetaceans and
some large predatory fish. In general, they are species beyond the level of
secondary consumers. This approach has advantages and disadvantages as
outlined below.
Advantages
(1) By definition, top predators are downstream, in terms of energy flow,
of changes within an ecosystem. This means that changes in
ecosystem structure that also affect the energy flows through the
system are likely to be reflected in changes at the top of food chains.
(2) Top predators often exploit marine resources at similar spatial and
temporal scales to those used by man, thus increasing the potential for
competition. It is a truism of marine-ecosystem management that it is
only possible to manage the activities of man; however, the data we
collect about the marine ecosystem – data that come from these
activities – are collected at similar spatial and temporal scales to those
that are relevant to understanding how resource variability is likely to
affect other predators that also forage at the same scales.
(3) Many predators are accessible during important parts of their life
histories mainly because they have terrestrial breeding seasons. This
also constrains their foraging ranges because of their need to return
regularly to the breeding site. Not only does this make it relatively easy
to provide consistent indices of population sizes, it also allows
estimation of regional productivity from the productivity of the
predators themselves. This advantage applies only to seabirds and
pinnipeds, and has the effect of narrowing the focus of interest in
using top predators as measures of ecosystem outputs to these groups.
This bias is reflected in many of the following chapters.
(4) Most of the species used for measuring the outputs from ecosystems
command a high level of public interest and studies of them are likely
to attract support over the long time periods needed to measure these
ecosystem outputs.

Disadvantages
(1) Measuring the changes in top-predator populations or in the behaviour,
performance or productivity of predators does not necessarily titrate
the effects of different management interventions within ecosystems.
4 I. L. Boyd et al.

(2) Top-predator responses are not necessarily predictive so they are

difficult to use in the context of classical fisheries science to set catch
levels, although there may be some circumstances where they can help
define the broad boundaries of catch limits (e.g. Boyd 2002).
(3) Not all situations in which there is a need for management have an
appropriate community of predators available for study. In fact,
predators appropriate for use in the context of fisheries management
are mainly confined to temperate and subpolar regions and even then
they are likely to be of most relevance to coastal and shelf-seas
management.
(4) By their very nature, top predators may be several trophic links remote
from the main drivers of change in ecosystems, especially if these
drivers affect the distribution and abundance of primary production.
This could lead to attenuation of signals from variation in inputs to the
ecosystem, either through the effects of physical forcing or through the
effects of management actions.
(5) Responses of different predators to the same management or
environmental drivers may differ, not only in terms of magnitude but
even in some cases in the direction of response. In reality, many
predator studies are of single species – or at best groups of similar
predators – and this makes it difficult to assess consistency of
responses. Ideally the emphasis should be on integrated multispecies
approaches but securing funding for this is often problematic.

This book sets out to explore the hypothesis that top predators can be
used in a whole-system approach to managing marine ecosystems. In some
circumstances these predators may also provide information relevant to the
management of specific resources. The emphasis on this hypothesis does
not preclude other approaches or imply that measuring predator responses
will always be informative. However, such an approach could potentially be
part of the set of measures, insights and interpretations used within sophis-
ticated management systems. Such an integrated approach is particularly
useful where there is a need to balance the competing demands for adequate
precaution in setting resource exploitation levels against the economic and
social demands to increase these levels of exploitation still further. It takes
the focus of attention away from the resource being managed and places it
onto the ecosystem in a way that is comprehensible to most components
of the decision-making hierarchy of the management structure and to the
public.
The book represents a collection of case studies and reviews of top preda-
tors as indicators of marine-ecosystem dynamics. Many of these studies are
 Introduction 5

syntheses of other published work because the intention was to provide an

overview of the subject that would stretch the boundaries of this field to a
non-specialist audience in marine science and resource management. The
results of some of these studies are already being incorporated explicitly into
resource-management procedures. The studies are weighted towards the
North Sea system because we considered that it was important to develop a
reasonably complete description of the state of knowledge within one partic-
ular system. To provide contrast we have also included several chapters on
the krill-based systems of the Southern Ocean. Consideration is also given
to the northwest Atlantic, Arctic, North Pacific and the Barents Sea. How-
ever, we emphasize that our intention is not to provide a comprehensive
survey of the topic as the use of predators to provide information for man-
agement is present in some form within all of the most productive fisheries
management zones in the world. One particular example that we have not
illustrated is that of the California Current, for which a substantial body of
work is available (Sydeman et al. 2001).
A consistent underlying theme within the book is the need to view whole
ecosystems as the products of an evolutionary process (Fowler & MacMa-
hon 1982) which has been challenged in very recent times by a new, pow-
erful, adaptable and highly selective predator in the form of man (Trites
et al. (Chapter 2 in this volume)). The consequences of this and other nat-
ural forcing of ecosystem change – not only for the absolute abundance
of species but also for their genetic structure, size structure and nutrient
turnover – are likely to have caused irreversible changes in ecosystems,
some of which may be evident in the changes within top-predator pop-
ulations (Iverson et al. (Chapter 7) and Wolf et al. (Chapter 19) in this
volume).
A challenge to observing this process using top predators appears to
be the non-linearity of responses shown by predators to changes in food
supply. Several authors have emphasized the importance of these non-
linearities (see chapters by Croxall (Chapter 11), Furness (Chapter 14) and
Constable (Chapter 22) in this volume). Empirical observation shows that
as resource availability declines there can be little change in predator pop-
ulation productivity up to a critical point, after which declines can occur
quickly. The reasons for these non-linearities probably relate to the ability
of top predators to switch between different groups of prey (Asseburg et al.
(Chapter 18 in this volume)) and in their use of rule-based approaches to
foraging which are adaptive to changes in food distribution and abundance
(Mori & Boyd 2004). These largely behavioural adjustments can enable
individuals to maintain high feeding rates even at relatively low levels of
food availability.
6 I. L. Boyd et al.

Many of the predators considered in the range of studies represented

in this book appear to depend to a large extent upon a small range of prey
species. These are usually represented by planktivorous omnivores, mainly
small fish species but also crustaceans. In the case studies described here,
sandeels (Ammodytes spp.) and krill (Euphausia) feature prominently; in
other ecosystems these are replaced by species like sardines (e.g. Sardinops
sagax), anchovies (e.g. Engraulis ringens) or capelin (e.g. Mallotus villosus)
(e.g. Chavez et al. 2003). Many of these are keystone species in their ecosys-
tems. While many predator species show a wide-ranging diet with complex
multispecies functional responses (Asseburg et al. (Chapter 18 in this vol-
ume)), these results suggest that energy flow to top predators may be chan-
nelled mainly through a narrow range of species involving relatively high
energy-transfer efficiencies.
Although the generality of this conclusion will need further study, its
implications for using top predators to indicate change in ecosystems are
wide-ranging. Firstly, top predators may be less remote, and therefore more
responsive, than first thought to physical drivers, such as those giving rise to
decadal ocean-climate oscillations in the Atlantic and Pacific Oceans, as well
as in the Southern Ocean (Trathan et al. (Chapter 3 in this volume)). This
is because there are fewer steps along the food chain than expected, thus
reducing the potential for signal attenuation. Secondly, they may be potent
indicators of general energy flow through marine ecosystems because they
are dependent upon the omnivorous keystone species so that large-scale
changes in the dynamics of energy flow are likely to affect the top predators.
Thirdly, they are less likely to be direct competitors with man than first
thought because they are likely to prey mainly at trophic levels below that
normally targeted by fisheries (Greenstreet (Chapter 15 in this volume)).
However, recent trends in fishing suggest that fisheries are beginning to
have an impact on the same trophic levels as the top predators (Pauly et al.
1998).
To an extent, the principle is now accepted that top predators in marine
ecosystems are responsive to changes in their environment and that these
responses can be measured and used to inform management. Now, the
focus is on attempting to understand the observed dynamics of top preda-
tors in terms of changes further down in the food chains.
The story of the North Atlantic fulmar (Fulmarus glacialis) illustrates this
shift (Thompson (Chapter 10 in this volume)), as do the communities of
predators foraging on krill from South Georgia in the Southern Ocean and
on sandeels in the North Sea. The species present in these communities
appear to forage at different spatial scales and capitalize on different prey
 Introduction 7

distributions in the water column (Camphuysen et al. (Chapter 6) and Scott

et al. (Chapter 4) in this volume). A wide range of variables can be mea-
sured cheaply and consistently across these predators, including indicators
of breeding success, population growth, and individual growth and survival.
All these variables have the potential to be used individually or as groups to
examine the dynamics of prey populations, even to the extent that they can
be used in specific circumstances to sample the age or size structure of prey
populations (Reid et al. (Chapter 17 in this volume)).
The methods used to measure most variables in many species of
seabirds and seals are well established and this has resulted in sets of data
collected using consistent methodologies over several decades (e.g. Croxall
(Chapter 11 in this volume)). These types of datasets are beginning to pro-
vide the foundation for approaches to combining indices from top predators
that integrate across those predator species that operate at similar spatial
and temporal scales among species (see Croxall (Chapter 11) and Constable
(Chapter 22) in this volume). We are on the verge of developing sophis-
ticated approaches to target management advice based on predator perfor-
mance, approaches that can predict how the physical dynamics of the oceans
may affect foraging success in these top predators (Scott et al. (Chapter 4)
and Trathan et al. (Chapter 3) in this volume).
An important element in developing a predictive approach to the
response of predators to the dynamics of their food supply is provided by
modelling predator behaviour from first principles (Ollason et al. (Chapter
20 in this volume)). Apart from the few occasions when it is possible to
relate predator responses directly to prey dynamics, models fitted to preda-
tor behaviour may be the only way of understanding the form of the func-
tions that relate predator responses to prey dynamics. It seems to us that
it is vital to know as much as possible about the non-linear form of these
functions through a combination of modelling and targeted experimental
studies. The different approaches to modelling are illustrated by the pre-
dictive models of Ollason et al. (Chapter 20 in this volume) and the post
hoc statistical fitting to data of Wolf et al. (Chapter 19 in this volume). The
approach taken by Enstipp et al. (Chapter 13 in this volume), which models
the energy budgets of predators, also provides an analysis that points to the
range of environmental productivity required to sustain predators.
Current systems used in the management of marine bioresources are
adapting slowly to the need to include information from predators. Two
contrasting approaches are illustrated by Constable (Chapter 22) and Tasker
(Chapter 24 in this volume). In general, traditional fisheries management
approaches cannot be easily adapted to include information from predators
8 I. L. Boyd et al.

but Constable points us in a direction that could lead to appropriate inte-

gration; in addition, the scenario models described by Schweder (Chapter
21 in this volume) for the Barents Sea are beginning to develop a frame-
work for integrating top-predator information into fisheries management
procedures. Although there are moves to include these approaches within
some fisheries management procedures, especially in the Southern Ocean,
it seems likely that the systems involving most northern-hemisphere fish-
eries, where biology is tensioned more strongly against social and economic
issues, will take much longer to fall into line. Here, as Tasker points out,
a less sophisticated and more pragmatic system of thresholds and targets
for predator populations is appropriate and more likely to gain acceptance.
It has been in the context of the North Sea that fisheries management tar-
gets have been adjusted based upon the breeding performance of the black-
legged kittiwake (Rissa tridactyla). This approach may represent the kind of
operational decision rule that is required in the future for integrating the tra-
ditional single-species approach to management of a fish population with
information for top predators that are also dependent on that resource.
Finally, the creation of marine protected areas based upon the distri-
bution of marine predators is a management procedure that has a wider-
ranging effect upon ecosystem sustainability than might be possible with
procedures targeted at the exploited components of an ecosystem (Hooker
(Chapter 23 in this volume)). This is underpinned by the principle that
maintaining healthy top-predator populations is closely linked with main-
taining a healthy ecosystem.
The current research effort is providing a range of management tech-
niques that are underpinned by a whole-system approach to bioresource
management. This book represents a synthesis that reflects the state of the
research field and is intended to provide managers, and those with interests
in marine-resource management, with the materials necessary to under-
stand what has been achieved to date. However, it is important to empha-
size that the approach to marine-bioresource management advocated here
is not an easy option and to be successful it will require targeted research
in several key areas:

(1) Coordinated data collection and management schemes like that

developed for the Southern Ocean by the Commission for the
Conservation of Antarctic Marine Living Resources (CCAMLR) (Agnew
1997, Constable (Chapter 22 in this volume)). That there is no
coordinated system of data collection and management for pinnipeds
within Europe is a reflection of the fractured nature of
 Introduction 9

marine-ecosystem research in the region, with competing national

interests and funding structures driving the science agenda.
(2) Greater intellectual input to the methods of integrating predator data
into management structures is required, although a prerequisite of this
is the provision of opportunity for integration through the
development of appropriately open-minded management regimes.
(3) Continuation and, where possible, enhancement of detailed process
studies involving marine predators and their response to changes in
food availability and, in some cases, investigation of the physical
drivers of these changes.
(4) Identification of the critical foraging and breeding habitat of marine
predators as a prerequisite to identifying regions that can be protected
not only for the conservation of the predators themselves but also the
ecological relationships that help to sustain them.

We hope that this book will illustrate that we are part of the way to achieving
these objectives.

REFERENCES
Agnew, D. J. (1997). The CCAMLR ecosystem monitoring programme. Antarct.
Sci., 9, 235–42.
Boyd, I. L. (2002). Integrated environment–prey–predator interactions off South
Georgia: implications for management of fisheries. Aquat. Conserv., 12, 119–26.
Chavez, F. P., Ryan, J., Lluch-Cota, S. E., & Niquen, M. (2003). From anchovies to
sardines and back: multidecadal change in the Pacific Ocean. Science, 299,
217–21.
Fowler, C. W. & MacMahon, J. A. (1982). Selective extinction and speciation: their
influence on the structure and functioning of communities and ecosystems.
Am. Nat., 119, 480–98.
Garlaschelli, D., Calderelli, G. & Pietronero, L. (2003). Universal scaling relations
in food webs. Nature, 423, 165–8.
Mori, Y. & Boyd, I. L. (2004). The behavioral basis for nonlinear functional
responses and optimal foraging in Antarctic fur seals. Ecology, 85, 398–410.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, F., Jr (1998). Fishing
down marine food webs. Science, 279, 860–3.
Pauly, D., Christensen, V., Guénette, S. et al. (2002). Towards sustainability in
world fisheries. Nature, 418, 689–95.
Plaganyi, E. E. & Butterworth, D. S. (2004). A critical look at the potential of
Ecopath with Ecosim to assist in practical fisheries management. Afr. J. Mar.
Sci., 26, 261–87.
Plaganyi, E. E., Butterworth, D. S. & Brandao, A. (2001). Towards assessing the
South African abalone Haliotis midae stock using an age-structured production
model. J. Shellfish Res., 20, 813–27.
Punt, A. & Butterworth, D. S. (1995). The effects of future consumption by the
Cape fur seal on catches and catch rates of the Cape hakes. 4. Modelling the
10 I. L. Boyd et al.

biological interaction between Cape fur seals Arctocephalus pusillus pusillus and
the Cape hake Merluccius capensis and M. paradoxus. S. Afr. J. Mar. Sci., 16,
255–85.
Sydeman, W. J., Hester, M. M., Thayer, J. A. et al. (2001). Climate change,
reproductive performance and diet composition of marine birds in the
southern California Current system, 1969–1997. Prog. Oceanogr., 49, 309–29.
Yodzis, P. (1998). Local trophodynamics and the interaction of marine mammals
and fisheries in the Benguela ecosystem. J. Anim. Ecol., 67, 635–58.
(2000). Diffuse effects in food webs. Ecology, 81, 261–6.
 2
Effects of fisheries on ecosystems: just another
top predator?
A . W . T R I T E S , V . C H R I S T E N S E N A N D D . P AU L Y

Apex predators – such as pinnipeds, cetaceans, seabirds and sharks – are

constrained by the sizes of prey they can consume and thus typically feed
within a narrow range of trophic levels. Having co-evolved with their prey,
they have influenced the behaviours, physiologies, morphologies and life-
history strategies of the species they target. In contrast, humans can con-
sume prey of any size from all trophic levels using methods that can rapidly
deplete populations. On an ecological time scale, fisheries, like apex preda-
tors, can directly affect the abundance of other species by consuming or out-
competing them; alternatively they can indirectly affect the abundance of
non-targeted species by removing other predators. However, there is grow-
ing evidence that the effects of fisheries go well beyond those imposed by
apex predators. Theory and recent observations confirm that the continued
development and expansion of fisheries over the past half century has led to
a decrease in the size and life spans of targeted species, with reproduction
of fish occurring at earlier ages and at smaller sizes. In addition, high lev-
els of fishing have altered the makeup of many ecosystems, depressing the
average trophic level of heavily fished ecosystems and speeding up the rate
of nutrient turnover within them. Inevitable consequences of fishing down
the food web are increased ecosystem instability, unsustainable fisheries
and an inability for the ecosystem to support healthy, abundant populations
of apex predators.
Outside of a general appreciation that fishing can directly reduce the
numbers of targeted and non-targeted (by-caught) species, there appears to
be little understanding of the effects that fishing can have on other species

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
12 A. W. Trites et al.

or on the ecosystem as a whole. This is due in part to the inherent spa-

tial and temporal complexities associated with studying marine ecosystems,
as well as the perceived expense and difficulties associated with monitor-
ing and tracking changes and responses in complex systems. Fisheries can
directly affect food webs by removing large numbers of targeted and by-
caught species, and by having a physical impact on the habitat of benthic
organisms through bottom trawling (Goni 1998). Such direct effects of fish-
eries on food webs are relatively easy to document. The less intuitive effects
of fisheries are those that are indirect. They include altered food-web inter-
actions, increasing rates of nutrient turnover caused by discarded unwanted
fish and organic detritus (offal), and continued mortality caused by lost gear
(ghost fishing).
Faced with a paucity of data to assess indirect effects of fisheries, it is easy
to be lulled into assuming that humans have no more effect on food webs
than do other apex predators – whether they be sharks, sea lions, whales
or birds. On some levels there are undoubtedly parallels between how fish-
eries and apex predators affect food webs, but on others the two groups
are clearly subject to different controls and exert different forces on marine
ecosystems.
Considerable advances have been made in recent years in compiling
and analysing extensive fishery datasets and in developing mathematical
descriptions of ecosystems. These tools provide significant insights into the
direct and indirect effects of fisheries on ecosystems. They also provide a
means for a better understanding of the roles that apex predators play in
marine ecosystems and the ultimate role that fisheries play in influencing
the dynamics of apex predators.

DIRECT EFFECTS OF FISHING

Humans have long had an association with the coastal regions of the world
and can trace the expansion of civilization to the ready supply of fish, inver-
tebrates and mammals that could be easily gathered, caught or hunted.
Until relatively recently, human activities were restricted to the nearshore
and surface waters. But all of this has changed over the past century as fossil
fuels and the application of new technologies have allowed fisheries to move
further from shore, to fish in deeper waters and to become more effective at
finding and capturing species from all levels of the food chain (Botsford et al.
1997, Merrett & Haedrich 1997, Hutchings 2000, Pauly et al. 2002, 2003).
The geographic and technical expansion of fisheries around the world
through the early to late 1900s was mirrored by steady increases in world
 Effects of fisheries on ecosystems 13

90

80
Global catch (× 106 t)

70

60

50

0
1970 1980 1990 2000
Year

Fig. 2.1 Global reported catch with (dotted line) and without (solid line) the
Peruvian anchovy. Total world catches have fluctuated around 82 million tonnes
since the late 1980s due to high catches of Peruvian anchovy which compensated
for the global decline in landings of all other species combined. Adapted from
Watson and Pauly (2001).

catches. However, global landings began stagnating in the early 1980s, and
have decreased since the late 1980s (Watson & Pauly 2001). This decreasing
trend is particularly apparent when the widely fluctuating catch of Peruvian
anchovy is discounted (Fig. 2.1).
Christensen et al. (2003b) reconstructed the biomass of commercially
important fishes that were present in the North Atlantic around 1900.
They considered the abundances of high-trophic-level species such as tuna,
sharks, mackerel, cod, flatfish and salmon – and relied on 23 spatialized
ecosystem models and multiple regressions that considered environmen-
tal and biological factors to predict abundances over spatial resolutions of
1
2
× 21 degree latitudes and longitudes. Plotting the densities of fishes in the
North Atlantic showed the relatively high productivity of the shelf regions
of Europe and eastern North America (Fig. 2.2). However, the data also
showed a major decline between 1900 and 1999 in the range and biomass
of the top predatory fishes that are typically found on dinner tables. Biomass
in the North Atlantic fell by 90% during the twentieth century, leading to
declines of catches throughout the North Atlantic, notably in eastern Canada
(Fig. 2.2). Similar downward trends in the biomasses of high-trophic-
level fishes have been seen elsewhere in the world where they have been
14 A. W. Trites et al.

Biomass Catch
> 11

1900 < 11
> 10
<9
> 2.0
< 2.0
< 1.0
< .75
<8
< .5
<7
< .4
<6
< .3
<5
< .2
<4
< .15
<3
< .1
<2
< .05
<1
< .01

1999

Fig. 2.2 Predicted biomass distributions and estimated fishing intensity for
high-trophic-level fishes (trophic level (TL) ≥ 3.75) in the North Atlantic in 1900
and 1999. Legend indicates biomass in tonnes per square kilometre. Adapted
from Christensen et al. (2003b).

investigated, such as northwest Africa or southeast Asia, where biomass

declined as fishing intensity increased (Christensen et al. 2003a, 2004).
Although it appears to have taken less than a century for North Atlantic
fisheries to reduce the biomass of the high-trophic-level fishes to under
10% of their original amounts (Christensen et al. 2003b), declines were
well underway before the advent of modern fishing technologies. The once-
abundant populations of marine mammals, turtles and large fishes are
believed to have incurred massive declines long before 1900 (Jackson et al.
2001). However, the pace of change has quickened considerably in recent
years to the point that industrial fleets may now only require a few decades
to reduce all fish populations to 10% of their unfished levels (Myers & Worm
2003).
Overall, there has been a serial depletion of species from marine ecosys-
tems (Pauly et al. 1998a, Jackson et al. 2001, Myers & Worm 2003). Histor-
ically, fishing started at the top of most food chains by removing the larger,
valuable and more easily caught species – then moved down to the next-
biggest species as those above were depleted and were no longer easy or
economical to catch. The downward shift towards taking species from lower
trophic levels is termed ‘fishing down the food web’ (Pauly et al. 1998a).
 Effects of fisheries on ecosystems 15

5 Sperm
Beaked whales
whales

Toothed
whales Steller
sea lions
Pisc. Large Deepwater

4 Seals Other
demersal
bird flatfish fish

fishes Squid and

Trophic level

Baleen octopus
whales Walrus and
bearded seal
Adult pollock
Pelagics Small
Jellyfish Juvenile flatfish
pollock
3 Benthic
feeders

Epifauna
Large
zooplankton
Herbivorous Infauna
2 plankton

1 Phytoplankton Detritus

Fig. 2.3 An energy-flow, food-web description of the eastern Bering Sea

highlighting the trophic level of each group of species and the relative strength of
the interactions (based on the amount of energy flowing from producers and
consumers). Here species or groups of species are placed according to their
trophic level (calculated as 1.0 plus the mean trophic level of the species that they
consume), and the size of each box is relative to the biomass of the species. In
general, trophic levels of functional groups tend to cluster about integer values.
Pisc., piscivorous. From Trites et al. (1999).

Trophic levels are the layers that make up food webs, wherein animals
are ranked according to how many steps they are above the primary pro-
ducers at the base of the food web (e.g. Fig. 2.3). Microscopic plants at the
bottom are assigned a trophic level of 1, while the herbivores and detriti-
vores that feed on the plants and detritus make up trophic level 2. Higher-
order carnivores, such as most marine mammals, are assigned trophic lev-
els ranging from 3 to 5 (Pauly et al. 1998b, Trites 2001). Animals that feed
from more than one trophic level have non-integer trophic levels. Thus,
knowing what an animal eats is all that is needed to calculate its trophic
level.
Pauly et al. (1998a) calculated the mean trophic level of reported catches
and found that it had declined over the years. The sharpest declines were
noted in the northwest Atlantic where the mean trophic level dropped from
a peak of 3.7 in 1965 to 2.8 in 1997. Smaller declines were noted in the north-
east Atlantic and the Mediterranean Sea, and have been reported elsewhere
as well (e.g. Arias-Gonzalez et al. 2004, Sanchez & Olaso 2004). Overall,
there has been an average worldwide trophic decline of 0.1 per decade in
16 A. W. Trites et al.

110

100
Maximum length (cm)
90 North Atlantic
80

70

Global coastal
60

50

40
1950 1960 1970 1980 1990 2000
Year
Fig. 2.4 Maximum length attained by species landed from 1950 to 2000.
Adapted from R. Watson, unpublished data.

the mean trophic level of species caught since the 1970s. The inevitable
consequence of such a decline is that species from the lowest trophic levels
may eventually become the mainstay of commercial fisheries.
In addition to progressively catching increasing numbers of species
from lower trophic levels, there are also clear signs that the fish being caught
are no longer as big as they once were. Globally, there has been about a
25% reduction in the mean maximum lengths of landed fish from coastal
waters (Fig. 2.4). These trends reflect the increasing importance to fisheries
of catching smaller species, as well as the selective forces of fisheries which
are resulting in individual fish maturing at smaller sizes.
Reductions in size-at-age and age-at-maturation of commercially
exploited fish have been reported in a number of ecosystems (Trippel 1995,
Rochet 1998, Law 2000). On the Scotian Shelf in eastern Canada, for exam-
ple, average weights of individual demersal fish have decreased by 41% to
51% since the 1970s (Zwanenburg 2000). Declines have also been reported
in the sizes of large demersal fishes in other high-latitude regions, and
have been detected in some – but not all – tropical regions (Pauly 1980,
Bianchi et al. 2000). Such changes in body sizes are particularly troubling,
given that survival and reproduction are functions of body size (Reiss 1989).
Small fish generally incur higher mortality rates and produce fewer eggs
than larger fish. Fisheries thus appear to have the potential to cause evolu-
tionary changes based on the apparent phenotypic and genetic response of
exploited species (Heino & Godo 2002, Hutchings 2004, Olsen et al. 2004).
 Effects of fisheries on ecosystems 17

Whether or not such effects of fisheries are long lasting or reversible is not
known. However there are indications that genetic drift may compromise a
population’s recovery after severe depletion (Hutchinson et al. 2003).
The most obvious direct effect of fisheries on marine ecosystems has
been reductions in the abundances of targeted species, such as cod on the
east coast of Canada (Walters & Maguire 1996). Less obvious has been
declines in the spatial ranges and degrees of overlap of depleted targeted
species with others in the ecosystem (e.g. Garrison & Link 2000), as well
as the potential losses of biodiversity (Agardy 2000). Losses of individuals
through bycatch and ghost fishing are poorly documented, but are known
to have directly affected many populations of the larger species such as
seabirds, turtles and marine mammals. Entanglement in fishing gear or
being caught and drowned by baited hooks have threatened a number of
species (e.g. Tasker et al. 2000). Tuna and shrimp fisheries have also been
responsible for large levels of bycatch of unwanted species, but the ecosys-
tem effects of bycatch are less well documented or understood. Finally, there
are the direct and immediate effects of bottom trawling that can modify
benthic habitat and community structure (e.g. Koslow et al. 2000, Jennings
et al. 2001), as well as severely decreasing benthic mega-fauna production
(Hermsen et al. 2003).

INDIRECT EFFECTS OF FISHING

Considerable progress has been made in understanding the potential indi-

rect effects of fishing through mathematical descriptions of ecosystems and
the accessibility of ecosystem modelling software (e.g. Walters et al. 1997,
Christensen & Walters 2004). A number of models have recently been
constructed that provide insights into how fisheries might alter food-web
interactions and increase the rates of nutrient turnover through discarded
organic detritus (offal) and unwanted species (e.g. Christensen 1998, Cox
et al. 2002, Arias-Gonzalez et al. 2004). For example, simulation models of
the eastern Bering Sea explored the ecosystem consequences of commer-
cial whaling and catching groundfish; the models showed that removing
whales had a small positive effect on groundfish by reducing competition
for food, and that reducing fishery removals of cannibalistic adults reduced
the amount of young pollock available for marine mammals to eat (Trites
et al. 1999, 2004). The models also revealed a high degree of potential com-
petition between seals and large flatfish and adult pollock, as well as the
large overlaps in the diets of pollock and baleen whales. Models have also
been used to show the importance of icefish in the Antarctic ecosystem,
18 A. W. Trites et al.

as well as the possible consequences that catching increased amounts of

krill might have on penguins and other species in the Antarctic ecosystem
(Alonzo et al. 2003, Bredesen 2003). The models generally go far beyond
the simple, single-species assessment of the effects of removing biomass,
and have revealed unexpected effects of fishing through indirect food-web
interactions.
Excessive catches of one species may lead to the collapse of an important
predator or prey in the system and may cause changes in the growth and
survival patterns of other species in the food web (Walters & Kitchell 2001).
Continued expansion of fisheries into deeper waters and the targeting and
marketing of new species may increase competition with predators previ-
ously unaffected by fisheries, such as beaked whales and sperm whales.
Changes at one level of a food web can also have cascading effects on oth-
ers; for example in the Black Sea a trophic cascade is believed to have been
launched by fishery removals of apex predators which caused planktivorous
fish to increase, and led to a decline in zooplankton biomass that in turn
allowed phytoplankton to increase (Daskalov 2002). This chain of events is
thought to explain the explosions of phytoplankton and jellyfish reported
in the Black Sea over the past 30 years (Daskalov 2002). In the Bering
Sea, a similar sort of question has been posed over whether the removal
of large baleen whales resulted in cascading declines of other species
of marine mammals (Springer et al. 2003) or increases in flatfish and
gadids (Trites et al. 1999). Mathematical models provide a structured frame-
work to test the consistency of trophic-cascade arguments and a means of
gaining further insight into the possible unexpected effects of fishing on
ecosystems.
Some groups of apex predators, such as marine mammals, may be
affected by fisheries even when the prey and species caught do not over-
lap. This has been termed ‘food-web competition’ (Trites et al. 1997). It
occurs at the base of the food pyramids and involves the primary produc-
tion required to sustain these pyramids (Fig. 2.5). Food-web competition
occurs when there is potential overlap of the trophic flows supporting a
given group (such as marine mammals) with the trophic flows supporting
another group (such as fisheries). The relationship between the size of fish-
ery catches and the amounts of primary production required to sustain fish-
eries and marine mammals suggests that the primary production available
to marine mammals may decline as catches increase (Trites et al. 1997).
This raises the possibility that fisheries in some areas may have entered
into ‘food-web competition’ with marine mammals.
 Effects of fisheries on ecosystems 19

Fig. 2.5 Schematic of food-web competition, illustrating how marine mammals

and fisheries may not directly compete (because they consume different species),
but could indirectly compete through the primary production required to sustain
each of their respective prey populations. Adapted from Trites et al. (1997).

Over the past 80 years, food-web research has sought to identify recur-
ring patterns and underlying mechanisms and constraints on ecosystem
structure (e.g. Elton 1927, Lindeman 1942, Pimm 1982, Cohen et al. 1990,
Trites 2003). Many of the conclusions that stem from such studies make
intuitive sense, such as the fact that most food chains are size-structured:
where most predators are larger than their prey and trophic level typi-
cally increases with increasing body size (Pope et al. 1994). Overall, system
biomass appears to be proportional to primary production (Pimm 1982),
and the proportion of species occupying top, intermediate and basal trophic
levels appears to be constant across food webs (Cohen 1978). There also
appears to be a relatively constant ratio of two to three species of prey
for every predator in an ecosystem (Martinez 1991), although numbers of
species of prey consumed by each species of predator tends to increase as
the size of the food web increases.
The lengths of most chains that form food webs are typically short (two
to five links) with maximum reported lengths of eight in tropical shelf seas,
20 A. W. Trites et al.

seven in tropical estuaries and six in oceanic upwelling areas (Christensen

& Pauly 1993). Food-chain lengths appear to be a function of both environ-
mental stability and energy-transfer efficiency between trophic levels (Trites
2003). Longer food chains exist associated with stable environments, while
shorter food chains exist in less predictable environments (Pimm 1991, Jen-
nings & Warr 2003). Species at the end of long food chains would be at risk
of extinction if the abundance of species lower in the food chain fluctuated
severely through natural or anthropogenic causes (fishing). Thus it appears
that fisheries have the potential to disrupt the biological structure of food
webs, particularly in ecosystems comprised primarily of long food chains.

EFFECTS OF APEX PREDATORS

A number of the parallels and differences that exist between fisheries and
apex marine predators offer additional insights into the effects that com-
mercial fisheries can have on ecosystems. As with fisheries, there is a long-
held belief that marine mammals significantly affect prey populations (Gul-
land 1987, Butterworth 1992, Tamura 2003). This is best demonstrated by
the declines of crabs, abalone and urchin numbers that have followed the
re-introductions and range expansions of sea otters (Estes 1996). Similarly,
declines of sea otters and other marine mammals in the Aleutian Islands
and Gulf of Alaska may be the result of ‘over-hunting’ by killer whales (Estes
et al. 1998, Doroff et al. 2003, Springer et al. 2003). As for other obser-
vations of the direct effects of marine mammals on ecosystems, examples
appear to be lacking. This is probably due to an absence of data on the abun-
dance of prey species prior to the recovery of exploited marine mammal
populations – or it might reflect a lack of suitable experimental controls or
monitoring of forage species to make proper comparisons and conclusions.
As with trawl fisheries, some apex predators such as walrus and grey
whales can influence the turnover of nutrients by feeding on species that
live in bottom sediments. Marine mammals can also influence growth rates
and the sizes at maturity of their prey – as demonstrated in lakes in Quebec,
Canada which are home to land-locked harbour seals that feed on trout.
Trout in these lakes are younger and spawn at younger ages than in adjacent
lakes without harbour seals (Power & Gregoire 1978). The trout also grow
faster and attain smaller sizes in the lakes inhabited by harbour seals.
While there are undoubtedly parallels between the effects of fisheries
and those of marine mammals on food chains, there are at least three impor-
tant differences. One is that mammals and all other species that make up
food webs are generally limited by the size of prey they can consume. They
 Effects of fisheries on ecosystems 21

also tend to be specialized feeders and hence draw their energy from a very
limited range of trophic levels (e.g. Fig. 2.3); in contrast, humans can feed on
any size of organism at any trophic level. A second major difference between
fisheries and apex predators is that predator populations in naturally occur-
ring systems are regulated through density-dependent processes that limit
reproduction and survival as prey populations decline; however, there is lit-
tle to regulate the rate of fishery catches apart from economic incentives,
which normally increase (rather than decline) as the species becomes rarer.
The third, and perhaps most significant, difference between the two is that
stable marine food webs are the result of a long period of natural selection
and co-evolution between predators and prey, whereas fisheries represent
an abrupt, knife-edged selective force that has potentially destabilizing con-
sequences.
To capture their prey, apex predators have evolved special sensory abil-
ities (e.g. vision and hearing), morphologies (e.g. dentition) and physiolo-
gies (e.g. diving and breath-holding abilities) (Trites 2002). They have also
evolved specialized feeding behaviours to capture prey that move diurnally
up and down the water column, or to capture prey that move seasonally
across broad geographic ranges. In response, fish and other cold-blooded
species of prey have evolved a number of strategies to increase their chances
of survival. One is cryptic coloration, such as flatfish that blend in with the
bottom when viewed from above but have white undersides to avoid detec-
tion when seen from below against a bright sea surface. Many species of
fish, invertebrates and zooplankton take refuge from predators in the deep,
dark waters during the day and move towards the surface to feed under
the cover of night. Another strategy that some species evoke is predator
swamping, such as the large aggregations of spawning salmon and her-
ring which reduce the numerical effect of predators on their prey popula-
tions. Schooling is another anti-predator behaviour that creates confusion
through the sheer volume of stimuli from a fleeing school, making it dif-
ficult for apex predators to actively select and maintain pursuit of single
individuals.
In addition to directly affecting the behaviours and life histories of other
species, some apex predators may also have indirectly influenced the evolu-
tion of non-targeted species in their ecosystems by consuming the predators
of these species. One such example of this is the relative lack of chemical
defences of kelp and other marine algae against urchin predation in systems
that also contained sea otters (Estes 1996). Another is the consumption by
cod of the potential predators and competitors of their young that has effect-
ively resulted in cod ‘cultivating’ their young (Walters & Kitchell 2001).
22 A. W. Trites et al.

Overall, it appears that predator–prey interactions have shaped each oth-

ers life-history strategies, and potentially those of their competitors as well
(Katona & Whitehead 1988, Bowen 1997, Trites 1997). These mechanisms
and adaptive traits have undoubtedly helped to maintain the integrity and
stability of marine ecosystems. However, none of these selected characteris-
tics are likely to be effective at maintaining populations targeted by fisheries.
In fact, many of the features that have allowed prey to flourish in the face of
apex predators now make fish more vulnerable to being caught by fisheries
(e.g. schooling behaviour, diurnal movement towards surface light, etc.).

CONCLUSIONS

Marine ecosystems encompass a broad range of habitat types and harbour a

wealth of species and genetic diversity. They consist of dynamic food webs
whose species have slowly co-evolved to form the systems present today.
Fisheries are a relative latecomer to the evolutionary predator–prey game,
and are playing by different sets of rules and with a different set of ultimate
consequences. However, the continued growth and expansion of fisheries
around the world suggests that there is little appreciation of the effects that
fishing has on ecosystems.
On many levels, fisheries have a lot in common with apex predators
in that they can reduce the abundance of their prey and can influence the
rates of growth and maturity of the species they target. Fisheries can also
influence rates of turnover and nutrient cycling. However, the effects of
fisheries go well beyond those of other apex predators, due in large part to
their capacity to remove large amounts of biomass from the world’s oceans
and the lack of biological controls or feedback to limit what and how much
they take.
Humans are biologically successful because they can feed from all
trophic levels and can consume any size of prey. They are also not impeded
by low prey abundance. The highly selective nature of fisheries means that
they have the potential to cause evolutionary changes in the species they
target that may be hard to reverse.
Current understanding of the full indirect effects that major removals
by fisheries can have on other components of the ecosystem is poor but
is improving with the development of ecosystem models and an empha-
sis on better monitoring of indicators of ecosystem change. Although not
all ecosystems are over-fished, e.g. Alaska (Pew Oceans Commission 2003,
US Commission on Ocean Policy 2004), they are an increasing minority
rather than the rule given that 75% of all fished populations are either fully
 Effects of fisheries on ecosystems 23

exploited, over-exploited or depleted (FAO 2002). Similarly, not all changes

that occur in ecosystems are the result of fishing (e.g. Caddy 2000, Benson
& Trites 2002). However, fishing alone has the capacity to operate outside
of the natural rules that govern populations and their ecosystems. In gen-
eral, large removals by fisheries can have destabilizing effects on marine
ecosystems, particularly on systems composed of highly reticulated food
webs. Intensive fishing may thus lead to large and long-lasting ecosystem
changes, and an inability of heavily fished ecosystems to support abundant,
healthy populations of apex predators.

ACKNOWLEDGEMENTS

Support for V. C. and D. P. was made possible by the Sea Around Us

Project, which was initiated and funded by the Pew Charitable Trust;
D. P. also received support from Canada’s National Science and Engi-
neering Research Council. A. W. T. was supported by the North Pacific
Marine Science Foundation through the North Pacific Universities Marine
Mammal Research Consortium.

REFERENCES

Agardy, T. (2000). Effects of fisheries on marine ecosystems: a conservationist’s

perspective. ICES J. Mar. Sci., 57, 761–5.
Alonzo, S. H., Switzer, P. V. & Mangel, M. (2003). An ecosystem-based approach
to management: using individual behaviour to predict the indirect effects of
Antarctic krill fisheries on penguin foraging. J. Appl. Ecol., 40, 692–702.
Arias-Gonzalez, J. E., Nunez-Lara, E., Gonzalez-Salas, C. & Galzin, R. (2004).
Trophic models for investigation of fishing effect on coral reef ecosystems.
Ecol. Model., 172, 197–212.
Benson, A. J. & Trites, A. W. (2002). Ecological effects of regime shifts in the
Bering Sea and eastern North Pacific Ocean. Fish Fisher., 3, 95–113.
Bianchi, G., Gislason, H., Graham, K. et al. (2000). Impact of fishing on size
composition and diversity of demersal fish communities. ICES J. Mar. Sci., 57,
558–71.
Botsford, L. W., Castilla, J. C. & Peterson, C. H. (1997). The management of
fisheries and marine ecosystems. Science, 277, 509–15.
Bowen, W. D. (1997). Role of marine mammals in aquatic ecosystems. Mar. Ecol.
Prog. Ser., 158, 267–74.
Bredesen, E. L. (2003). Krill and the Antarctic: finding the balance. M.Sc. thesis,
University of British Columbia, Canada.
Butterworth, D. S. (1992). Will more seals result in reduced fishing quotas? S. Afr.
J. Sci., 88, 414–16.
Caddy, J. F. (2000). Marine catchment basin effects versus impacts of fisheries on
semi-enclosed seas. ICES J. Mar. Sci., 57, 628–40.
24 A. W. Trites et al.

Christensen, V. (1998). Fishery-induced changes in a marine ecosystem: insight

from models of the Gulf of Thailand. J. Fish Biol., 53, 128–42.
Christensen, V. & Pauly, D. (1993). Flow characteristics of aquatic ecosystems. In
Trophic Models of Aquatic Ecosystems, eds. V. Christensen & D. Pauly. ICLARM
Conference Proceedings 26. Manila, Philippines: ICLARM, pp. 338–52.
Christensen, V. & Walters, C. J. (2004). Ecopath with Ecosim: methods,
capabilities and limitations. Ecol. Model., 172, 109–39.
Christensen, V., Garces, L. R., Silvestre, G. T. & Pauly, D. (2003a). Fisheries impact
on the South China Sea Large Marine Ecosystem: a preliminary analysis using
spatially-explicit methodology. In Assessment, Management and Future Directions
for Coastal Fisheries in Asian Countries, eds. G. T. Silvestre, L. R. Garces,
I. Stobutzki, et al. World Fish Center Conference Proceedings 67. Penang,
Malaysia: WorldFish Centre, pp. 51–62.
Christensen, V., Guénette, S., Heymans, J. J. et al. (2003b). Hundred-year decline
of North Atlantic predatory fishes. Fish Fisher., 4, 1–24.
Christensen, V., Amorim, P., Diallo, I. et al. (2004). Trends in fish biomass off
Northwest Africa, 1960–2000. In Marine Fisheries, Ecosystems, and Societies in
West Africa: Half a Century of Change, eds. P. Chavance, M. Ba, D. Gascuel,
J. M. Vakily & D. Pauly. Paris: Institut de Recherche pour le Development;
Brussels: European Commission, pp. 377–86.
Cohen, J. E. (1978). Food Webs and Niche Space. Princeton, NJ: Princeton University
Press.
Cohen, J. E., Briand, F. & Newman, C. M. (1990). Community Food Webs: Data and
Theory. New York: Springer-Verlag.
Cox, S. P., Essington, T. E., Kitchell, J. F. et al. (2002). Reconstructing ecosystem
dynamics in the central Pacific Ocean, 1952–1998. II. A preliminary
assessment of the trophic impacts of fishing and effects on tuna dynamics.
Can. J. Fish. Aquat. Sci., 59, 1736–47.
Daskalov, G. M. (2002). Overfishing drives a trophic cascade in the Black Sea. Mar.
Ecol. Prog. Ser., 225, 53–63.
Doroff, A. M., Estes, J. A., Tinker, M. T., Burn, D. M. & Evans, T. J. (2003). Sea
otter population declines in the Aleutian archipelago. J. Mamm., 84, 55–64.
Elton, C. (1927). Animal Ecology. London: Sidgwick and Jackson.
Estes, J. A. (1996). The influence of large, mobile predators in aquatic food webs:
examples from sea otters and kelp forests. In Aquatic Predators and their Prey,
eds. S. P. R. Greenstreet & M. L. Tasker. Oxford, UK: Fishing News Books,
pp. 65–72.
Estes, J. A., Tinker, M. T., Williams, T. M. & Doak, D. F. (1998). Killer whale
predation on sea otters linking oceanic and nearshore ecosystems. Science, 282,
473–6.
FAO (Food and Agriculture Organization) (2002). The State of World Fisheries and
Aquaculture. Rome, Italy: Fisheries Department, Food and Agriculture
Organization.
Garrison, L. P. & Link, J. S. (2000). Fishing effects on spatial distribution and
trophic guild structure of the fish community in the Georges Bank region.
ICES J. Mar. Sci., 57, 723–30.
Goni, R. (1998). Ecosystem effects of marine fisheries: an overview. Ocean Coastal
Managmt, 40, 37–64.
 Effects of fisheries on ecosystems 25

Gulland, J. A. (1987). The impact of seals on fisheries. Mar. Policy, 11, 196–204.
Heino, M. & Godo, O. R. (2002). Fisheries-induced selection pressures in the
context of sustainable fisheries. Bull. Mar. Sci., 70, 639–56.
Hermsen, J. M., Collie, J. S. & Valentine, P. C. (2003). Mobile fishing gear reduces
benthic megafaunal production on Georges Bank. Mar. Ecol. Prog. Ser., 260,
97–108.
Hutchings, J. A. (2000). Collapse and recovery of marine fishes. Nature, 406,
882–5.
(2004). The cod that got away. Nature, 428, 899–900.
Hutchinson, W. F., Oosterhout, C. van, Rogers, S. I. & Carvalho, G. R. (2003).
Temporal analysis of archived samples indicates marked genetic changes in
declining North Sea cod (Gadus morhua). Proc. R. Soc. Lond. B, 270, 2125–32.
Jackson, J. B. C., Kirby, M. X., Berger, W. H. et al. (2001) Historical overfishing and
the recent collapse of coastal ecosystems. Science, 293, 629–38.
Jennings, S. & Warr, K. J. (2003). Smaller predator–prey body size ratios in longer
food chains. Proc. R. Soc. Lond. B, 270, 1413–17.
Jennings, S., Dinmore, T. A., Duplisea, D. E., Warr, K. J. & Lancaster, J. E. (2001).
Trawling disturbance can modify benthic production processes. J. Anim. Ecol.,
70, 459–75.
Katona, S. & Whitehead, H. (1988). Are cetacea ecologically important? Oceanogr.
Mar. Biol. Annu. Rev., 26, 553–68.
Koslow, J. A., Boehlert, G. W., Gordon, J. D. M. et al. (2000). Continental slope and
deep-sea fisheries: implications for a fragile ecosystem. ICES J. Mar. Sci., 57,
548–57.
Law, R. (2000). Fishing, selection, and phenotypic evolution. ICES J. Mar. Sci., 57,
659–68.
Lindeman, R. L. (1942). The trophic-dynamic aspect of ecology. Ecology, 23,
399–418.
Martinez, N. D. (1991). Artifacts or attributes? Effects of resolution on the Little
Rock Lake food web. Ecol. Monogr., 61, 367–92.
Merrett, N. R. & Haedrich, R. L. (1997). Deep Sea Demersal Fish and Fisheries.
London: Chapman and Hall.
Myers, R. A. & Worm, B. (2003). Rapid worldwide depletion of predatory fish
communities. Nature, 423, 280–3.
Olsen, E. M., Heino, M., Lilly, G. R. et al. (2004). Maturation trends indicative of
rapid evolution preceded the collapse of northern cod. Nature, 428, 932–5.
Pauly, D. (1980). A new methodology for rapidly acquiring basic information on
tropical fish stocks: growth, mortality and stock-recruitment relationships. In
Stock Assessment for Tropical Small-Scale Workshop, 19–21 September, 1979,
University of Rhode Island, eds. S. Saila & P. Roedel. Kingston, RI: International
Center for Marine Resources Development, pp. 154–72.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, F., Jr (1998a). Fishing
down marine food webs. Science, 279, 860–3.
Pauly, D., Trites, A. W., Capuli, E. & Christensen, V. (1998b). Diet composition
and trophic levels of marine mammals. J. Mar. Sci., 55, 467–81.
Pauly, D., Christensen, V., Guénette, S. et al. (2002). Towards sustainability in
world fisheries. Nature, 418, 689–95.
26 A. W. Trites et al.

Pauly, D., Alder, J., Bennett, E. et al. (2003). The future for fisheries. Science, 302,
1359–61.
Pew Oceans Commission (2003). America’s Living Oceans: Charting a Course for Sea
Change. A Report to the Nation. May 2003. Arlington, VA: Pew Oceans
Commission.
Pimm, S. L. (1982). Food Webs. London: Chapman and Hall.
(1991). The Balance of Nature? Chicago, IL: University of Chicago Press.
Pope, J. G., Shepherd, J. G. & Webb, J. (1994). Successful surf-riding on size
spectra: the secret of survival in the sea. Phil. Trans. R. Soc., 343, 41–9.
Power, G. & Gregoire, J. (1978). Predation by freshwater seals on the fish
community of Lower Seal Lake, Quebec. J. Fish. Res. Board Can., 35, 844–50.
Reiss, M. J. (1989). The Allometry of Growth and Reproduction. Cambridge, UK:
Cambridge University Press.
Rochet, M. J. (1998). Short-term effects of fishing on life-history traits of fishes.
ICES J. Mar. Sci., 55, 371–91.
Sanchez, F. & Olaso, I. (2004). Effects of fisheries on the Cantabrian Sea shelf
ecosystem. Ecol. Model., 172, 151–74.
Springer, A. M., Estes, J. A., Vliet, G. B, van et al. (2003). Sequential megafaunal
collapse in the North Pacific Ocean: an ongoing legacy of industrial whaling?
Proc. Natl Acad. Sci. U. S. A., 100, 12 223–8.
Tamura, T. (2003). Regional assessments of prey consumption and competition by
marine cetaceans in the world. In Responsible Fisheries in the Marine Ecosystem,
eds. M. Sinclair & G. Valdimarsson. Rome, Italy: Food and Agriculture
Organization; Wallingford, UK: CABI Publishing, pp. 143–70.
Tasker, M. L., Camphuysen, C. J., Cooper, J. et al. (2000). The impacts of fishing
on marine birds. ICES J. Mar. Sci., 57, 531–47.
Trippel, E. A. (1995). Age at maturity as a stress indicator in fisheries. Bioscience,
45, 759–71.
Trites, A. W. (1997). The role of pinnipeds in the ecosystem. In Pinniped
Populations, Eastern North Pacific: Status, Trends and Issues, eds. G. Stone,
G. Goebel & S. Webster. Boston, MA: New England Aquarium, Conservation
Department, pp. 31–9.
(2001). Marine mammal trophic levels and interactions. In Encyclopedia of Ocean
Sciences, eds. J. Steele, S. Thorpe & K. Turekian. London: Academic Press,
pp. 1628–33.
(2002). Predator–prey relationships. In Encyclopedia of Marine Mammals, eds,
W. F. Perrin, B. Wursig & H. G. M. Thewissen. San Diego, CA: Academic
Press, pp. 994–7.
(2003). Food webs in the ocean: who eats who and how much? In Responsible
Fisheries in the Marine Ecosystem, eds. M. Sinclair & G. Valdimarsson. Rome,
Italy: Food and Agriculture Organization, Rome; Wallingford, UK: CABI
Publishing, pp. 125–43.
Trites, A. W., Christensen, V. & Pauly, D. (1997). Competition between fisheries
and marine mammals for prey and primary production in the Pacific Ocean.
J. Northw. Atl. Fish. Sci., 22, 173–87.
Trites, A. W., Livingstone, P., Mackinson, S. et al. (1999). Ecosystem Change and the
Decline of Marine Mammals in the Eastern Bering Sea: Testing the Ecosystem Shift
 Effects of fisheries on ecosystems 27

and Commercial Whaling Hypotheses. Fisheries Centre Research Reports 7(1).

Vancouver, Canada: Fisheries Centre.
Trites, A. W., Bredesen, E. L. & Coombs, A. P. (2004). Whales, whaling and
ecosystem change in the Antarctic and Eastern Bering Sea: insights from
ecosystem models. In Investigating the Roles of Cetaceans in Marine Ecosystems.
CIESM Workshop Monographs 25. Monaco: CIESM, pp. 85–92.
US Commission on Ocean Policy (2004). An Ocean Blueprint for the 21st Century.
Final report of the US Commission on Ocean Policy, Washington, DC.
Walters, C. & Kitchell, J. F. (2001). Cultivation/depensation effects on juvenile
survival and recruitment: implications for the theory of fishing. Can. J. Fish.
Aquat. Sci., 58, 39–50.
Walters, C. & Maguire, J. J. (1996). Lessons for stock assessment from the
northern cod collapse. Rev. Fish Biol. Fish., 6, 125–37.
Walters, C. J., Christensen, V. & Pauly, D. (1997). Structuring dynamic models of
exploited ecosystems from trophic mass-balance assessments. Rev. Fish Biol.
Fish., 7, 139–72.
Watson, R. & Pauly, D. (2001). Systematic distortions in world fisheries catch
trends. Nature, 414, 534–6.
Zwanenburg, K. C. T. (2000). The effects of fishing on demersal fish communities
of the Scotian Shelf. ICES J. Mar. Sci., 57, 503–9.
 3
Physical forcing in the southwest Atlantic:
ecosystem control
P . N . T R AT H A N , E . J . M U R P H Y, J . F O R C A D A ,
J . P . C R O XA L L , K . R E I D A N D S . E . T H O R P E

In the southwest Atlantic sector of the Southern Ocean, temporal variability

in the physical environment has been recorded since the early part of the last
century. For example, sea-surface temperature at South Georgia shows peri-
odicity of approximately 3 to 4 years. Variability at South Georgia also reflects
temperature fluctuations in the Pacific, with the Pacific leading South Geor-
gia by approximately 3 years. Increased krill biomass at South Georgia coin-
cides with cold periods. In contrast, periods of reduced predator breeding
performance are strongly correlated with warm anomaly periods, but these
lag behind by a number of months. For some predators the most critical
periods appear to be prior to the breeding season during the summer and
early autumn of the preceding year. Such relationships between predator
breeding performance and the physical environment most probably reflect
prey (krill) availability.

PHYSICAL VARIABILITY IN THE SOUTHERN OCEAN

In the Southern Ocean, phase relationships in physical anomalies (e.g.

Lemke et al. 1980, Zwally et al. 1983) and the movement of such anoma-
lies have been identified and linked with transport by ocean currents (e.g.
Lemke et al. 1980, Jacka & Budd 1991), principally the Antarctic Circumpo-
lar Current (ACC) which connects the major ocean basins in an unbroken
flow. Indeed, it is now thought that the ACC plays an important role in
the transfer of climatic variability between the Pacific, Atlantic and Indian

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Physical forcing in the southwest Atlantic 29

oceans (Murphy et al. 1995, White & Peterson 1996, Connolley 2002,
Trathan & Murphy 2002). Anomalies in sea-ice extent are now thought to
precess eastwards at speeds consistent with the ACC (Murphy et al. 1995).
Similarly, anomalies in other physical factors are also thought to propagate
at speeds consistent with the flow (Jacobs & Mitchell 1996, White & Peter-
son 1996) – including anomalies in atmospheric pressure, wind stress, sea-
surface temperature and sea height. This precession of anomalies has been
described as the Antarctic Circumpolar Wave (ACW) (White & Peterson
1996).
Processes such as that exemplified by the ACW potentially provide an
increased ability to forecast variability over large spatial scales (cf. White &
Cherry 1999). As physical anomalies propagate eastwards with the ACC,
environmental variability should be predictable with lag periods commen-
surate with the flow. For areas where physical variability is a dominant fea-
ture of the ecosystem, the ability to predict change is of ecological and/or
commercial importance (Trathan &Murphy 2002).

IMPACTS ON BIOLOGICAL COMPONENTS

OF THE ECOSYSTEM

In the southwest Atlantic at South Georgia (Fig. 3.1), substantial levels of

both physical and biological variability have been recorded since the early
part of the last century when the island was the centre of a major shore-
based whaling industry. Attempts to understand the high levels of variation
(e.g. Priddle et al. 1988, Fedulov et al. 1996, Murphy et al. 1998, Trathan &
Murphy 2002) have highlighted the need for a better understanding of the
physical environment and, in particular, the phase relationships of physical
and ecosystem attributes.
Early analyses suggested that biological variation was linked with
changes in mean temperature (Harmer 1931, Kemp & Bennet 1932). More
recently, Mackintosh (1972) suggested that variability in the abundance of
krill – a key component in the diet of many of the great whales, seals and
seabirds – was linked with temperature, which in turn was linked with vari-
ability in regional sea-ice and oceanography. Hence, the physical environ-
ment probably drives a large proportion of the observed biological variabil-
ity at South Georgia (Maslennikov & Solyankin 1988, Priddle et al. 1988,
Fedulov et al. 1996), particularly physical variability in the southern portion
of the ACC (Hofmann et al. 1998, Tynan 1998, Nicol et al. 2000, Trathan
et al. 2003).
Fig. 3.1 Map of the Scotia Sea showing South Georgia, the 3000-m isobath, the longitudinal extent of
the region of sea-ice grid cells and the single sea-surface temperature (SST) grid cell at SG1. See Box 3.1
for further explanation.
 Physical forcing in the southwest Atlantic 31

Variability in krill abundance at South Georgia is probably related to the

amount entrained within the flow of the ACC at sites upstream of South
Georgia (Murphy et al. 1998, Brierley et al. 1999), or to levels of spatial
and temporal variability in the transport mechanism itself (the ACC). Such
variability has important consequences for those higher trophic levels that
depend upon krill as prey. At South Georgia a large number of species –
including fish, squid, seabirds and marine mammals – feed upon krill
(Croxall et al. 1985), with some species feeding almost exclusively upon krill
at certain times of the year. Various processes in these predator populations,
including changes in diet composition (Barlow et al. 2002), variability in
foraging-trip duration (McCafferty et al. 1998) and variability in breeding
performance (Lunn & Boyd 1993), are now thought to be related to variabil-
ity in krill biomass and/or availability.
Our increased understanding of the role of ocean currents in transport-
ing krill to South Georgia and of the relationship between sea-surface tem-
perature (SST) and the abundance of krill at South Georgia (Trathan et al.
2003), means that physical variability could potentially be used to predict
responses in higher trophic levels. Therefore, in this chapter we examine
variability in SST at South Georgia, and relate it to variability in the breed-
ing performance of some krill-dependent predators.
To do this, we examine some of the large-scale physical processes affect-
ing the dynamics of the southwest Atlantic Ocean (Box 3.1) and relate the
observed signals to the reproductive performance of predators breeding at
South Georgia (Box 3.2). We consider two large-scale physical processes:
sea-ice and SST.

Box 3.1 Testing for predictability in the physical environment

Understanding how physical processes affect the ecosystem is not

simple as the coupled ice–ocean–atmosphere system in the South-
ern Ocean is complex and relationships change rapidly. Perturba-
tions at South Georgia result from processes that operate at different
spatial and temporal scales; variability occurs over days, as weather
systems move through the area (cf. Priddle et al. 1988), over months
as the seasons change and over years, as warm or cold periods influ-
ence the region (cf. Kemp & Bennet 1932). For our analyses, we used
data that were of a spatial and temporal resolution such that small-
scale, short-lived variability would be integrated, whereas large-scale
and long-lived variability would remain evident.
32 P. N. Trathan et al.

We used the US National Oceanic and Atmospheric Admin-

istration (NOAA) sea-ice dataset (Reynolds et al. 2002) (avail-
able at http://dss.ucar.edu/datasets/ds277.0/data/oiv2/mnly/data);
monthly data at a resolution of 1◦ latitude by 1◦ longitude for the
period December 1981 to April 2003 were used. Multiple grid cells
across the Scotia Sea (Fig. 3.1) were selected; these were located
between 55◦ 30 W and 35◦ 30 W and between 89◦ 30 S and 45◦
30 S. Monthly anomalies in northerly sea-ice extent were calculated
along each meridian using a climatology derived from the long-term
average of each calendar month (for December to April, n = 22;
and for May to November n = 21); the individual meridional series
were then averaged to provide a single Scotia Sea sea-ice anomaly
index.
The SST dataset used was the NOAA operational global
analysis (Reynolds & Smith 1994, Reynolds et al. 2002) (avail-
able at http://dss.ucar.edu/datasets/ds277.0/data/oiv2/mnly/data);
monthly data at 1◦ × 1◦ resolution for the period December 1981 to
April 2003 were used. A single grid cell close to South Georgia (Fig.
3.1) was selected as representative of the region (see Trathan and
Murphy (2002)); this cell was located at 54◦ 30 S, 34◦ 30 W (SG1).
Monthly anomalies were calculated using a climatology derived from
the long-term average of each calendar month (n = 22 or n = 21).

Scotia Sea sea-ice and South Georgia SST

Lagged autocorrelation analyses for sea-ice (Fig. 3.2a) showed sig-
nificant correlation at short lag periods (<6 months) as well as at
longer lag periods; at the longer lags both negative (1 to 2 years) and
positive (3 to 4 years) correlations were evident. This result supports
suggestions (White & Peterson 1996) that sea-ice extent is cyclical
across years, with periods of more extensive sea-ice cover and peri-
ods of reduced ice extent. These periods alternate in a regular man-
ner, with cycles lasting approximately 3 to 4 years. The lagged auto-
correlation analyses for SST (Fig. 3.2b) also showed significant nega-
tive (1 to 2 years) and positive (3 to 4 years) correlation. As with sea-ice
extent, SST showed a cyclical pattern across years, with periodicity
at approximately 3 to 4 years.
Lagged crosscorrelation analyses between the Scotia Sea sea-ice
anomalies and the SST anomalies for SG1 showed significant nega-
tive crosscorrelation (Fig. 3.2c). The strongest correlations occurred
 Physical forcing in the southwest Atlantic 33

(a) (b)
1.0 1.0

Autocorrelation function
Autocorrelation function

0.5 0.5

0.0 0.0

~3 to 4 years ~3 to 4 years

−0.5 −0.5
0 20 40 60 0 20 40 60
Lag (months) Lag (months)

(c) (d)

~0 years
0.4 0.4

Crosscorrelation function
Crosscorrelation function

0.2 0.2

0.0 0.0

−0.2 −0.2
SG1 leads SG1 leads EI Niño leads
Sea-ice leads
~3 years
−0.4 −0.4

−60 −40 −20 0 20 40 60 −60 −40 −20 0 20 40 60

Lag (months) Lag (months)

Fig. 3.2 (a) Lagged autocorrelation function for sea-ice anomalies (northerly
√
extent) for the Scotia Sea with 95% confidence limits ( ± 1.96/ n where n =
256). (b) Lagged autocorrelation function for SST anomalies at SG1 with 95%
√
confidence limits ( ± 1.96/ n where n = 256). (c) Lagged crosscorrelation
function between SST anomalies at SG1 and Scotia Sea sea-ice anomalies
√
(northerly extent) with 95% confidence limits ( ± 1.96/ n where n = 256).
(d) Lagged crosscorrelation function between SST anomalies in the Pacific
El Niño and SST anomalies at the SG1 region with 95% confidence limits
√
( ± 1.96/ n where n = 256).

at short lag periods (<6 months), although significant correlations

also occurred at longer lag periods (3 years), with SST leading
sea-ice.
To test for spurious crosscorrelations, a process of de-trending
and whitening was carried out for the Scotia Sea sea-ice anomalies
and the SST anomalies (Box 3.2). Following this process, signifi-
cant crosscorrelations remained at the lag periods already identified,
although the significance of the relationships was reduced.

South Georgia SST and Southern Ocean teleconnections

Trathan and Murphy (2002) reported strong correlations between
the SST from each of the El Niño regions in the Pacific and
South Georgia, with the El Niño 4 region showing the strongest
34 P. N. Trathan et al.

relationship. The El Niño 4 index was therefore used in these anal-

yses. Lagged crosscorrelation analyses between the El Niño region
and SG1 (Fig. 3.2d) showed the existence of significant correlations;
both positive correlations (2 to 3 years) and negative correlations
(<6 months and 4 to 5 years) were present. Using an analogous pro-
cess to de-trend and whiten the data (Box 3.2), correlations were still
evident, although the significance was somewhat reduced.

Physical forcing at South Georgia

Our analyses show that cyclical patterns occur in the physical envi-
ronment at South Georgia and that these are strongly influenced by
factors and processes that originate outside the region. Results indi-
cate that sea-ice extent and SST are both related to global climate
processes originating elsewhere in the Southern Ocean, namely in
the western Pacific, a region that is also highly variable (Trenberth
& Hoar 1996, Fedorov & Philander 2000). At present, the manner
in which these teleconnections operate and structure the physical
environment at South Georgia remains unclear (Connolley 2002,
Trathan & Murphy 2002), nevertheless, the impacts are unam-
biguous.

Box 3.2 Testing for spurious crosscorrelations

To determine whether spurious crosscorrelations arose from auto-

correlation within each time series, the series were filtered to con-
vert them into approximate white noise (Brockwell & Davis 1991).
For this, an autoregressive Yule–Walker model (Chatfield 2004) was
fitted to each series (xt ) using the Akaike Information Criterion to
choose the autoregressive orders. A filtered series (xt ) was then esti-
mated as


xt = (xt − x) − αr (xt−r − x),
r

where x was the sample mean and α r were the estimated autoregres-
√
sive coefficients of order r. The 95% confidence limits ( ± 1.96/ n
where n = 256) were used to test whether crosscorrelations were
significantly different from zero.
 Physical forcing in the southwest Atlantic 35

SEA-ICE VARIABILITY IN THE SCOTIA SEA

Sea-ice is a dominant feature of the Southern Ocean, affecting the habi-

tat of many different species across a wide range of trophic levels (Brier-
ley & Thomas 2002). Sea-ice, and particularly the marginal ice zone, is
thought to be a key habitat for Antarctic krill (Brierley et al. 2002). Sea-ice
is also highly variable, both within and between years (Zwally et al. 1983,
Jacka & Budd 1991). Our results (Box 3.1) show the existence of tempo-
ral and spatial structure within the regional sea-ice extent, with periodicity
equating to an approximate 3- to 4-year cycle. Previous analyses of sea-ice in
the southwest Atlantic (Murphy et al. 1995) have shown similar patterns of
variability.

SST VARIABILITY AT SOUTH GEORGIA

Spatial and temporal structure also exist in SST at South Georgia, with peri-
odicity equating to an approximate 3- to 4-year cycle (Box 3.1). Previous anal-
yses of SST in this area (Venegas et al. 1997, Trathan & Murphy 2002) have
shown that the correlation at 4-years dominates and is consistent with the 4-
year period reported for the ACW by White and Peterson (1996). However,
Trathan and Murphy (2002) have shown that the period and strength of this
relationship varies through time, being stronger during the late 1980s and
early 1990s when compared with the late 1990s. Our analyses confirm that
this relationship continues to vary and is currently weaker than in previous
years.
Given that there is strong periodicity evident in both sea-ice anoma-
lies and SST anomalies, it is not surprising that strong correlations exist
between the two data series (Box 3.1). This strong relationship between
regional sea-ice and SST is not unexpected, as both datasets reflect the link-
age between the ocean and the atmosphere.

SST RELATIONSHIPS BETWEEN SOUTH GEORGIA AND

T H E E L N I Ñ O 4 R E G I O N I N T H E P A C I F I C

Strong correlations also exist between the SSTs observed at South Georgia
and those found in the Pacific (Box 3.1); the strongest correlations occurred
with the Pacific leading South Georgia. This suggests that SSTs at South
Georgia are related to the Pacific with a lag period of approximately 2 to
3 years, with warmer (colder) water occurring at South Georgia following
warm (cold) events in the Pacific.
36 P. N. Trathan et al.

Trathan and Murphy (2002) suggested that the observed variability in

SST at South Georgia was consistent with transport of anomalies by the
ACC via a process such as the ACW (White & Peterson 1996). As our under-
standing about southern-hemisphere teleconnections increases (White &
Peterson 1996, Christoph et al. 1998, Stössel & Kim 1998) and the manner
in which anomalies propagate into other areas is elucidated (Peterson &
White 1998, Connolley 2002), large-scale variability in the physical environ-
ment at South Georgia will be better understood (Whitehouse et al. 1996,
Brierley et al. 1999, Trathan et al. 2003). This will lead to a better appre-
ciation of the relative roles of local and global ocean–atmosphere signals
and their relative contribution towards meso-scale and large-scale variabil-
ity. Such understanding will also lead to a better appreciation of the causes
and consequences of biological variability in the marine ecosystem.

RELATIONSHIPS BETWEEN SOUTH GEORGIA SST AND

PREDATORS AT BIRD ISLAND

SST reflects a fundamental factor driving biological variability at South

Georgia. This is shown by the consistent relationships between SST and
the breeding performance of both gentoo penguins and Antarctic fur seals
(Box 3.3). For homeotherms such as birds and mammals, this relationship
is unlikely to reflect a physiological cost, but is more likely to reflect variabil-
ity in food supply. Trathan et al. (2003) have shown that the acoustic density
of Antarctic krill at South Georgia is negatively related to temperature, with
acoustic biomass reflecting variability in oceanographic conditions. Trathan
et al. (2003) also showed, that over the past two decades, periods of high krill
biomass have only occurred during anomalously cold periods. This finding
is consistent with the correlations observed here, where we suggest that
reduced levels of prey availability are linked with anomalously warm peri-
ods. If this is the case, then low prey availability must affect the population
processes of both gentoo penguins and Antarctic fur seals.
At South Georgia, observations of the length–frequency distribution of
krill in the diet of Antarctic fur seals (Reid et al. 1999) have shown that
krill recruitment occurs episodically and that cohort growth takes place
over a number of years (Murphy & Reid 2001). Furthermore, as strong
recruitment pulses are evident and dominate the local population, it is
likely (assuming favourable oceanographic transport) that biomass is also
recruitment driven (Murphy & Reid 2001). Thus, it is plausible that warm
SST anomalies (and reduced sea-ice extent) affect krill recruitment and
thus biomass, which in turn affects the productivity of higher trophic
levels.
 Physical forcing in the southwest Atlantic 37

Box 3.3 Biological responses to variability in the physical

environment
The biological data used were from the Bird Island Monitoring Pro-
gramme (BIMP) at South Georgia (Croxall & Prince 1979, Croxall
et al. 1988). Data on the breeding performance of two species were
used, gentoo penguins (Pygoscelis papua) and Antarctic fur seals (Arc-
tocephalus gazella). These time series are among the longest acces-
sible, with data from 1988/9 to 2002/3 available. Both species are
important predators of Antarctic krill (Euphausia superba) with typ-
ically more than 50% krill in their diets (Williams 1991, Staniland
et al. 2003). In addition, both species consume a number of fish
(such as mackerel icefish Champsocephalus gunneri), which in turn
are also consumers of krill (Everson 2000).
Gentoo penguins are resident at South Georgia and rarely travel
far from the island either during chick rearing (Croxall & Prince
1980, Croxall et al. 1988) or during winter (Tanton et al. 2004). Our
analyses used information on the number of nests, the total num-
ber of chicks fledged and breeding success (defined as the average
number of chicks fledged per nest per breeding pair).
Antarctic fur seals are not resident at South Georgia and are typ-
ically thought to migrate out of the region after the breeding season
(Boyd et al. 2002). Fur seal demographic studies are carried out at
Bird Island on a specifically designated beach (Boyd et al. 1995). Dur-
ing the breeding season, the total number of births per day and the
various causes of pup mortality were recorded; from this are derived
the total number of live births per year.

Variability in predator breeding performance

The reproductive performance of gentoo penguins (Fig. 3.3a) and
Antarctic fur seals (Fig. 3.3b) shows high levels of inter-annual vari-
ability, with some years of extremely low reproductive output. The
years of lowest productivity for gentoos, that is all years within the
lowest quartile (1990/1, 1993/4, 1997/8 and 1999/2000), were also
those years when there was low productivity for Antarctic fur seals
(again years within the lowest quartile).

Relationships between South Georgia SST and South Georgia

predators
Lagged crosscorrelation analyses between the monthly SST at SG1
(for the months prior to the breeding season) and gentoo penguin
38 P. N. Trathan et al.

(a)
Nests
Years of extremely low performance
6000 Chicks 2.0
Success
Total number counted
1.5

Breeding success
4000

1.0

2000
0.5

0 0.0
1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003
Year
1989 = 1988–9

(b)
Born
Years of extremely low performance 1000
1000 Died
Survivors

800 800
Total number counted

Total number counted

600 600

400 400

200 200

0 0
1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

Year
1989 = 1988–9

Fig. 3.3 (a) The reproductive performance of gentoo penguins showing the
number of nests, the number of chicks and the breeding success (average
number of chicks fledged per nest per year) of the study population over the
period 1988/9 to 2002/4. (b) The reproductive performance of Antarctic
fur seals showing the number of live births, the number of deaths and the
number of pups surviving in the study population over the period 1988/9
to 2002/3.

productivity (number of chicks fledged) showed that significant neg-

ative correlations were present (Fig. 3.4a). The strongest correlations
were present during the January-to-March period prior to hatching.
After de-trending and whitening (Box 3.2), significant correlations
were still evident, but reduced in significance. Averaging SST over
 Physical forcing in the southwest Atlantic 39

(a) Fledging success

Start of Dec
nesting Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct

1.0 0.5 0.0 0.5 1.0

Crosscorrelation coefficient

(b)

5000

4000
Fledging success

3000

2000

1000

0
0.5 1.0 1.5 2.0
SST (°C)

Fig. 3.4 (a) Crosscorrelation coefficients between monthly SST at SG1 and
gentoo penguin reproductive output (number of chicks fledged). The dashed
lines indicate the 95% confidence intervals. (b) Averaged November-to-January
SST against the number of chicks fledged; the dashed line indicates the fitted
polynomial regression (F3, 11 = 12.24, R2 = 0.7696, p = 0.001; predicted chicks
fledged = 4871.54 – (2412.99 × SST) + (3242.86 × SST2 ) – (1434.23 × SST3 )).

3-monthly intervals and fitting a polynomial regression showed that

the November-to-January period accounted for the highest level of
variance, with 76.9% explained (Fig. 3.4b).
Lagged crosscorrelation analyses between the seasonal SST at
SG1 and Antarctic fur seal pup productivity (number surviving at
40 P. N. Trathan et al.

(a) Pup survival

Start of
pupping Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct

1.0 0.5 0.0 0.5 1.0

Crosscorrelation coefficient

(b)

600
Pup survival

500

400

300

0.5 1.0 1.5 2.0

SST (°C)

Fig. 3.5 (a) Crosscorrelation coefficients between monthly SST at SG1 and
Antarctic fur seal pup survival. The dashed lines indicate the 95% confidence
intervals. (b) Averaged November-to-January SST against the number of pups
surviving; the dashed line indicates the fitted polynomial regression
(F3, 11 = 9.84, R2 = 0.7284, p = 0.002; predicted pup survival =
704.84 – (436.49 × SST) + (510.44 × SST2 ) – (177.98 × SST3 )).

birth) also showed the presence of significant negative correla-

tions (Fig. 3.5a); however, consecutive months showed that these
were variable. After de-trending and whitening (Box 3.2), signif-
icant correlations were still evident, but were reduced in signifi-
cance. Averaging SST over 3-monthly intervals and fitting a polyno-
mial regression showed that the November-to-January period again
 Physical forcing in the southwest Atlantic 41

accounted for the highest level of variance, with over 72.8% ex-
plained (Fig. 3.5b).
The indices of breeding performance of both the predators con-
sidered in this study showed strong relationships with SST in the
period preceding the breeding season. This suggests that the physi-
cal environment has significant impact on the population processes
of these species and that it affects the manner in which they achieve
breeding condition.

Relationships between SST and recruitment have been observed else-

where in the southwest Atlantic, with variability in the recruitment of squid
(Waluda et al. 1999, 2004) and fish (North et al. 1998, North 2004) linked
to SST. Such relationships are not unexpected for planktonic and nektonic
species where temperatures may directly affect juvenile development. It is
perhaps more surprising that such relationships exist for higher trophic
levels. That these occur, suggests that these species are limited by prey avail-
ability, are rather sensitive to fluctuations in prey availability and are possi-
bly operating close to their ecological limit.
An ability to understand the primary factors controlling ecosystem pro-
cesses and ecosystem productivity, and the capacity to predict the future sta-
tus is of clear conservation benefit. Furthermore, given the need for precau-
tionary and sustainable exploitation of harvested stocks in conjunction with
the maintenance of ecological relationships, the ability to make realistic pre-
dictions about the ecosystem is critical to the management process. This
is now particularly the case given future climate-change scenarios (IPCC
2001).
Changes to the pattern of El Niño Southern Oscillation development
since the late 1970s have been emphasized by Trenberth and Hoar (1996),
who related the more frequent occurrence of El Niño events and the less fre-
quent occurrence of La Niña events to decadal changes in climate through-
out the Pacific (see also Fedorov and Philander (2000)). McGowan et al.
(1998) have since shown that such climate–ocean variability perturbs bio-
logical communities along coastal North America. Here we show (Box 3.3)
that Pacific climate–ocean variability also has an impact on population-level
processes in the southwest Atlantic. The emerging consequences of this are
potentially profound; population- and ecosystem-level changes may be rapid
and far-reaching and decadal-scale changes, or regime shifts (Steele 1998,
Hunt et al. 2002), may be evident where population-level processes trigger
ecosystem changes.
42 P. N. Trathan et al.

ACKNOWLEDGEMENTS

We thank all of the Zoological Field Assistants who have helped with the
Bird Island long-term monitoring studies. This work was carried out in sup-
port of the British Antarctic Survey DYNAMOE Core Science Programme
at South Georgia.

REFERENCES
Barlow, K. E., Boyd, I. L., Croxall, J. P. et al. (2002). Are penguins and seals in
competition for Antarctic krill at South Georgia? Mar. Biol., 140, 205–13.
Boyd, I. L., Croxall, J. P., Lunn, N. J., & Reid, K. (1995). Population demography of
Antarctic fur seals: the costs of reproduction and implications for life-histories.
J. Anim. Ecol., 64, 505–18.
Boyd, I. L., Staniland, I. J. & Martin, A. R. (2002). Distribution of foraging by
female Antarctic fur seals. Mar. Ecol. Prog. Ser., 242, 285–94.
Brierley, A. S. & Thomas, D. N. (2002). Ecology of Southern Ocean pack ice. Adv.
Mar. Biol., 43, 171–814.
Brierley, A. S., Demer, D. A., Watkins, J. L. & Hewitt, R. P. (1999). Concordance of
interannual fluctuations in acoustically estimated densities of Antarctic krill
around South Georgia and Elephant Island: biological evidence of same year
teleconnections across the Scotia Sea. Mar. Biol., 134, 675–81.
Brierley, A. S., Fernandes, P. G., Brandon, M. A. et al. (2002). Antarctic krill under
sea ice: elevated abundance in a narrow band just south of ice edge. Science,
295, 1890–2.
Brockwell, P. J. & Davis, R. A. (1991). Time Series: Theory and Methods. New York:
Springer-Verlag.
Chatfield, C. (2004). The Analysis of Time Series: An Introduction. London:
Chapman and Hall/CRC.
Christoph, M., Barnett, T. P. & Roeckner, E. (1998). The Antarctic Circumpolar
Wave in a coupled ocean–atmosphere GCM. J. Clim., 11, 1659–72.
Connolley, W. M. (2002). Long term variability of the Antarctic Circumpolar Wave.
J. Geophys., 108, article 8076.
Croxall, J. P. & Prince, P. A. (1979). Antarctic seabird and seal monitoring studies.
Polar Rec., 19, 573–95.
(1980). The food of Gentoo Penguins Pygoscelis papua and Macaroni Penguins
Eudyptes chrysolophus at South Georgia. Ibis, 122, 245–53.
Croxall, J. P., Prince, P. A. & Ricketts, C. (1985). Relationships between prey
life-cycles and the extent, nature and timing of seal and seabird predation in
the Scotia Sea. In Antarctic Nutrient Cycles and Food Webs, eds. W. R. Siegfried,
P. R. Condy & R. M. Laws. Berlin: Springer-Verlag, pp. 516–33.
Croxall, J. P., McCann, T. S., Prince, P. A. & Rothery, P. (1988). Reproductive
performance of seabirds and seals at South Georgia and Signy Island, South
Orkney Islands, 1976–1987: implications for Southern Ocean monitoring
studies. In Antarctic Ocean and Resources Variability, ed. D. Sahrhage. Berlin:
Springer-Verlag, pp. 261–85.
 Physical forcing in the southwest Atlantic 43

Everson, I. (2000). Role of krill in marine food webs: the Southern Ocean. In Krill:
Biology, Ecology and Fisheries, ed. I. Everson. Oxford UK: Blackwell Science,
pp. 194–201.
Fedorov, V. & Philander, S. G. (2000). Is El Niño changing? Science, 288,
1997–2002.
Fedulov, P. P., Murphy, E. J. & Shulgovsky, K. E. (1996). Environment–krill
relations in the South Georgia marine ecosystem. CCAMLR Sci., 3, 13–30.
Harmer, S. F. (1931). Southern whaling. Proc. Linn. Soc. Lond., Session 142
(1929–30), 85–163.
Hofmann, E. E., Klinck, J. M., Locarnini, R. O., Fach, B. & Murphy, E. J. (1998).
Krill transport in the Scotia Sea and environs. Antarct. Sci., 10, 406–15.
Hunt, G. L., Stabeno, P., Walters, G. et al. (2002). Climate change and control of the
southeastern Bering Sea pelagic ecosystem. Deep-Sea Res. Part II, 49, 5821–53.
IPCC (Intergovernmental Panel on Climate Change) (2001). Climate Change 2001:
The Scientific Basis, eds. J. T. Houghton, Y. Ding, D. J. Griggs et al.
Contribution of Working Group I to the Third Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge, UK: Cambridge
University Press. Online at http://www.ipcc.ch/.
Jacka, T. H. & Budd, W. F. (1991). Detection of temperature and sea ice extent
changes in the Antarctic and Southern Ocean. In International Conference on
the Role of the Polar Oceans in Global Change, eds. G. Weller, C. L. Wilson &
B. A. B. Severin. Fairbanks, Alaska: Geophysical Institute, University of Alaska,
pp. 63–70.
Jacobs, G. A. & Mitchell, J. L. (1996)., Ocean circulation variations associated with
the Antarctic Circumpolar Wave. Geophys. Res. Lett., 23, 2947–50.
Kemp, S. & Bennet, A. G. (1932). On the distribution and movements of whales on
the South Georgia and South Shetland whaling grounds. Discovery Rep., 6,
165–90.
Lemke, P., Trinkl, E. W. & Hasselmann, K. (1980). Stochastic dynamic analysis of
polar sea ice variability. J. Phys. Oceanogr., 10, 2100–20.
Lunn, N. J. & Boyd, I. L. (1993). Effects of maternal age and condition on
parturition and the perinatal period of Antarctic fur seals. J. Zool. Lond., 229,
55–67.
Mackintosh, N. A. (1972). Life cycle of Antarctic krill in relation to ice and water
conditions. Discovery Rep., 36, 1–94.
Maslennikov, V. V. & Solyankin, E. V. (1988). Patterns of fluctuations in the
hydrological conditions of the Antarctic and their effect on the distribution of
Antarctic krill. In Antarctic Ocean and Resources Variability, ed. D. Sahrhage.
Berlin: Springer-Verlag, pp. 209–13.
McCafferty, D. J., Boyd, I. L., Walker, T. R. & Taylor, R. I. (1998). Foraging
responses of Antarctic fur seals to changes in the marine environment. Mar.
Ecol. Prog. Ser., 166, 285–99.
McGowan, J. A., Cayan, D. R. & Dorman, L. M. (1998). Climate–ocean variability
and ecosystem response in the northeast Pacific. Science, 281, 210–7.
Murphy, E. J. & Reid, K. (2001). Modelling Southern Ocean krill population
dynamics: biological processes generating fluctuations in the South Georgia
ecosystem. Mar. Ecol. Prog. Ser., 217, 175–89.
44 P. N. Trathan et al.

Murphy, E. J., Clarke, A., Symon, C. J. & Priddle, J. (1995). Temporal variation in
Antarctic sea-ice: analysis of a long-term fast-ice record from the South Orkney
Islands. Deep-Sea Res., Part I, 42, 1045–62.
Murphy, E. J., Watkins, J. L., Reid, K. et al. (1998). Interannual variability of the
South Georgia marine ecosystem: Biological and physical sources of variation
in the abundance of krill. Fish. Oceanogr., 7, 381–90.
Nicol, S., Pauly, T., Bindoff, N. L. et al. (2000). Ocean circulation off east
Antarctica affects ecosystem structure and sea-ice extent. Nature, 406, 504–7.
North, A. W. (2004). Mackerel icefish size and age at South Georgia and Shag
Rocks. CCAMLR Sci., 11.
North, A. W., White, M. G. & Trathan, P. N. (1998). Interannual variability in the
early growth rate and size of the Antarctic fish Gobionotothen gibberifrons
(Lonnberg). Antarct. Sci, 10, 416–22.
Peterson, R. G. & White, W. B. (1998). Slow oceanic teleconnections linking the
Antarctic Circumpolar Wave with the tropical El Niño-Southern Oscillation.
J. Geophys. Res., 103, 24 573–83.
Priddle, J., Croxall, J. P., Everson, I. et al. (1998). Large-scale fluctuations in
distribution and abundance of krill: a discussion of possible causes. In
Antarctic Ocean and Resources Variability, ed. D. Sahrhage. Berlin:
Springer-Verlag, pp. 169–82.
Reid, K., Watkins, J. L., Croxall, J. P. & Murphy, E. J. (1999). Krill population
dynamics at South Georgia 1991–1997, based on data from predators and nets.
Mar. Ecol. Prog. Ser., 177, 103–14.
Reynolds, R. W. & Smith, T. M. (1994). Improved global sea surface temperature
analyses using optimum interpolation. J. Clim. Res., 7, 929–48.
Reynolds, R. W., Rayner, N. A., Smith, T. M., Stokes, D.C. & Wang, W. Q. (2002).
An improved in situ and satellite SST analysis for climate. J. Clim., 15, 1609–25.
Staniland, I. J., Taylor, R. I. & Boyd, I. L. (2003). An enema method for obtaining
fecal material from known individual seals on land. Mar. Mamm. Sci., 19,
363–70.
Steele, J. H. (1998). Regime shifts in marine ecosystems. Ecological Applic., 8,
S33–6.
Stössel, A. & Kim, S.-J. (1998). An interannual Antarctic sea-ice–ocean model.
Geophys. Res. Lett., 25, 1007–10.
Tanton, J. L., Reid, K., Croxall, J. P. & Trathan, P. N. (2004). Winter distribution
and behaviour of gentoo penguins Pygoscelis papua at South Georgia. Polar
Biol., 27, 299–303.
Trathan, P. N. & Murphy, E. J. (2002). Sea surface temperature anomalies near
South Georgia: relationships with the Pacific El Niño regions. J. Geophys Res.,
108, article 8075.
Trathan, P. N., Brierley, A. S., Brandon, M. A. et al. (2003). Oceanographic
variability and changes in Antarctic krill (Euphausia superba) abundance at
South Georgia. Fish. Oceanogr., 12, 569–83.
Trenberth, K. E. & Hoar, T. L. (1996). The 1990–1995 El Niño Southern
Oscillation event: longest on record. Geophys. Res. Lett., 23, 57–60.
Tynan, C. T. (1998). Ecological importance of the Southern Boundary of the
Antarctic Circumpolar Current. Nature, 392, 708–10.
 Physical forcing in the southwest Atlantic 45

Venegas, S. A., Mysak, L. A. & Straub, D. N. (1997). Evidence for interannual and
interdecadal climate variability in the South Atlantic. Geophys. Res. Lett., 23,
2673–6.
Waluda, C. M., Trathan, P. N. & Rodhouse, P. G. (1999). Influence of
oceanographic variability on recruitment in the Illex argentinus (Cephalopoda:
Ommastrephidae) fishery in the South Atlantic. Mar. Ecol. Prog. Ser., 183,
159–67.
(2004). Synchronicity in southern hemisphere squid stocks and the influence of
the Southern Oscillation and Trans Polar Index. Fish. Oceanogr., 13, 1–12.
White, W. B. & Cherry, N. J. (1999). The influence of the Antarctic circumpolar
wave upon New Zealand temperature and precipitation during
autumn–winter. J. Clim., 12, 960–76.
White, W. B. & Peterson, R. G. (1996). An Antarctic circumpolar wave in surface
pressure, wind, temperature and sea-ice extent. Nature, 380, 699–702.
Whitehouse, M. J., Priddle, J. & Symon, C. (1996). Seasonal and annual change in
seawater temperature, salinity, nutrient and chlorophyll a distributions around
South Georgia, South Atlantic. Deep-Sea Res., 43, 435–43.
Williams, T. D. (1991). Foraging ecology and diet of gentoo penguins Pygoscelis
papua at South Georgia during winter and an assessment of their winter prey
consumption. Ibis, 133, 3–13.
Zwally, H. J., Parkinson, C. L. & Comiso, J. C. (1983). Variability of Antarctic sea ice
and changes in carbon dioxide. Science, 220, 1005–12.
 4
The use of biologically meaningful
oceanographic indices to separate the effects of
climate and fisheries on seabird breeding success
B. E. SCOTT, J. SHARPLES, S. WANLESS, O. N. ROSS,
M . F R E D E R I K S E N A N D F . D AU N T

An important issue when considering seabird breeding success is what fac-

tors affect prey availability. If availability reflects absolute prey abundance,
different species preying on the same prey population should show syn-
chronized variation in breeding success. If, on the other hand, species-
specific foraging techniques coupled with prevailing oceanographic condi-
tions result in differential access to prey, then, breeding success is likely to
vary asynchronously between species. Furthermore, for each species, long-
term variation in breeding success should be predictable using appropriate
oceanographic covariates. Currently, commercial fishing quotas are set on
the assumption that prey abundance is the only important factor for multi-
species management. Therefore, it is essential to understand prey availabil-
ity in the context of both climate change and fishing pressure. This requires
an integrated approach and in this chapter we demonstrate the potential of
combining long-term demographic data from seabirds with output from a
one-dimensional physical–biological model. Using data from the North Sea,
we examine relationships between breeding performance and biologically
meaningful indices of the physical environment during a period of years
with and without an industrial fishery. We speculate how the contrasting
responses shown by two seabird species might reflect differences in prey
availability mediated by foraging technique.
Over the last 20 to 30 years, seabirds in the North Sea have shown con-
siderable temporal variability in breeding success (Ratcliffe 2004). These

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Biologically meaningful oceanographic indices 47

changes have frequently been attributed to variation in feeding conditions,

in particular availability of lesser sandeels (Ammodytes marinus), the prin-
cipal prey of many seabirds during the breeding season (Furness & Tasker
2000). In general, surface-feeding species such as the black-legged kitti-
wake (Rissa tridactyla) have been more severely affected than diving species
such as the common guillemot (Uria aalge) (Monaghan 1992). The lesser
sandeel is also the target of the largest single-species fishery in the North
Sea, and the impact of industrial fishing on seabird populations has been
a major conservation and fisheries issue (Furness 2002). More recently,
attention has shifted to the potential impact of climate change on the
North Sea ecosystem and how this might disrupt predator–prey relation-
ships (Edwards & Richardson 2004). One of the best-studied parts of the
North Sea is the area around the Firth of Forth off the coast of southeast
Scotland. An industrial fishery has operated in this area for some of the
time that data on seabird demography and environmental conditions have
been collected (Rindorf et al. 2000). This area is thus an ideal setting, not
only for investigating relationships between predator performance and the
physical environment, but also for separating out the effects of climate and
fisheries.

NORTH SEA SEASONAL OCEANOGRAPHIC CYCLE

Oceanographic conditions prevailing during the time of seabird reproduc-

tion (April to July) are most likely to exert a direct influence on the availabil-
ity of seabird prey and hence the birds’ breeding success. It is thus neces-
sary to consider the timing of seasonal events in the North Sea. As a shal-
low sea less than 200 m in depth, the seasonal cycle of the North Sea is
relatively simple to understand and model in oceanographic terms (Otto
et al. 1990, Turrell 1992). The North Sea’s physical characteristics are dom-
inated by tides, winds and solar radiation (Otto et al. 1990). During the win-
ter months, the lower levels of radiation together with the stronger winds
and tidal friction leave the water column completely mixed. In the spring,
increasing amounts of sunlight and less windy conditions allow a decrease
in vertical mixing. In those areas that are deep enough or have weaker tidal
currents such that the effect of tidal mixing does not reach the surface, the
surface layer begins to warm up (Pingree et al. 1978, Mann & Lazier 1996).
This warming creates a difference in density between the upper and lower
layers of the water column called ‘stratification’ (see glossary in Box 4.1). The
onset of stratification allows plankton to remain above the ‘critical depth’
(see Box 4.1) needed for population growth. Consequently, the timing of
48 B. E. Scott et al.

Box 4.1 Glossary

Bank regions. Large banks in the seabed topography off southeast
Scotland, typically rising 20 to 40 m above the surrounding seabed,
and measuring 10 to 30 km east–west and 50 to 100 km north–south.
Critical depth. If phytoplankton are continuously mixed between the
sea surface and the critical depth, the light energy they receive is
just sufficient to compensate for respiratory losses. If they are mixed
in a region shallower than the critical depth, then growth exceeds
respiratory losses and biomass can increase. If they are mixed deeper
than the critical depth, then respiratory losses exceed growth and
phytoplankton begin to die.
Primary production. The growth of phytoplankton in the ocean.
Phytoplankton are single-celled plants, typically between 5 and
100 µm in size, and requiring both sunlight and nutrients in order
to photosynthesize and grow. They are the ocean’s primary produc-
ers, forming the base of the marine food chain.
Shallow sea front regions. These are also known as ‘tidal mixing
fronts’ or ‘shelf-sea fronts’. These fronts separate areas of shelf sea
that are permanently vertically mixed (shallow water and/or strong
tidal currents) from areas that thermally stratify during summer
(deeper water and/or weaker tidal currents). They mark the bound-
ary where the tendency towards summer stratification driven by
solar heating is just countered by the tendency to redistribute heat
through the water column by tide-induced turbulent mixing.
Spring bloom. As the solar irradiance increases in spring there is
more light available for phytoplankton photosynthesis and more
heat available for stratifying the water column. If the tendency
towards stratification is able to overcome the mixing by tides and
winds, the development of a warm surface layer isolates some phyto-
plankton, along with dissolved nutrients, in the surface layer. The
stratification prevents these phytoplankton from being mixed into
the deeper, darker water, and with ample light and nutrients they
grow (bloom) rapidly. The bloom peaks quickly, but while the light
in the surface layer continues to increase the nutrients are used up
and cannot be easily re-supplied from the deeper water because of
the inhibiting effect that the stratification has on mixing. The phyto-
plankton become nutrient-limited and, along with losses to grazing
by herbivorous zooplankton, the bloom decays.
 Biologically meaningful oceanographic indices 49

Stratification. A water column is stratified when the density of the

water has some vertical variability. This could be because the sur-
face water has been warmed (reducing its density compared with the
deeper water), and/or because the surface water has a lower salinity.
Stratification inhibits vertical mixing of heat, nutrients, phytoplank-
ton, etc. and is a key process in controlling the light and nutrient
environments experienced by phytoplankton.

stratification is generally believed to herald the beginning of the seasonal

flush of ‘primary production’, referred to as the ‘spring bloom’ (see Box 4.1)
(Miller 2004).

VARIABILITY IN THE SPRING BLOOM

The variability both in timing of the spring bloom and in the seasonal cycle
of primary production at a given location in the North Sea is driven by the
degree of mixing of the water column (Pingree et al. 1975, Simpson 1981,
Fèvre 1986). Therefore, as the depth and speed of tides at any location are
predictable (Pingree et al. 1978, Simpson & Bowers 1981), the variation in
mixing, and hence primary production, is due to the inter-annual differ-
ences in the amount of wind, radiation and freshwater input received at that
location. Thus, local meteorological forcing, such as daily wind speeds and
the amount of sunlight and rain, drive variation in the timing and amount
of production at the lowest trophic levels.

THE SPRING BLOOM AS AN INDICATOR

Almost 100 years ago, a hypothesis was formulated that the timing of
the spring bloom, and therefore the availability of appropriate food, would
greatly influence the survival of larval fishes (Hjort 1914). This idea was
expanded upon by Cushing (1975), who coined the match–mismatch theory
stating that high survival of fish larvae is expected in those years when the
timing of spawning and hatching is such that larvae overlap appropriately
with the timing of the spring bloom. Only recently has a study confirmed
that fish recruitment does indeed increase when such an overlap occurs
(Platt et al. 2003). However, the lack of support for the match–mismatch
theory does not stem from a scarcity of studies addressing this question.
Instead, it reflects the difficulties associated with sampling marine ecosys-
tems repeatedly over appropriate temporal and spatial scales required to
50 B. E. Scott et al.

simultaneously establish the timing of the spring bloom and estimate its
effect on fish survival and growth.
Although the environmental features that trigger spring blooms have
long been well understood in a general sense (Mann & Lazier 1996,
Miller 2004), it is only recently that physical oceanographic modelling
has advanced sufficiently to capture accurately the biological dynamics of
these events at the temporal and spatial scales appropriate to the feed-
ing behaviour of individual animals (Franks 1992, Sharples 1999, Waniek
2003). These types of models, in particular the one-dimensional physical–
biological coupled model of Sharples (1999), allow the monitoring, in one
location, of daily or hourly changes in vertical structure of the water col-
umn and the amount of primary production arising at any given depth at
that location. However, marine ecologists are still some way from under-
standing the impact of between-year variation in the seasonal production
cycle on higher trophic levels. This is because it is prohibitively expensive
to continuously and simultaneously sample phytoplankton, zooplankton,
and larval and adult fish. Therefore, a way to improve our understanding of
marine-ecosystem functioning is to combine the quantitative predictions of
these coupled physical–biological models with concurrent measurements
of the foraging behaviour and breeding success of highly visible top preda-
tors such as seabirds. If the top predators can be shown to be good inte-
grators of important signals being amplified as they move up the trophic
levels, then we will have more reliable and immediate indicators of the cur-
rent state of the ecosystem (Bertram et al. 2001, Gjerdrum et al. 2003).

OCEANOGRAPHY OF THE STUDY AREA AND

REGION-SPECIFIC ONE-DIMENSIONAL
PHYSICAL–BIOLOGICAL MODELLING

Our study area (Firth of Forth, 55◦ 30 to 57◦ N, 3◦ W to 0◦ 30 E) contains
two of the hydrographic regions found within the North Sea (Otto et al.
1990): ‘Bank regions’ and ‘Shallow sea front regions’ (see Box 4.1). Both
these water types are important foraging areas for seabirds breeding on the
Isle of May, one of the main colonies in the area (see Daunt et al. (Chapter
12 in this volume) and Camphuysen et al. (Chapter 6 in this volume), for a
description of the foraging distributions of these seabirds).
We used a one-dimensional physical–biological coupled model
(Sharples 1999, see Box 4.2) to derive the inter-annual variability in
the seasonal primary production patterns within these two regions (see
Box 4.3). The one-dimensional model was parameterized using daily,
 Biologically meaningful oceanographic indices 51

Box 4.2 The one-dimensional physical--biological model

The one-dimensional physical–biological coupled numerical model
is based on that of Sharples (1999) and Sharples et al. (in press).
The physical component of the model is driven by tidal forcing, sur-
face heating and surface winds, calculating the vertical water column
structure of currents, temperature (i.e. stratification) and light. A tur-
bulence closure scheme (Canuto et al. 2001) is used to calculate the
rates of turbulent mixing driven by tidal and wind stresses. The bio-
logical component calculates the response of a single phytoplankton
species (in terms of chlorophyll concentration) to the light and nutri-
ent environment, with the turbulent mixing controlling the vertical
fluxes of phytoplankton and dissolved inorganic nutrients.
The model has been re-written with a graphical interface, allow-
ing user input of all physical, chemical and biological parameters
required to drive the modelled processes. The model was initially
calibrated using the current- and temperature-profile information
provided by the moorings at the two sites between March and July
2001, yielding a reliable agreement between modelled and observed
tidal currents, vertical temperature structure and primary produc-
tion. Meteorological information from 1974 to 2003 (see Box 4.3)
was then used to calculate physical and biological time series over
the 30-year period.

Box 4.3 Collection of fine-scale oceanographic, meteorological

and bird breeding success data

Moorings
In order to monitor at fine temporal and vertical resolution and
to collect the data needed to parameterize the one-dimensional
physical–biological model for the study area, moorings were placed
in two regions in which seasonal production cycles were expected
to differ: the bank region (depth 45 m, 56◦ 15 N, 02◦ 00 W) and
the shallow sea front region (depth 65 m, 56◦ 15 N, 01◦ 15 W).
The moorings provided information, at a 10-min resolution, on the
changes in vertical structure (at 5- to 10-m intervals), such that it
was possible to define the depth of the surface mixed layer and
52 B. E. Scott et al.

the strength of the thermocline at any point in time. The mooring

in the shallow sea front had two current meters, one fluorometer
and eight mini-loggers (temperature recorders). The mooring in the
bank region had one current meter, one fluorometer and seven mini-
loggers. Each mooring was in operation from March to October for
both 2001 and 2002.
Meteorological data
The most appropriate daily meteorological data for the study area,
as needed to run the one-dimensional model, were collected from
the Leuchars and Mylnefield meteorological stations, Scotland, UK.
The database consists of hourly and daily weather indices for the last
30 years including wind speed, wind direction, irradiance, dew-point
temperature and freshwater input (rain and river runoff). The data
were obtained from the British Oceanographic and Atmospheric
Data Centres (BODC and BADC).
Seabird breeding success
Standardized data on seabird breeding success were collected at
the Isle of May, southeast Scotland (56◦ 11 N, 02◦ 34 W). Breed-
ing success, measured as the mean number of fledged chicks per
pair, was estimated from 1982 to 2003 for the common guillemot (a
pursuit-diver) and from 1985 to 2003 for the black-legged kittiwake
(a surface-feeder ).

local meteorological information and can therefore be used to recreate

oceanographic conditions as far back in time as such data are available
(see Box 4.2 and Box 4.3). The daily output of the model (Fig. 4.1) includes
top and bottom temperatures and chlorophyll levels, as well as numerous
other physical and biological variables, and allows calculation of the annual
timing of the onset of stratification (taken as the time when the difference
between top and bottom temperature exceeds 0.5 ◦ C) and the timing of the
start of the spring bloom (when chlorophyll levels exceed 2 mg m−3 ).

SANDEEL LIFE-HISTORY STAGES AND SEASONAL

OCEANOGRAPHIC CYCLE

Primary production during the spring bloom provides food for zooplank-
ton populations which, in turn, are the main food source of sandeels
(Covill 1959, Monteleone & Peterson 1986). Therefore the timing, length
 Biologically meaningful oceanographic indices 53

16 Surface temperature 14
Bottom temperature
14 12

Chlorophyll (mg m−3)

Surface chlorophyll
Temperature ( C)

10
12
o

Start of spring bloom

8
10 Start of stratification
6
8
4

6 2

4 0
1 31 61 91 121 151 181 211 241 271 301 331 361
Julian date

Fig. 4.1 An example of daily output of surface chlorophyll, surface temperature

and bottom temperature from the one-dimensional physical–biological model for
1992. The solid arrow marks the timing of the start of the spring bloom, defined
as the date when chlorophyll levels exceed 2 mg m−3 and stay above that value for
five consecutive days. The dashed arrow marks the timing of the start of
stratification, defined as the date when the difference between top and bottom
temperatures exceeds 0.5 ◦ C and stays above that value for five consecutive days.

and intensity of the bloom are all potentially important factors in determin-
ing food availability and hence the time spent foraging by larval, juvenile
and adult sandeels. Sandeels in the North Sea spend the vast majority of
their life buried in the sand (Reay 1970, Winslade 1974a, 1974b, 1974c,
Pearson et al. 1984). In the Firth of Forth area they may only come out of the
sands to feed between April and September (Worsøe 1999). In addition, the
breeding component of the population emerges to spawn in late Decem-
ber and early January, and the eggs hatch by late February (Reay 1970,
Winslade 1974b). Once hatched, the distance over which the larvae may
be advected from spawning locations appears to be variable and dependent
on wind speeds, wind directions and how fast the larvae attain the ability to
make vertical migrations (Proctor et al. 1998, Munk et al. 2002, Jensen et al.
2003).
As soon as sandeels leave the protection of the sands and forage within
the water column, they are subject to predation by a wide range of preda-
tors – such as larger fish (Greenstreet (Chapter 15 in this volume)), seabirds
(Daunt et al. (Chapter 12 in this volume)) and marine mammals. Therefore,
for sandeels to leave the substrate, the gain from food intake must override
predation and starvation risks. In fact, it has been shown experimentally
that low food availability significantly increased the time sandeels remained
54 B. E. Scott et al.

buried in the sand (Winslade 1974a). Thus, it is reasonable to assume that

the timing of the spring bloom will influence the timing of emergence of
adult sandeels as well as the growth and survival of larvae and juveniles.

THE SANDEEL FISHERY

An industrial sandeel fishery targeting predominately adult sandeels was

in operation in the Firth of Forth from 1990 to 1999. The fishery oper-
ated mostly in June but also extended into May and July in some years.
Total annual catches ranged from 20 000 to over 100 000 t (Rindorf et al.
2000). The fishery was closed in 2000, but a catch of 3000 to 4600 t has
been allowed each year up to the present (2004) for scientific purposes
(P. Wright, FRS Marine Laboratory, Aberdeen, personal communication).

SEABIRD BREEDING SUCCESS IS LINKED TO

AVAILABILITY OF SANDEELS VIA THE SPRING BLOOM

The breeding season is the most energetically demanding part of the seabird
life cycle, and a successful outcome is critically dependent on the availabil-
ity of sufficient amounts of high-quality food. If the initiation of the annual
increase in primary production is the driving factor for the emergence of
adult sandeels, timing of the spring bloom may be very important for the
birds. The availability of adult sandeels at the right time is important in the
early stages of the breeding season (egg laying and incubation) and thus may
be a critical factor determining annual breeding success in seabirds (ICES
2004). During the chick-rearing period (typically in June), adult sandeels
seemingly become less available as they disappear out of the diets of both
kittiwakes and guillemots (Harris & Wanless 1985, Lewis et al. 2001). A
likely explanation is that adult sandeels spend more time in the sands as the
availability of their own food is declining; by this time of the year, primary
production is falling rapidly due to the lack of free nutrients for phytoplank-
ton growth (Miller 2004). Therefore, the birds must now depend more on
juvenile sandeels and other prey species (such as sprat Sprattus sprattus) to
feed their chicks and themselves. Spring conditions and their effect on the
timing of primary production will have influenced the growth and survival
of juvenile fish. It is therefore reasonable to assume that fledging success
is also influenced by the timing and location of spring blooms. In short,
spring-bloom timing is expected to influence all components of breeding
success.
 Biologically meaningful oceanographic indices 55

TIMING OF STRATIFICATION AND THE SPRING BLOOM

IN THE BANK AND SHALLOW SEA FRONT REGIONS

To investigate annual variability in the timing of stratification and the spring

bloom for the two regions, we ran the one-dimensional physical–biological
coupled model over 30 years (1974–2003). In the bank region, we found
that the mean date on which the spring bloom started was 19 April with a
standard deviation of only 4.1 days. This constancy was maintained despite
the amount of wind mixing in the weeks leading up to the bloom varying
by an order of magnitude between years (mean amount of force of mixing
from winds, wind stress, for March ranges from 0.02 to 0.20 N m−2 ). The
mean start date for the bloom in the shallow sea front region was similar
(21 April ± 5.4 days) and dates in the two regions were highly correlated
(rp = 0.74, p < 0.001).
Differences in water depth in the two regions resulted in consistent dif-
ferences in the timing of the spring bloom relative to stratification. Thus
in the shallower bank regions, the critical depth required for the modelled
phytoplankton to achieve net growth is greater than the water-column depth
and, on average, the spring bloom occurs 9 days before stratification (mean
stratification date 28 April ± 12.3 days). In contrast, in the deeper shallow
sea front region the spring bloom occurred at the same time as the water
column became stratified (21 April ± 11.4 days). Any annual difference in
the date of the spring bloom between the bank and the shallow sea front
regions is therefore linked to the greater dependence on the development
of stratification within the deeper region. Because stratification is not a pre-
requisite for growth in the shallower bank region, the initiation of the bloom
is more simply related to increasing solar irradiance. This phenomenon has
also been found in the southern North Sea (Haren et al. 1998) and may well
be a feature of bank regions throughout the North Sea. The reliance on sea-
sonal solar radiation levels for the timing of the spring bloom, in either
region, explains the consistency of bloom dates and their proximity to the
spring equinox.

SEABIRD BREEDING SUCCESS, TIMING OF THE SPRING

BLOOM AND THE SANDEEL FISHERY

We compared the breeding success of kittiwakes and guillemots on the Isle

of May to the timing of the spring bloom and stratification in the bank and
the shallow sea front regions (see Box 4.3). For both species, results were
similar for the two oceanographic regions and for both indices of timing.
56 B. E. Scott et al.

(a)
Effect of spring-bloom date
Kittiwakes (F1,16 = 6.20, P = 0.02)
1.4
Breeding success (chicks per pair) Effect of fishery
(F1,16 = 44.92, P < 0.001)
1.2
No interaction
1.0 (F1,16 = 2.40, P = 0.14)

0.8

0.6

0.4

0.2

0
95 100 105 110 115 120 125
Spring bloom in bank region (Julian date)

(b)

0.90
Breeding success (chicks per pair)

Guillemots
0.85

0.80

0.75

0.70

0.65

0.60
95 100 105 110 115 120 125
Spring bloom in bank region (Julian date)

Fig. 4.2 Breeding success of (a) black-legged kittiwakes (1985–2003) and (b)
common guillemots (1982–2003) on the Isle of May in relation to the start date
of the spring bloom in the bank region, as estimated by the one-dimensional
physical–biological model. Years with no commercial fishery for sandeels are
represented by filled squares and years with a fishery with open squares.

To avoid unnecessary repetition, we therefore present quantitative results

for the bank region and the timing of spring bloom only. Kittiwakes bred
more successfully in the years when the spring bloom and stratification in
either region occurred later. Breeding success increased by 0.13 chicks per
pair for every 5 days delay in timing of the spring bloom (Fig. 4.2). There
 Biologically meaningful oceanographic indices 57

was no evidence that the effect of date of the spring bloom on breeding
success was different in fishing and non-fishing years (interaction: F1,16 =
2.40, p = 0.14), but breeding success was 0.33 ± 0.05 (mean ± SE) chicks
higher in years without fishing than in years with fishing (F1,16 = 44.92,
p < 0.001). The final model containing both the effects of spring-bloom date
and the sandeel fishery explained 74% (p < 0.001) of the variance in breed-
ing success. Separating years with or without a fishery, the effect of climate
alone explained 56% of the variance in breeding success in years without a
fishery and 10% of the variance in years with a fishery. This suggests that
important climatic variables are more easily identified in the absence of the
confounding effects of a fishery.
A similar analysis carried out for guillemots, revealed that neither the
timing of the spring bloom, nor stratification in either oceanographic
region, had a significant effect on breeding success (Fig. 4.2).

DISCUSSION

With the use of a detailed one-dimensional physical–biological model we

have shown how integrated and biologically meaningful region- and year-
specific oceanographic variables can provide new insights into the mech-
anisms that link multiple meteorological conditions with seabird breed-
ing success. In particular, we have used the timing of stratification and
the spring bloom as possible indicators of sandeel availability. While pre-
vious studies have demonstrated statistical relationships between larger-
scale oceanographic or climatic variables and seabird performance (e.g.
Aebischer et al. 1990), our approach allows identification of proximate fac-
tors that directly affect seabird prey availability. In a system where exper-
imental manipulation is impossible, this type of model is an extremely
promising tool in the search for causal relationships among ecosystem
components.
Information on the breeding success for seabirds with contrasting for-
aging strategies in an area where an industrial sandeel fishery has operated
intermittently, and which has experienced large climatic changes (Edwards
et al. 2002), provides a rare opportunity to investigate the effects of both
climate and fishing activities. Our results suggest that breeding success of
kittiwakes is higher in years when the spring bloom occurs relatively late
throughout the study area. Because kittiwakes are surface-feeders and are
more dependent on prey being present in the upper water column, this
finding leads us to speculate that a later spring bloom increases the avail-
ability of prey in the upper water column during the breeding season. We
58 B. E. Scott et al.

suggest that the mechanism by which this is achieved is through a slowing

down in the growth of juvenile sandeels, possibly coupled with a delay in
the emergence of adult sandeels. With elevated levels of food occurring rel-
atively late in the season, the trade-off between predation and feeding may
force fish to take longer to acquire adequate levels of food needed to survive
the next winter. Either the slowing of fish growth, or the extension of the
time window during which sandeels are feeding actively in the water col-
umn, could potentially provide an increase in prey availability, especially to
surface-feeders.
These results accord well with those of Rindorf et al. (2000) who found
that breeding success of Isle of May seabirds was higher when sandeel
abundance peaked later in the season. They also help interpret the find-
ings of Frederiksen et al. (2004) that kittiwake breeding success at the Isle
of May was reduced when local winter sea temperature was high in the pre-
vious year. There is a weak negative correlation between winter sea tem-
perature and the timing of the spring bloom in the current as well as the
previous year (rp = −0.22 −0.27 respectively). Although spring-bloom tim-
ing here is modelled rather than measured, it is a more proximate and
thus a much more interpretable correlate of breeding success than a single
weather variable such as sea temperature. Using this biologically meaning-
ful and locally specific index we have confirmed the conclusion of Frederik-
sen et al. (2004) that, under similar climatic conditions, kittiwakes suffer
an additional reduction in breeding success in years when a commercial
sandeel fishery is operating. Our approach allows us to quantify fishing and
climate effects separately and indicates that the presence of a local sandeel
fishery decreases breeding success of Isle of May kittiwakes by 0.33 chicks
per pair whereas every 5-day delay in the date of the spring bloom increases
breeding success by 0.13 chicks per pair.
For guillemots, annual variation in breeding success was not explained
by the timing of spring blooms, stratification or the presence of a sandeel
fishery (Fig. 4.2). Guillemots are pursuit-divers and are therefore less con-
strained in their foraging depths than kittiwakes (see Daunt et al. (Chapter
12 in this volume)). They thus have access to sandeels in the whole water
column and are probably less likely to encounter food limitation. This con-
trast in how oceanographic factors affect the mechanism of prey availability
helps explain why annual variations in breeding success of different species
at the same colony may not be in phase (guillemot and kittiwake breeding
success were not significantly correlated: rp = 0.26, p > 0.2), and also why
breeding success fluctuates less from year to year for guillemots than for
kittiwakes (Fig. 4.2).
 Biologically meaningful oceanographic indices 59

Although we have found here that for some species breeding success
is linked to annual variation in the timing of spring blooms, seabird pop-
ulation growth is also affected by other demographic parameters. Indeed,
because seabirds are long-lived, population growth rate is most sensitive
to variation in adult annual survival (Croxall & Rothery 1991). Outside the
breeding season, Isle of May seabirds range much more widely than our
study area, in some cases over the entire North Atlantic Ocean. Seabirds
only recruit into the breeding population when they are several years old,
and during the pre-breeding period they range even more widely than
adults. As encouraging as our present results are, identifying, measuring
and modelling oceanographic variables at the appropriate spatial and tem-
poral scale to understand interactions between seabird survival and recruit-
ment still presents a major challenge.
More studies of this kind involving the use of one-dimensional physical–
biological models as tools for connecting past and predicted changes in
climate to higher trophic levels will bring us closer to identifying critical
linkages within ecosystems. In a constantly changing environment, where
future climate change is likely to have profound consequences for marine
ecosystems, these models could prove invaluable tools for understanding
and predicting impacts on higher trophic levels.

ACKNOWLEDGEMENTS

This work was funded by the European Commission project ‘Interactions

Between the Marine Environment, Predators and Prey: Implications for
Sustainable Sandeel Fisheries (IMPRESS; QRRS 2000-30864)’. We thank
Mike Harris for establishing the long-term seabird studies, the Joint Nature
Conservation Committee for funding under their Seabird Monitoring
Programme, Scottish Natural Heritage for access to the Isle of May, Fish-
ery Research Services Marine Laboratory Aberdeen – in particular Simon
Greenstreet, Mike Heath, Helen Fraser, Gayle Holland, Sarah Hughes, John
Dunn, George Slessor – and the crew of the HMV Clupea for support with
collection and analysis of the mooring data.

REFERENCES
Aebischer, N. J., Coulson, J. C. & Colebrook, J. M. (1990). Parallel long-term trends
across four marine trophic levels and weather. Nature, 347, 753–5.
Bertram, D. F., Mackas, D. L. & McKinnell, S. M. (2001). The seasonal cycle
revisited: interannual variation and ecosystem consequences. Prog. Oceanogr.,
49, 283–307.
60 B. E. Scott et al.

Canuto, V. M., Howard, A., Cheng, Y. & Dubovikov, M. S. (2001). Ocean

turbulence. Part I: one-point closure model – momentum and heat vertical
diffusivities. J. Phys. Oceanogr., 31, 1413–26.
Covill, R. W. (1959). Food and feeding habits of larvae and postlarvae of Ammodytes
americanus. Bull. Bingham Oceanogr. Coll., 17, 125–46.
Croxall, J. P. & Rothery, P. (1991). Population regulation of seabirds: implications
of their demography for conservation. In Bird Population Studies: Relevance to
Conservation and Management, eds. C. M. Perrins, J.-D. Lebreton & G. J. M.
Hirons. Oxford, UK: Oxford University Press, pp. 272–96.
Cushing, D. H. (1975). Marine Ecology and Fisheries. Cambridge, UK: Cambridge
University Press.
Edwards, M. & Richardson, A. J. (2004). Impact of climate change on marine
pelagic phenology and trophic mismatch. Nature, 430, 881–4.
Edwards, M., Beaugrand, G., Reid, P. C., Rowden, A. A. & Jones, M. B. (2002).
Ocean climate anomalies and the ecology of the North Sea. Mar. Ecol. Prog.
Ser., 239, 1–10.
Franks, P. J. S. (1992). New models for the exploration of biological processes at
fronts. ICES J. Mar. Sci., 54, 161–7.
Frederiksen, M., Wanless, S., Harris, M. P., Rothery, P. & Wilson, L. J. (2004). The
role of industrial fishery and oceanographic change in the decline of North Sea
black-legged kittiwakes. J. Appl. Ecol., 41, 1129–39.
Furness, R. W. (2002). Management implications of interactions between fisheries
and sandeel-dependent seabirds and seals in the North Sea. ICES J. Mar. Sci.,
59, 261–9.
Furness, R. W. & Tasker, M. L. (2000). Seabird–fishery interactions: quantifying
the sensitivity of seabirds to reductions in sandeel abundance, and
identification of key areas for sensitive seabirds in the North Sea. Mar. Ecol.
Prog. Ser., 202, 354–64.
Gjerdrum, C., Vallee, A. M. J., St Clair, C. C. et al. (2003). Tufted puffin
reproduction reveals ocean climate variability. Proc. Natl Acad. Sci. U. S. A.,
100, 9377–82.
Haren, H. Van, Mills, D. K & Wetsteyn, L. P. M. J. (1998). Detailed observations of
the phytoplankton spring bloom in the stratifying central North Sea. J. Mar.
Res., 56, 655–80.
Harris, M. P. & Wanless, S. (1985). Fish fed to young guillemots, Uria aalge, and
used in display on the Isle of May, Scotland. J. Zool. Lond., 207, 441–58.
Hjort, J. (1914). Fluctuations in the great fisheries of northern Europe viewed in the
light of biological research. Rapp. P.-V. Reun., Cons. Int. Explor. Mer 20, 1–228.
ICES (International Council for the Exploration of the Sea) (2004). Report of the
Working Group on Seabird Ecology, ICES CM 2004/C:05. Copenhagen,
Denmark: ICES.
Jensen, H., Wright, P. J. & Munk, P. (2003). Vertical distribution of pre-settled
sandeel (Ammodytes marinus) in the North Sea in relation to size and
environmental variables. ICES J. Mar. Sci., 60, 1342–51.
Le Fèvre, J. (1986). Aspects of the biology of frontal systems. Adv. Mar. Biol., 23,
164–299.
 Biologically meaningful oceanographic indices 61

Lewis, S., Wanless, S., Wright, P. J. et al. (2001). Diet and breeding performance of
black-legged kittiwakes Rissa tridactyla at a North Sea colony. Mar. Ecol. Prog.
Ser., 221, 277–84.
Mann, K. H. & Lazier, J. R. N. (1996). Dynamics of Marine Ecosystems. Oxford, UK:
Blackwell Science.
Miller, G. B. (2004). Biological Oceanography. Oxford UK: Blackwell Science.
Monaghan, P. (1992). Seabirds and sandeels: the conflict between exploitation and
conservation in the northern North Sea. Biodiversity Conserv., 1, 98–111.
Monteleone, D. M. & Peterson, W. T. (1986). Feeding ecology of American sand
lance Ammodytes americanus larvae from Long Island Sound. Mar. Ecol. Prog.
Ser., 30, 133–43.
Munk, P., Wright, P. J. & Pihl, N. J. (2002). Distribution of the early larval stages of
cod, plaice and lesser sandeel across haline fronts in the North Sea. Estuar.
Coast. Shelf Sci., 55, 139–49.
Otto, L., Zimmerman, J. T. F., Furnes, G. K. et al. (1990). Review of the physical
oceanography of the North Sea. Neth. J. Sea Res., 26, 161–238.
Pearson, W. H., Woodruff, D. L. & Sugarmann, P. C. (1984). The burrowing
behaviour of sand lance, Ammodytes hexapterus: effects of oil-contaminated
sediment. Mar. Environ. Res., 11, 17–32.
Pingree, R. D., Pugh, P. R., Holligan, P. M. & Forster, G. R. (1975). Summer
phytoplankton blooms and red tides along tidal fronts in the approaches to the
English Channel. Nature, 258, 672–7.
Pingree, R. D., Bowman, M. J. & Esaias, W. E. (1978). Headland fronts. In Oceanic
Fronts in Coastal Processes, eds. M. J. Bowman & W. E. Esaias. Berlin:
Springer-Verlag, pp. 78–86.
Platt, T., Fuentes-Yaco, C. & Frank, K. T. (2003). Spring algal bloom and larval fish
survival. Nature, 423, 398–9.
Proctor, R., Wright, P. J. & Everitt, A. (1998). Modelling the transport of larval
sandeels on the north-west European shelf. Fish. Oceanogr., 7, 347–54.
Ratcliffe, N. (2004). Causes of seabird population change. In Seabird Populations of
Britain and Ireland, eds. P. I. Mitchell, S. F. Newton, N. Ratcliffe & T. E. Dunn.
London: T. & A. D. Poyser, pp. 407–37.
Reay, P. J. (1970). Synopsis of Biological Data on North Atlantic Sandeels of the genus
Ammodytes (A. tobianus, A. dubius, A. americanus and A. marinus). FAO
Fisheries Synopsis 82. Rome, Italy: FAO.
Rindorf, A., Wanless, S. & Harris, M. P. (2000). Effects of sandeel availability on
the reproductive output of seabirds. Mar. Ecol. Prog. Ser., 202, 241–52.
Sharples, J. (1999). Investigating the seasonal vertical structure of phytoplankton
in shelf seas. Mar. Models, 1, 3–38.
Sharples, J., Ross, O. N., Scott, B. E., Greenstreet, S. P. R. & Fraser, H. Inter-annual
variability in the timing of stratification and the spring bloom in a temperate
shelf sea. Cont. Shelf Res, in press.
Simpson, J. & Bowers, D. (1981). Models of stratification and frontal movement in
shelf seas. Deep-Sea Res. Part I, 28, 727–38.
Simpson, J. H. (1981). The shelf-sea fronts: implications of their existence and
behaviour. Phil. Trans. R. Soc. Lond. A, 302, 531–46.
62 B. E. Scott et al.

Turrell, W. R. (1992). New hypotheses concerning the circulation of the northern

North Sea and its relation to North Sea fish stock recruitment. ICES J. Mar.
Sci., 49, 107–23.
Waniek, J. J. (2003). The role of physical forcing in the initiation of spring blooms
in the northeast Atlantic. J. Mar. Syst., 39, 57–82.
Winslade, P. (1974a). Behavioural studies on the lesser sand eel Ammodytes
marinus (Raitt) I. The effect of food availability on the activity and the role of
olfaction in food detection. J. Fish Biol., 6, 565–76.
(1974b). Behavioural studies on the lesser sand eel Ammodytes marinus (Raitt) II.
The effect of light intensity on activity. J. Fish Biol., 6, 577–86.
(1974c). Behavioural studies on the lesser sand eel Ammodytes marinus (Raitt) III.
The effect of temperature on activity and environmental control of the annual
cycle of activity. J. Fish Biol., 6, 587–99.
Worsøe, L. A. (1999). Emergence and growth of juvenile sandeel (Ammodytes
marinus): Comparison of two areas in the North Sea. M.Sc thesis, Marine
Biological Laboratory, Helsingør, University of Copenhagen, Denmark.
 5
Linking predator foraging behaviour and diet
with variability in continental shelf ecosystems:
grey seals of eastern Canada
W . D . B O W E N , C . A . B E C K , S . J . I V E R S O N , D . AU S T I N
AND J. I. MCMILLAN

Upper-trophic-level marine predators are presumed to respond to environ-

mental variability. However, the nature of these responses has been stud-
ied in few pinnipeds, particularly during the non-breeding season. Between
1992 and 2003, we measured a suite of behavioural, dietary and life-history
variables in grey seals; variables which were expected to vary in response to
changes in prey availability. We found significant inter-annual variation in
some diving variables indicative of foraging effort and in the species com-
position of their diets. Postpartum body mass of adult females did not vary
inter-annually, but duration of offspring investment (lactation length), total
energy investment (offspring weaning mass) and the difference in weaning
mass of male and female pups did. There was considerable inter-annual
variation in the estimated biomass of grey seal prey species from summer
bottom-trawl surveys; however, there was little correlation between grey seal
response variables with those estimates. There could be several reasons for
this result, but three stand out. First, grey seal numbers on the Scotian Shelf
have increased exponentially over the past four decades, implying overall
favourable environmental conditions. Grey seals may have adjusted their
behaviour and diet to account for variability in prey characteristics other
than biomass. Secondly, foraging grey seals and their prey were not sam-
pled at the same time of year. Finally, trends in the biomass of many of the
species eaten by grey seals are poorly estimated, thus limiting our under-
standing of predator responses to ecosystem state.

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
64 W. D. Bowen et al.

Ecosystems are complex systems in which interactions among species

occur at multiple spatial and temporal scales (Allen 1985, Levin 1992). Sus-
tainable use of marine ecosystems will depend on a better understanding
of the mechanisms underlying responses of ecosystems to natural forcing
and human impacts. To do this will require a variety of approaches, includ-
ing process studies and long-term measurements at multiple spatial and
temporal scales. For many decades, measurements of lower trophic lev-
els (e.g. zooplankton and fishes) have been used to determine how marine
ecosystems change over time (e.g. Mahon et al. 1998, Sherman et al. 1998).
Although upper-trophic-level predators have been monitored as indicators
of ecosystem changes in the Southern Ocean for a number of decades
(Reid & Croxall 2001, Reid et al. (Chapter 17 in this volume), Trathan
et al. (Chapter 3 in this volume)), and it has been suggested that they may
also be useful in other ecosystems (e.g. Montevecchi et al. (Chapter 8),
Furness (Chapter 14) and Tasker (Chapter 24) in this volume), the value of
top marine predators as indicators of ecosystem state has not been widely
investigated.
The grey seal (Halichoerus grypus) is a size-dimorphic member of
the family Phocidae, with males being about 50% heavier than females
(McLaren 1993). This species has several attributes that make them poten-
tially useful indicators of ecosystem state. They are large (>100 kg) and
long-lived (∼40 years) and, thus, individuals must have evolved to cope with
variability at various temporal (months to decades) and spatial (<1 km to
1000 km) scales. Most females give birth each year to a single pup, begin-
ning at age 4 or 5 years and continuing for several decades. Female grey
seals are capital breeders and females with low body mass at parturition
tend to wean smaller pups or wean pups prematurely (Iverson et al. 1993,
Mellish et al. 1999, Pomeroy et al. 1999), increasing the probability of
juvenile mortality (Coulson 1960, Hall et al. 2001). These strong mater-
nal effects on offspring provide a basis for expecting variation in life-history
traits in response to environmental variability.
Grey seals are the most abundant pinniped inhabiting the Scotian Shelf
and adjacent areas. The number of pups born on Sable Island has increased
exponentially for the past four decades, with a doubling time of ∼6 years
(Bowen et al. 2003). Thus, they represent a significant source of preda-
tion mortality in fish, particularly during the 1990s. Grey seals disperse
widely over the continental shelf of the northwest Atlantic during the non-
breeding season (Stobo et al. 1990) and are capable of foraging in many
habitats throughout this range. They are generalist predators of demer-
sal and pelagic fishes, but typically a small number of prey dominate the
 Predator behaviour and environmental variability 65

Table 5.1. Response variables measured in grey seals

Variables Period

Behavioural
Individual Mean duration, depth, bottom time, % time at depth, 1992–2001
dives % square dives, descent and ascent rates, surface
time between dives, number of dives per day, total
duration diving per day, total bottom time per day
Dive bouts Mean duration, depth, % bout at depth, % square dives 1992–2001
per bout, % V-shaped dives per bout, number of
dives per bout, total time in bout per day
Diet % species composition, diversity, energy density 1994–2002
Life history
Maternal Postpartum body mass, lactation length 1992–2001
Offspring Weaning body mass, change in male and female 1992–2001
weaning mass

diet at any one time or place (Bowen et al. 1993, Bowen & Harrison 1994),
probably reflecting local prey abundance. Given their broad geographic dis-
tribution and their accessibility at the main breeding site on Sable Island,
grey seals may provide an opportunity to monitor changes in the Scotian
Shelf ecosystem. Here we address two questions: (1) do behavioural, dietary
and life-history variables of grey seals vary over time; (2) do these responses
correlate with particular features of environmental variability?

DATA COLLECTION

The data were collected from 1992 to 2003 on Sable Island (43◦ 55 N,
60◦ 00 W), a crescent-shaped, partially vegetated sandbar approximately
300 km southeast of Halifax, Nova Scotia, Canada. Sable Island is the largest
haul-out and breeding colony for grey seals in the northwest Atlantic popu-
lation. Seals congregate on the island in May and June to moult and again
in late December and January to rear offspring and mate. Thousands of
grey seals also haul out on the island throughout the year between foraging
trips. We studied a suite of behavioural, dietary and life-history variables to
investigate responses of grey seals to environmental variability (Table 5.1).

DISTRIBUTION

We determined foraging locations of 70 grey seals using satellite-

relay data loggers (SRDLs; Wildlife Computers, Redmond, WA, USA or
ST-18s; Telonics, Mesa, AZ, USA) fitted to seals in either May/June or
66 W. D. Bowen et al.

Fig. 5.1 Annual distribution of grey seals based on locations of 70 adults fitted
with Argos satellite tags on Sable Island (n = 18 for May/June, n = 38 for
September/October, n = 14 for January). To determine the spatial distribution
of the population (i.e. percentage usage), we divided the study area into 5 × 5
cells and counted the number of seals that entered each cell. Multiple use of a
cell by an individual seal was scored as a single use to avoid biasing the
population distribution by the behaviour of individual seals. Too few seals were
tagged in each year to permit the analysis of inter-annual changes in the use of
space. The 100-m isobath is indicated by a grey line.

September/October from 1995 to 2001 and in January 2003 (described in

Austin et al. (2004)). Locations were determined from data collected by Ser-
vice Argos. Locations for each seal (including auxiliary locations) were fil-
tered using a three-stage algorithm (Austin et al. 2003) to remove erroneous
data.
Most grey seal locations were confined to the continental shelves off
eastern Canada and the United States, although transit within and among
shelves occasionally occurred over deeper waters (Fig. 5.1). Within this
range, the areas <100 m depth were used particularly often. Some off-
shore banks are clearly delimited by the distribution of locations, with Sable
Island, and Western and Middle banks (areas near Sable Island) being used
by most seals (Fig. 5.1). This representation presumably provides a reason-
able illustration of the spatial scale of the population and foraging areas that
 Predator behaviour and environmental variability 67

Fig. 5.2 Percentage of square-shaped dives ( ± 1 SE) used by grey seals between
September and December of each year. Number of seals studied is given above
the symbol. Percentage of square-shaped dives differed among years (MANOVA:
F8,81 = 2.4, p = 0.024).

probably underlie variation in grey seal response variables. Nevertheless,

habitat use by individual seals is highly variable (Austin et al. 2004).

RESPONSE VARIABLES

Diving behaviour

We measured characteristics of 8089 ± 403 dives per individual in 93

(46 males and 47 females) adult grey seals over an average of 51 ± 2.5
days during the period from September to December. Using the analyti-
cal methods of Beck et al. (2003a, 2003b), we found that diving behaviour
showed significant inter-annual variability at the scale of individual dives
(MANOVA: year – F63, 511 = 1.38, p = 0.035). Although depth, duration,
and descent and ascent rates of individual dives did not vary inter-annually,
surface time between dives did (1994 excluded because of small sample
size; F8,81 = 2.5, p = 0.016), with surface intervals being significantly
shorter in 1993 compared with other years. The percentage of square-
shaped dives also differed significantly among years (Fig. 5.2). Inter-
annual variation was also evident in bout characteristics (MANOVA: year –
F45, 340 = 1.72, p = 0.004), with an increasing linear trend in the percent-
age of bouts spent at depth (F7,83 = 2.2, p = 0.04).
68 W. D. Bowen et al.

(a)
200

180
Dives per day

160

140

120

100
1990 1992 1994 1996 1998 2000 2002

(b)
20

18
Time in foraging bouts (hours per day)

16

14

12

10

4
1990 1992 1994 1996 1998 2000 2002

Year

Fig. 5.3 Inter-annual indices of foraging effort ( ± 1 SE) (a) Mean number of
dives per day. (b) Time spent in foraging bouts. Sample sizes are as given in Fig.
5.2. Based on MANOVA, both indices varied inter-annually: dives per day,
F 8,81 = 2.9 and p = 0.007; hours in foraging bouts, F7,74 = 2.7 and p = 0.014.

Two of the three indices (number of dives per day, time spent in for-
aging bouts and time spent at depth) of foraging effort varied significantly
among years. Dives per day increased through the mid 1990s and declined
through the late 1990s into 2001 (Fig. 5.3a). Hours spent in foraging bouts
was relatively stable from 1992 to 1997, then declined significantly in
 Predator behaviour and environmental variability 69

1998 only to increase to former levels in 1999, and subsequently decline

(Fig. 5.3b). Number of dives per day was positively correlated with the pro-
portion of square dives exhibited each year (r = 0.85, p = 0.002).

Diet

We used quantitative fatty acid signature analysis (QFASA; Iverson et al.,

2004 (Chapter 7 in this volume)) to derive estimates of diet using a prey
fatty acid library of 27 species of fish and invertebrates collected from the
study area. Blubber biopsies were collected from the posterior flank of seals
shortly after they arrived on the breeding colony in January 1994 through
2002. Individual adult grey seals consumed between 1 and 10 prey species
in the 4 months prior to arriving at the breeding colony, averaging 4.3 ±
0.11 prey species per individual. Diet diversity varied significantly among
years (Fig. 5.4a). However, despite this variation, energy density (kJ g−1 ) of
the diet did not vary among years (ANOVA: F8, 322 = 0.84, p = 0.57; aver-
age 5.5 ± 0.02 kJ g−1 ; range 4.6–6.6 kJ g−1 ).
Species composition of the diet also varied significantly among years
(Fig. 5.4b). In each year, two to five species accounted for over 80% of the
diet by weight, with northern sand lance (Ammodytes dubius) and redfish
(Sebastes spp.) dominating the diet. The proportions of nine different prey
species – including cod (Gadus morhua), gaspereau (Alosa pseudoharengus),
squid (Illex illecebrosus) and thorny skate (Raja radiata) – differed signifi-
cantly among years. However, it was the proportions of pollock (Pollachius
virens), redfish, sand lance, witch flounder (Glyptocephalus cynoglossus), and
winter skate (Raja ocellata) that exhibited the greatest inter-annual variabil-
ity in the diet. Pollock was significantly more abundant in the diet in 1994,
1998 and 2000–2001 than in other years. Sand lance accounted for sig-
nificantly less of the diet in 1994 and 1998 than in other years, whereas
redfish made up a particularly small proportion of the diet in 2000. Sev-
eral other demersal fishes also contributed more to the diet in 1994, 1998
(witch flounder) and 2000 (winter skate Raja radiata) compared with other
years.
We hypothesized that grey seal diving behaviour might vary with the
ratio of the dominant prey types in the diet, as these prey species dif-
fer in many characteristics. We found a positive correlation between the
ratio of redfish to sand lance in the diet and the number of dives per day
(r = 0.71, n = 9, p = 0.03), but no significant correlation with the propor-
tion of square-shaped dives (p = 0.61) or total time spent in diving bouts
(p = 0.13).
(a) 51 46 40 53 20 57 44 11 9

0.50

Standardized Shannon–Weaver Index

0.45

0.40

0.35

0.30

0.25

0.20

(b) 1.0
Other demersals
0.9 Squid
Flounders
0.8
Gadoids Small pelagics
0.7

0.6
Proportion

0.5
Sand lance
0.4

0.3

0.2
Redfish
0.1

0.0
1993 1994 1995 1996 1997 1998 1999 2000 2001
Year

Fig. 5.4 (a) Inter-annual estimates of Shannon–Weaver standardized index

(range 0 to 1) of diet diversity ( ± 1 SE). Diet diversity varied among years
(ANOVA: F8,322 = 5.1, p < 0.001): 1994 differed from 1993, 1995 and 1999;
1995 differed from 1998; 1998 differed from 1999. Diet estimates were based on
fatty acids extracted from blubber cores using the methods of Iverson et al.
(2001). Fatty acid composition was analysed according to Iverson et al. (2002)
and Budge et al. (2002) and used to estimate diet composition following Iverson
et al. (2004, Chapter 7 in this volume). The number of seals studied is given
above the symbols. (b) Percentage contribution of prey species to the diet of grey
seals differed significantly among years (p = 0.0001): 1993 differed from 1994
and 1998; 1994 differed from 1995–7 and 1999–2000; 1995 differed from 1998
and 2000; 1996 differed from 1998; 1997 differed from 1998 and 2000; 1998
differed from 1999 and 2000. Inter-annual variability in diet was tested using
permutation tests. The Kulback–Liebler distances among the mean diet
composition for each year were calculated and then the diet estimates of
individuals were randomly assigned 10 000 times each year. Then the mean diet
composition for each year and the distance between the new yearly means again
was computed creating a distribution against which to test the observed distance.
A Bonferroni correction was applied to the resulting p values to account for the
multiple comparisons. Sample sizes are as in Fig. 5.4a.
 Predator behaviour and environmental variability 71

Life history

Variability in food availability might be reflected in the body mass of females

arriving at the breeding colony to give birth. In turn, inter-annual variability
in maternal postpartum mass (MPPM) might affect the duration of energy
investment (lactation length) and the allocation of energy to offspring (pup
weaning mass). Thus, each year we weighed a sample of known-age females
at 3 days postpartum and determined lactation length and weaning date
based on daily surveys throughout the colony.
MPPM did not differ significantly among years (Fig. 5.5a). Lactation
length in grey seals exhibited significant quadratic variation with maternal
age (W. D. Bowen, unpublished observations). Thus, inter-annual estimates
of lactation length were adjusted to account for differences in the mean age
of females sampled among years. When this was done there was a signifi-
cant difference in mean lactation length among years (Fig. 5.5b) with 1993–5
having shorter lactation periods than other years. However, year accounted
for only 2.4% of the variation and therefore is of doubtful biological signif-
icance. As with lactation length, weaning mass also varied in a quadratic
manner with maternal age. Mean weaning mass (corrected for maternal
age) varied significantly among years (Fig. 5.5c). Weaning mass in 1998
was significantly greater than in 2000, but there were no other significant
differences among years.
Male pups are heavier than female pups, and therefore require a greater
absolute energy investment by mothers. Thus, variation in the average dif-
ference in the weaning mass of male and female pups may indicate years in
which adult females were in relatively better condition, having stored more
energy prior to arrival at the breeding colony. We calculated the difference
between male and female weaning masses each year, corrected for the effect
of maternal age on weaning mass. Differences between the mean weaning
mass of male and female pups were larger than average in 1993, 1998 and
2002 (Fig. 5.6). However, the interpretation of these findings is not clear
as other life-history measures did not indicate that females were in better
condition or made greater overall energy investment in their offspring in
those years.

ENVIRONMENTAL VARIABILITY

Environmental variability on the Scotian Shelf, the marine ecosystem pri-

marily used by grey seals from Sable Island, is summarized in Zwanenburg
et al. (2002). Inter-annual changes in water temperatures and salinities on
72 W. D. Bowen et al.
 Predator behaviour and environmental variability 73

Fig. 5.6 Inter-annual estimates of the difference in mean weaning mass of male
and female grey seal pups.

the Scotian Shelf are among the most variable in the North Atlantic. On
the northeastern Scotian Shelf the cold, intermediate-layer water, repre-
sented by Misaine Bank at 100 m depth, fell sharply by the mid 1980s
and remained below normal through 1995, returning to the climatological
mean in the period 1997 to 2002 (DFO 2003). Despite these changes, the
estimated mean composition of the winter grey seal diet was similar during
the cold period and during years when temperature had returned to average
conditions (MANOVA: F8,322 = 1.8, p = 0.09; Table 5.2).
There were large inter-annual changes in the estimated biomass of some
prey species during the 1990s (Fig. 5.7). However, to a considerable degree
the magnitude of those changes are more difficult to assess as the bottom-
trawl inconsistently samples species such as redfish, sand lance, capelin and

←
Fig. 5.5 Inter-annual estimates of: (a) maternal postpartum mass (MPPM),
(b) lactation length and (c) pup weaning mass of grey seals. Estimated means of
MPPM did not differ among years (univariate general linear model (GLM) with
maternal age as a covariate, F9, 267 = 1.4, p = 0.21); estimated mean lactation
length among years (adjusted for differences in mean maternal age) differed
inter-annually (univariate GLM with maternal age and maternal age 5 squared as
covariates, F10, 811 = 2.0, p = 0.03); estimated mean weaning mass varied
among years (univariate GLM with maternal age and maternal age 5 squared as
covariates, F10, 1328 = 2.4, p = 0.01). Error bars are 95% confidence limits.
74 W. D. Bowen et al.

Table 5.2. Percentage composition of grey seal diets during a cold-water

period (1993–6) and after a return to climatological average temperatures
(1997–2001) based on average annual water temperature at 100 m,
Misaine Bank

Period

Species Cold Average

Redfish 33.8 29.7

Sand lance 38.8 37.8
Other forage fisha 6.6 8.4
Cod 1.8 1.0
Other demersalsb 6.8 8.4
Floundersc 5.0 5.3
Skatesd 4.3 6.6
Squid 2.9 2.7

a
Capelin (Mallotus villosus), herring (Clupea harengus), mackerel (Scomber scombrus),
snakeblenny (Lumpenus lumpretaeformis), gaspereau (Alosa pseudoharengus).
b
Pollock, haddock (Melanogrammus aeglefinus), lumpfish (Cyclopterus lumpus).
c
American plaice (Hippoglossoides platessoides), yellowtail ( Limanda ferruginea), witch
flounder, winter flounder, turbot (Rheinhardtius hippoglossoides).
d
Thorny skate (Raja radiata), winter skate (Raja ocellata).

pollock. This inconsistent sampling is suggested by the dramatic apparent

increase in capelin biomass in 1994 and pollock biomass in 1996 (Fig. 5.7).
These are some of the more frequently consumed grey seal prey, thus mak-
ing it difficult to determine how grey seal consumption may respond to prey
abundance. Nevertheless, the abundance of sand lance clearly increased
over the course of our study, whereas capelin decreased and redfish biomass
seems to have been relatively stable (Fig. 5.7). Flounders also appeared to
have increased during the later part of our study, whereas pollock biomass
fluctuated, but in general declined. Despite inter-annual variation in both
estimated diet and prey abundance, the proportions of the dominant prey
(i.e. redfish, sand lance, pollock) in the diet were not significantly correlated
with prey abundance.

SEALS AS INDICATORS OF ECOSYSTEM STATE

The idea that upper-trophic-level predators can provide information that

could be used to improve the management of marine species is attractive
because such predators sample their environment at a range of spatial and
temporal scales that are difficult and expensive to achieve using research
 500

Biomass (× 1000 t) 400

300

200

100 Skates

Flounders

0 Pollock
1990 1992 1994 1996 1998 2000 2002

8000

6000
Biomass (× 1000 t)

4000

2000 Redfish

Capelin

0 Sand lance
1990 1992 1994 1996 1998 2000 2002

Year

Fig. 5.7 Biomass estimates of selected grey seal prey from July bottom-trawl
surveys, 1990–2002. Inter-annual variability in fish abundance was derived from
synoptic, stratified-random, bottom-trawl surveys conducted each July. Estimates
of species biomass, corrected for catchability to provide a better indication of true
relative abundance, were combined for the trawl survey strata primarily used by
grey seals. However, species such as sand lance, redfish, capelin and pollock are
poorly sampled by bottom trawls such that the resulting biomass estimates are
biased and observed trends may provide only a rough indication of true trends.
76 W. D. Bowen et al.

and commercial vessels. Nevertheless, the use of predators for this purpose
requires an understanding of how predator responses are linked to the vari-
ability in particular ecosystem components (Croxall et al. 1988, Boyd & Mur-
ray 2001, Hindell et al. 2003). We found significant inter-annual variation in
aspects of foraging behaviour, diet and several life-history variables of grey
seals over the 10 years of our study. Presumably, differences in foraging
behaviour and diet are causal, reflecting the need to use different foraging
tactics to locate and capture different prey species (e.g. redfish versus sand
lance) (Bowen et al. 2002). Similarly, differences in behaviour and diet are
presumably related to changes in the availability of prey. However, the links
among these variables are not clear in our data.
The continental-shelf ecosystems inhabited by grey seals in east-
ern Canada have exhibited considerable variability over the past several
decades – involving changes in physical and biological oceanography, fish-
eries exploitation rates and species abundance – with a general shift from
a system dominated by demersal fishes to one dominated by pelagic fish
species (Rice 2000, Swain & Sinclair 2000, Zwanenburg et al. 2002). Thus,
there were considerable changes in ecosystem state to test whether grey
seals revealed those changes. However, only grey seal pup production at
Sable Island was monitored over those earlier several decades. Measure-
ments of the behavioural, diet and life-history variables were only initiated
in the early 1990s after many of the larger changes had already occurred.
Grey seals are large, long-lived mammals with K-selected life histo-
ries. Despite the large environmental changes observed over the past four
decades on the Scotian Shelf (Zwanenburg et al. 2002), the grey seal pop-
ulation size has increased steadily from only a few thousand seals in the
1960s to about 175 000 in 1995 (Mohn & Bowen 1996). Pup production
on Sable Island increased exponentially, at a rate near the maximum possi-
ble (rmax ), through the late 1990s (Bowen et al. 2003). Although there is no
a-priori reason to have expected exponential population growth, the fact that
it occurred suggests that – from a grey seal perspective – the environment
was favourable throughout this period. This demographic history provides
an essential context for interpreting the performance of the response vari-
ables measured in this study.
Diving behaviour ought to reflect characteristics of the prey available to
pinnipeds since all foraging necessarily occurs during diving. The relation-
ship between diving behaviour and changes in prey availability is perhaps
best understood in Antarctic fur seals (Arctocephalus gazella) (e.g. Bengtson
1988, Boyd et al. 1994, McCafferty et al. 1998). Females in this species
altered both trip duration and number of dives in response to changes in
 Predator behaviour and environmental variability 77

krill abundance and the amount of fish and squid in the diet. However,
these conclusions are limited to the period of offspring provisioning and
thus may not be representative of responses at other times of the year, or
in males. We studied diving behaviour of adult male and female grey seals
over the 4 months prior to arrival at the breeding colony. During this period
both sexes gain mass (Beck et al. 2003c), indicating that this is a period of
heavy feeding. Although most variables describing individual dives or bouts
of dives exhibited little inter-annual variability, number of dives per day, pro-
portion of square-shaped dives, proportion of dive bout spent at depth and
total time spent in diving bouts per day varied among years. However, for
the most part, inter-annual variation in foraging behaviour was not related
to differences in diet or estimated prey biomass. Number of dives per day
was positively correlated with the ratio of the two dominant prey in the diet,
redfish and sand lance. However, this finding is difficult to interpret with-
out knowing how predator foraging tactics differ for these prey types.
Inter-annual variation in pinniped diets is generally assumed to reflect
changes in prey abundance and encounter rates (Bowen & Siniff 1999).
Although demersal species accounted for ∼25% of the grey seal diet in
some years, diets were dominated by sand lance and redfish. The percent-
age of those two species in the diet varied significantly among years. How-
ever, there was no correlation between this variation and estimates of prey
biomass from trawl surveys conducted within grey seal habitat. There are
a number of possible reasons for this. Firstly, the estimate of prey biomass
was derived from the survey conducted in July, whereas our diet sam-
ples were collected about 5 months after the survey. Although fatty-acid-
based estimates of diet should integrate intake over several months (Iverson
et al. 2004), both prey availability and grey seal distribution are presumably
dynamic such that the July survey may not be a good measure of prey avail-
able to seals months later. Secondly, the trawl survey is known to sample
both redfish and sand lance inconsistently. Thus, the true abundance of
these species may not be reflected by the survey. Thirdly, the small number
of grey seals sampled in some years (e.g. 1997, 2000 and 2001) may not
have been representative of grey seal diets. Obtaining a representative sam-
ple may be difficult for a wide-ranging predator exploiting a spatially het-
erogeneous habitat. Fourthly, although we know little about the ontogeny
of foraging behaviour in grey seals and other pinnipeds, it is reasonable to
expect that learning plays an important role in the diet of individual seals
resulting in strong individual differences in diet among individuals for-
aging in the same habitat (Estes et al. 2003). Individual prey preferences
may partially obscure responses at the population level, particularly when
78 W. D. Bowen et al.

overall prey resources are not limited. Finally, and perhaps most impor-
tantly, given the favourable prey environment (as judged by the rapid rate of
population increase), it is possible that grey seals were foraging in the range
of the asymptotic limb of the non-linear functional response curve (Furness
(Chapter 14 in this volume)) where consumption is relatively insensitive to
changes in prey biomass. If true, the interpretation of predator responses
will be contingent on demography.
The significance of changes in foraging behaviour and diet to the preda-
tor can only be determined through their effects on demography (Croxall
et al. 1988). However, annual estimates of survival and fecundity are diffi-
cult to measure in most pinnipeds. Maternal and offspring size and con-
dition are attractive because they can be easily measured, ought to reflect
changes in prey availability and can affect demography. We found that
MPPM, an index of foraging success, did not vary among years. Duration
of maternal investment (i.e. lactation length) and pup weaning mass exhib-
ited significant, but relatively little, inter-annual variation. Interestingly, in
the year (1998) that the difference between male and female weaning mass
was greatest, combined pup weaning mass was also the highest, perhaps
suggesting that adult females were in particularly good condition that year.
However, in general, these life-history response variables in grey seals were
not informative of ecosystem state. As noted above, our data were collected
during a period when this population was experiencing favourable envi-
ronmental conditions and exponential population growth. The same vari-
ables measured during a period of population decline or stability may have
responded quite differently.
For the present, we conclude that despite the large changes in estimates
of invertebrate- and fish-species abundances (Zwanenburg et al. 2002),
MPPM, lactation length and offspring weaning mass provided little indica-
tion of those environmental changes. Although we observed greater inter-
annual variability in the foraging behaviour and diet, those response vari-
ables for the most part were also not informative with respect to specific
ecosystem changes that occurred during the 1990s. However, we believe
it would be premature to suggest that grey seals and similar species will
not be useful monitors on the basis of this initial exploratory analysis. It
is possible that grey seal diets are better indicators of abundance for many
of the species consumed than are the bottom-trawl surveys routinely used
for this purpose. Comparison of species estimates in the diet of grey seal
against reconstructed prey-population abundance from catch-at-age models
might provide a means of validating this hypothesis. Measurement of for-
aging behaviour and diet response variables at other times of the year may
 Predator behaviour and environmental variability 79

be more informative because they coincide more closely with measures of

prey distribution and abundance from surveys. Combining information on
the spatial distribution of foraging with diving variables and diet may be a
more sensitive indicator of response by grey seals to prey abundance (Boyd
et al.2002). Finally, the exponential growth observed in this population over
the past four decades cannot persist indefinitely. As indicated by fur seal
responses to krill abundance (Boyd et al. 1994) and the wealth of data on
seabirds (e.g. Montevecchi 1993), we expect that many of the response vari-
ables examined here will provide more information about ecosystem state
when the grey seal population is eventually limited by its food supply.

ACKNOWLEDGEMENTS

We thank B. Beck, D. Boness, S. Budge, D. Coltman, M. Cooper, S. Insley,

S. Lang, D. Lidgard, S. McCulloch, M. Muelbert, D. Parker, L. Rea, T. Schulz,
W. Stobo, G. Thiemann, S. Tucker and D. Tully for assistance with the field
work. We also thank G. Forbes for providing logistic support on Sable Island
and the Canadian Coast Guard for ship and helicopter support. Funding
was provided by the Canadian Department of Fisheries and Oceans, and
the Natural Science and Engineering Research Council of Canada. We also
thank an anonymous reviewer for helpful comments on an earlier version
of this chapter.
REFERENCES

Allen, P. M. (1985). Ecology, thermodynamics, and self-organization: towards a

new understanding of complexity. Can. Bull. Fish. Aquat. Sci., 213, 3–26.
Austin, D., McMillan, J. I. & Bowen, W. D. (2003). A three-stage algorithm for
filtering erroneous Argos satellite locations. Mar. Mamm. Sci., 19, 371–83.
Austin, D., Bowen, W. D. & McMillan, J. I. (2004). Intraspecific variation in
movement patterns: modeling individual behaviour in a large marine predator.
Oikos, 105, 15–30.
Beck, C. A., Bowen, W. D., McMillan, J. I. & Iverson, S. J. (2003a). Sex differences
in the foraging behaviour of a size dimorphic capital breeder: the grey seal.
Anim. Behav., 66, 777–89.
(2003b). Sex differences in diving at multiple temporal scales in a size-
dimorphic capital breeder. J. Anim. Ecol., 72, 979–93.
Beck, C. A., Bowen, W. D. & Iverson, S. J. (2003c). Seasonal energy storage and
expenditure in a phocid seal: evidence of sex-specific trade-offs. J. Anim. Ecol.,
72, 280–91.
Bengtson, J. L. (1988). Long-term trends in the foraging patterns of female
Antarctic fur seals at South Georgia. In Antarctic Ocean Resource Variability, ed.
D. Sahrhage. Berlin: Springer-Verlag, pp. 286–91.
Bowen, W. D. & Harrison, G. D. (1994). Offshore diet of grey seals Halichoerus
grypus near Sable Island, Canada. Mar. Ecol. Prog. Ser., 112, 1–11.
80 W. D. Bowen et al.

Bowen, W. D. & Siniff, D. B. (1999). Distribution, population biology, and feeding

ecology of marine mammals. In Biology of Marine Mammals, eds. J. E. Reynolds,
III & S. A. Rommel. Washington, DC: Smithsonian Press, pp. 423–84.
Bowen, W. D., Lawson, J. W. & Beck, B. (1993). Seasonal and geographic variation
in the species composition and size of prey consumed by grey seals
(Halichoerus grypus) on the Scotian shelf. Can. J. Fish. Aquat. Sci., 50, 1768–78.
Bowen, W. D., Tully, D., Boness, D. J., Bulheier, B. & Marshall, G. (2002).
Prey-dependent foraging tactics and prey profitability in a marine mammal.
Mar. Ecol. Prog. Ser., 244, 235–45.
Bowen, W. D., McMillan, J. I. & Mohn, R. (2003). Sustained exponential
population growth of the grey seal on Sable Island. ICES J. Mar. Sci., 60, 1265–
374.
Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Boyd, I. L., Arnould, J. P. Y., Barton, T. & Croxall, J. P. (1994). Foraging behaviour
of Antarctic fur seals during periods of contrasting prey abundance. J. Anim.
Ecol., 63, 703–13.
Boyd, I. L., Staniland, I. J. & Martin, A. R. (2002). Distribution of foraging by
female Antarctic fur seals. Mar. Ecol. Prog. Ser., 242, 285–94.
Budge, S. M., Iverson, S. J., Bowen, W. D. & Ackman, R. G. (2002). Among- and
within-species variability in fatty acid signatures of marine fish and
invertebrates on the Scotian Shelf, Georges Bank, and southern Gulf of St.
Lawrence. Can. J. Fish. Aquat. Sci., 59, 886–98.
Coulson, J. C. (1960). The growth of grey seal calves on the Farne Islands,
Northumberland. Trans. Nat. Hist. Soc., Northumberland, 13, 86–100.
Croxall, J. P., McCann, T. S., Prince, P. A. & Rothery, P. (1988). Reproductive
performance of seabirds and seals at South Georgia and Signy Island, South
Orkneys Islands, 1976–1987: implications for Southern Ocean monitoring
studies. In Antarctic Ocean Resource Variability, ed. D. Sahrhage. Berlin:
Springer-Verlag, pp. 261–85.
DFO (Department of Fisheries and Oceans) (2003). State of the Eastern Scotian Shelf
Ecosystem. Ecosystem Status Report. Dartmouth, Nova Scotia: Department of
Fisheries and Oceans.
Estes, J. A., Riedman, M. L., Staedler, M. M., Tinker, M. T. & Lyon, B. E. (2003).
Individual variation in prey selection by sea otters: patterns, causes and
implications. J. Anim. Ecol., 72, 144–55.
Hall, A., McConnell, B. & Barker, R. (2001). Factors affecting first-year survival in
grey seals and their implications for life history. J. Anim. Ecol., 70, 138–49.
Hindell, M. A., Bradshaw, C. J. A., Harcourt, R. G. & Guinet, C. (2003). Ecosystem
monitoring: are seals a potential tool for monitoring change in marine
systems? In Marine Mammals: Fisheries, Tourism and Management Issues, eds.
N. Gales, M. Hindell & R. Kirkwood. Collingwood, Australia: CSIRO, pp. 330–
43.
Iverson, S. J., Bowen, W. D., Boness, D. J. & Oftedal, O. T. (1993). The effect of
maternal size and milk energy output on pup growth in grey seals (Halichoerus
grypus). Physiol. Zool., 66, 61–88.
Iverson, S. J., Lang, S. & Cooper, M. (2001). Comparison of the Bligh and Dyer and
Folch methods for total lipid determination in a broad range of marine tissue.
Lipids, 36, 1283–7.
 Predator behaviour and environmental variability 81

Iverson, S. J., Frost, K. J. & Lang, S. (2002). Fat content and fatty acid composition
of forage fish and invertebrates in Prince William Sound, Alaska: factors
contributing to among and within species variability. Mar. Ecol. Prog. Ser., 241,
161–81.
Iverson, S. J., Field, C., Bowen, W. D. & Blanchard, W. (2004). Quantitative fatty
acid signature analysis: a new method of estimating predator diets. Ecol.
Monogr., 74, 211–35.
Levin, S. A. (1992). The problem of pattern and scale in ecology. Ecology, 73,
1943–67.
Mahon, R., Brown, S. K., Zwanenburg, K. C. T. et al. (1998). Assemblages and
biogeography of demersal fishes of the east coast of North America. Can. J.
Fish. Aquat. Sci., 55, 1704–38.
McCafferty, D. J., Boyd, I. L., Walker, T. R. & Taylor, R. I. (1998). Foraging
responses of Antarctic fur seals to changes in the marine environment. Mar.
Ecol. Prog. Ser., 166, 285–99.
McLaren, I. A. (1993). Growth in pinnipeds. Biol. Rev., 68, 1–79.
Mellish, J.-A. E., Iverson, S. J. & Bowen, W. D. (1999). Individual variation in
maternal energy allocation and milk production in grey seals and
consequences for pup growth and weaning characteristics. Physiol. Biochem.
Zool., 72, 677–90.
Mohn, R. & Bowen, W. D. (1996). Grey seal predation on the eastern Scotian Shelf:
modelling the impact on Atlantic cod. Can. J. Fish. Aquat. Sci., 53, 2722–38.
Montevecchi, W. A. (1993). Birds as indicators of changes in marine prey stocks. In
Birds as Monitors of Environmental Change, eds. R. W. Furness and J. J. D.
Greenwood. London: Chapman and Hall, pp. 217–66.
Pomeroy, P. P., Fedak, M. A., Rothery, P. & Anderson, S. (1999). Consequences of
maternal size for reproductive expenditure and pupping success of grey seals at
North Rona, Scotland. J. Anim. Ecol., 68, 235–53.
Reid, K. & Croxall, J. P. (2001). Environmental response of upper trophic-level
predators reveals a system change in an Antarctic marine ecosystem. Proc. R.
Soc. Lond. B, 268, 377–84.
Rice, J. C. (2000). Evaluating fishery impacts using metrics of community
structure. ICES J. Mar. Sci., 57, 682–8.
Sherman, K., Solow, A. Jossi, J. & Kane, J. (1998). Biodiversity and abundance of
the zooplankton of the Northeast Shelf ecosystem. ICES J. Mar. Sci., 55, 730–8.
Stobo, W. T., Beck, B. & Horne, J. K. (1990). Seasonal movements of grey seals
(Halichoerus grypus) in the Northwest Atlantic. In Population Biology of
Sealworm (Pseudoterranova decipiens) in Relation to its Intermediate and Seal
Hosts, ed. W. D. Bowen. Can. Bull. Fish. Aquat. Sci., 222, 199–213.
Swain, D. P. & Sinclair, A. F. (2000). Pelagic fishes and the cod recruitment
dilemma in the Northwest Atlantic. Can. J. Fish. Aquat. Sci., 57, 1321–5.
Zwanenburg, K. C. T., Bowen, D. W., Bundy, A. et al. (2002). Decadal changes in
the Scotian Shelf large marine ecosystem. In Large Marine Ecosystems of the
North Atlantic: Changing States and Sustainability, eds. K. Sherman & H. R.
Skjoldal. Amsterdam, the Netherlands: Elsevier Science.
 6
Distribution and foraging interactions of
seabirds and marine mammals in the North Sea:
multispecies foraging assemblages and
habitat-specific feeding strategies
C. J. CAMPHUYSEN, B. E. SCOTT AND S. WANLESS

The top-predator community in the northwest North Sea consists of 50

species of seabirds and marine mammals, most of which are piscivorous.
Sandeels are important prey for many species, and reduced sandeel abun-
dance has had detectable consequences for breeding success, most notably
in surface-feeding seabirds. In recent years, breeding success and popu-
lation trends of seabirds nesting along the east coast of Britain have dif-
fered among species, suggesting species-specific responses to fluctuating
prey stocks. A large-scale, multi-disciplinary study of top-predator distribu-
tion patterns and at-sea foraging behaviour was conducted in the north-
west North Sea to investigate some of the behavioural mechanisms under-
lying these species-specific population responses. This approach provided
new insights into the ways in which marine predators utilize a shared
prey resource. At-sea distributions of some of the smaller seabirds, such as
black-legged kittiwakes, suggested individuals avoided feeding in inshore
areas used by the larger Larus gulls. This resulted in an apparently counter-
intuitive, positive relationship between annual breeding success and for-
aging range, with productivity tending to be lower in years when oceano-
graphic conditions led to good foraging areas occurring closer inshore.
Combining distributional data with information on activity patterns showed
that northern gannets used different foraging strategies in nearshore
and offshore habitats and that chick-rearing common guillemots utilized

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Distribution and foraging interactions 83

spatially segregated, colony-specific feeding areas. Many surface-feeding

and plunge-diving seabirds relied heavily on facilitation by pursuit-diving
predators, such as auks and cetaceans.
Sandeels Ammodytidae are major prey for top predators in the North
Sea such as seabirds (Furness 1990, Lewis et al. 2001), cetaceans (Santos
et al. 2004) and pinnipeds (Hammond & Fedak 1994). Severe effects of
sandeel stock collapses on some species have been reported (Bailey et al.
1991), but the relationship between prey density and availability to preda-
tors remains poorly understood. Some seabirds fail to reproduce in years
when sandeel stocks are low (Monaghan et al. 1992), while other species
adjust their foraging successfully or change prey (Martin 1989). Rindorf
et al. (2000) investigated the potential impact of the industrial sandeel fish-
ery on seabirds, assuming that breeding success of seabirds depended on
sandeel availability and that the fishery may have reduced sandeel availabil-
ity to a level at which avian reproductive output is affected. It appeared that
breeding success was significantly reduced when sandeel availability to the
fishery in June was low, but also that the timing of peak sandeel availability
influenced reproductive output such that success was lower when availabil-
ity peaked early.
Recent studies of factors influencing the availability of sandeels to four
common seabirds off the British east coast, using a combination of data
loggers on individual birds (Hamer et al. (Chapter 16 in this volume),
Daunt et al. (Chapter 12 in this volume)) and observations at the colony,
have provided detailed insight into the foraging activities. Such studies are
essentially single-species investigations, and to examine the complicated
interplay between predators, a large-scale study of the at-sea distribution,
foraging behaviour, feeding interactions and hydrographical characteristics
of the feeding areas of all avian and mammalian top predators was con-
ducted. In this chapter we present data from systematic surveys of the north-
west North Sea (area surveyed 54◦ to 59◦ N, 2◦ E – British east coast, Fig. 6.1,
Box 6.1) in nine summers between 1991 and 2003. We use these results to
focus on intra- and interspecific interactions between predators in differ-
ent areas, examine their tendency to participate in feeding assemblages and
investigate area usage in terms of multispecies foraging opportunities and
broad-scale habitat characteristics.

TOP-PREDATOR COMMUNITY AT SEA

The seabird breeding population at the mainland coast between Banff

and Humberside (54◦ to 58◦ 30 N) in 2000 was estimated at 680 000
84 C. J. Camphuysen et al.

Fig. 6.1 Study area (54◦ to 59◦ N, 2◦ E to the coast) and locations of seabird
colonies and oceanographic areas mentioned in the text. Isobaths for 30-, 50- and
100-m depths are shown, horizontal lines indicate ship-based transects (see Box
6.1).
 Distribution and foraging interactions 85

Box 6.1 Recording seabirds and marine mammals at sea: general

methods
At-sea densities of seabirds, seals, whales, dolphins and harbour
porpoises were assessed during nine acoustics surveys by the fish-
eries research vessel Tridens in the northwest North Sea in June
and July 1991–2003. Additional censuses were conducted on board
RV Pelagia for a sub-sample of transects in the Wee Bankie area
in June 2003. Census techniques were standardized strip-transect
counts using 5- or 10-min intervals, and using a snap-shot for fly-
ing birds (Tasker et al. 1984) with special emphasis given to record-
ing foraging behaviour and feeding assemblages (Camphuysen &
Garthe 2004). Birds and mammals were detected by eye and identi-
fied by using 10 × 40 binoculars. Surveys were conducted along 20
transects perpendicular to the British east coast (Fig. 6.1), running
from approximately 10 km from the coast out to a latitude of 2◦ E
in the central North Sea. Additional data from CTD casts to sample
water masses, and acoustic information on fish distribution, were
collected during the 1999–2003 surveys.

pairs, comprising 19 species – with common guillemot Uria aalge (30% of

the total population), black-legged kittiwake Rissa tridactyla (25%), Atlantic
puffin Fratercula arctica (22%) and northern gannet Morus bassanus (7%)
being most abundant (Mitchell et al. 2004). The relative abundances of
seabirds recorded at the breeding colonies were mirrored in the numbers
of birds seen within approximately 100 km of the coast. Small numbers
of two species of Puffinus shearwaters, two storm-petrels Hydrobatidae,
one Phalaropus phalarope, four skuas Stercorariidae and three Larus gulls
occurred as non-breeding visitors. With divers Gaviidae, grebes Podicipedi-
dae and seaduck Anatidae included, the overall summer seabird commu-
nity comprised 39 species. Marine mammals present in the area included
harbour seal Phoca vitulina, grey seal Halichoerus grypus and at least nine
cetaceans, with harbour porpoise Phocoena phocoena, white-beaked dolphin
Lagenorhynchus albirostris and minke whale Balaenoptera acutorostrata being
the most abundant and widespread. Thus in total, the avian and mam-
malian top-predator community comprised at least 50 species.
Densities of both seabirds and seals declined with increasing distance
from the coast, with values markedly lower beyond 100 km (Fig. 6.2a).
(a) 40

Density (bars) and n species richness (line)

35 Auks
Tems
Kittiwake
30
Large Larus gulls
Small Larus gulls
25
Skuas
Seaduck
20
Cormorants
Gannet
15
Storm-petrels
Shearwaters
10 Fulmar
Divers
5 Species richness

0
20 40 60 80 100 150 200 250 300

(b) 100%
Auks
90%
Tems
80% Kittiwake
Large Larus gulls
70% Small Larus gulls
Skuas
60%
Seaduck
50% Cormorants
Gannet
40% Storm-petrels
Shearwaters
30%
Fulmar
20% Grebes
Divers
10%

0%
Coastal 20 40 60 80 100 150 200 250 300

(c) 10 000
Seabirds
Mar. mammals

1000
Biomass (kg km −2)

100

10

1
20 40 60 80 100 150 200 250 300
Distance strata off UK coast (km)

Fig. 6.2 Changes in top-predator community by distance from land.

(a) Densities (n km−2 ; stacked bars) and jack-knife estimate of species richness
(n ± 95% Confidence interval; line). (b) Biomass as a percentage of total
breeding population 1999–2003 (Mitchell et al. 2004) from summer censuses of
divers Gaviidae and seaduck Anatidae in nearshore waters (Pollock & Barton
2004); percentage of the breeding population and counts of non-breeding bird
species from the coast derived from ship-based surveys in June–July 1991–2003.
(c) Biomass estimates for all seabirds and marine mammals (Mar. mammals;
pinnipeds and cetaceans combined).
 Distribution and foraging interactions 87

In the case of seabirds, species richness also declined substantially with

distance (Fig. 6.2a). Predator groups for which >60% of individuals were
recorded within 40 km of the coast included divers, grebes, cormorants,
shearwaters, seaduck, skuas, Larus gulls, terns Sternidae and seals. Groups
with a slightly more offshore distribution (>70% of individuals recorded
within 80 km of the coast) included northern gannet, phalaropes, black-
legged kittiwake, auks, whales and harbour porpoise. Predators found fur-
thest away from the coast (50% to 75% of individuals recorded >80 km
from land) included European storm-petrel Hydrobates pelagicus, dolphins
and North Atlantic fulmar Fulmarus glacialis. In biomass terms, the seabird
community within 80 km of the coast was dominated by pursuit-diving
auks, whereas deep-plunging northern gannets and surface-feeding north-
ern fulmars were most important further offshore (Fig. 6.2b). However,
marine mammal biomass greatly exceeded that of seabirds in all areas
(Fig. 6.2c).

FORAGING RANGE

In general, the highest densities of foraging seabirds between the Farne

Islands and Moray Firth/Witch Ground were observed within 100 km of the
coast (Figs 6.1 and 6.2a). The offshore boundary of this feeding zone was
typically quite abrupt, being characterized by high densities of black-legged
kittiwakes, common guillemots and razorbills Alca torda. A comparison of
annual observations along 11 transects running perpendicular to the coast
between the Farn Deeps and the Moray Firth (Fig. 6.1), indicated that the
mean (±SE) of this boundary occurred between 33 ± 12 km (1998) and
60 ± 5 km (1997) of the coast (range 5 to 100 km for individual transects). In
1999, the boundary was difficult to identify, with high densities of foraging
seabirds recorded 35 km from the coast on one of the transects, but with
the transition zone between high and low feeding densities being diffuse
on the other 10 transects.
Concurrent with these surveys, the foraging locations of several seabird
species were recorded using data loggers deployed on breeding adults
(Daunt et al. (Chapter 12 in this volume), Hamer et al. (Chapter 16 in this
volume)). This provided a unique opportunity to compare findings from
ship-based surveys with data from seabirds of known origin and breed-
ing status. In the case of common guillemots, birds carrying fish – pre-
sumably back to the colony either to feed chicks or for display – were
frequently recorded during survey transects. Flight directions of individ-
uals heading towards land suggested that breeders from different colonies
88 C. J. Camphuysen et al.

were using spatially discrete foraging areas. Comparing these results with
information on diving locations, obtained using activity loggers deployed on
chick-rearing adults on the Isle of May in 2003, showed that there was close
agreement between the two methods, with birds feeding predominantly on
the western side of the Wee Bankie. In addition, the at-sea surveys suggested
that common guillemots from the Farne Islands and St Abb’s Head were
using the southern part of the Marr Bank, while birds from Fowlsheugh
foraged mainly in the northern part (Fig. 6.1). These results indicate max-
imum foraging ranges of 50 km for common guillemots from the Isle of
May, 55 km for St Abb’s Head, 70 km for the Farne Islands and at least 110
km for Fowlsheugh.

FORAGING-HABITAT CHARACTERISTICS

The study area is part of the Northeast Atlantic shelves province of the
Atlantic coastal biome (Longhurst 1999) and contains two distinct hydro-
graphic regions: North Atlantic waters, which occupy most of the central
North Sea, and Scottish coastal waters (Otto et al. 1990, Scott et al. (Chapter
4 in this volume)). During the winter months, lower levels of solar radia-
tion combined with stronger winds and tidal friction leave the water col-
umn throughout the North Sea completely mixed. Only in the spring does
the surface layer in deeper areas begin to warm due to increasing amounts
of sunlight and decreasing winds. This warming creates a difference in den-
sity between the upper and lower layers of the water column and the onset
of the resulting stratification allows plankton to stay above the critical depth
needed for population growth and marks the beginning of seasonal primary
production (Scott et al. (Chapter 4 in this volume)). In shelf seas, shallow
sea fronts, also known as tidal mixing fronts, separate inshore areas that
are permanently vertically mixed due to their shallow depth and/or strong
tidal currents, from areas that stratify due to deeper depths and/or weaker
tidal currents (Simpson 1981, Scott et al. this volume). Top predators fre-
quently congregate around these shallow sea fronts that are associated with
increased abundances of fish, larvae and zooplankton (Pingree et al. 1975,
Pingree & Griffiths 1978, Richardson et al. 1986). The exact locations of the
fronts change over the spring and summer months in response to weather
conditions, and the monthly and daily rhythm of tidal speeds. A ‘stratifica-
tion index’, defined as the difference in density between the sea surface and
the bottom, can be used to identify the locations of fronts (Heath & Brander
2001). The offshore boundary of the area used by many seabirds and marine
mammals repeatedly identified from at-sea surveys, typically coincided with
 Distribution and foraging interactions 89

this frontal zone, where the stratification index ranged from 0.6 to 0.8
(cf. Ollason 2000).

FORAGING BEHAVIOUR AND MULTISPECIES FEEDING

ASSOCIATIONS (MSFAs)

Small, short-lived MSFAs (Box 6.2) were frequently recorded in the coastal
foraging zone, particularly around the shallow sea front. The tendency to
participate in such MSFAs differed among the various species (Table 6.1).
Black-legged kittiwakes frequently acted as catalysts or initiators in MSFA
formation, large gulls and skuas quickly joined in, with the former acting
as scroungers or suppressors, while the latter were peripheral, aerial klep-
toparasites (see Box 6.2 for definitions of these terms). Small species such
as storm-petrels and terns rarely joined feeding aggregations, except at the
periphery, possibly because such birds are likely to lose out in direct com-
petition with other predators. Auks were normally joined by other seabirds
and rarely joined existing aggregations (0.3% of cases, n = 3277 MSFAs
recorded within 100 km of the coast). The most common type of MSFA in
coastal waters formed over groups of feeding common guillemots and/or
razorbills (76%, n = 3277), puffins (13%) or harbour porpoises (3%). Within
40 km of the coast, about one-quarter of MSFAs (26%, n = 1518) were tar-
geted by large Larus gulls, and the arrival of these species rapidly prevented
further access by catalysts. In contrast, only 6% (n = 1759) of MSFAs more
than 40 km from land were targeted by large gulls, and black-legged kitti-
wake foraging activities tended to be concentrated in these aggregations.
The apparent avoidance by black-legged kittiwakes of the inshore areas
used by the large gulls resulted in a counter-intuitive, positive relationship
between kittiwake annual breeding success and foraging range (rS = 0.68,
n = 9, p < 0.05) such that success tended to be lower in years when the
shallow sea front occurred closer inshore. Northern gannets joined 18% of
MSFAs (n = 3277), and their arrival typically rapidly disrupted the forag-
ing opportunities of all the other participants, including other gannets and
auks.
Large differences in feeding activity, as well as in the frequency of occur-
rence of MSFAs, were recorded when comparing transects crossing the
shallow sea front. On some occasions only large flocks of inactive (resting or
preening) seabirds were encountered while on others high numbers of birds
and MSFAs were recorded. A dedicated cruise in 2003 revealed that forag-
ing activity in these areas varied during the day in relation to changes in tidal
currents, suggesting that physical processes may help drive prey towards the
Table 6.1. Proportion of surface-feeding and plunge-diving seabirds participating in MSFAsa , behavioural characteristics and
role within MSFAs (see Box 6.2), main feeding areab and the type of diving predator producing the MSFA (see Box 6.2)

MSFA
Species (%) Feeding behaviour and MSFA role Distance Producer

Arctic skua (Stercorarius parasiticus) 93 Kleptoparasite, joining Nearshore —

Great skua (S. skua) 89 Kleptoparasite, joining Nearshore —
Herring gull (Larus argentatus) 90 Klepto/surface-seizing, suppressor Nearshore Auks
Great black-backed gull (L. marinus) 88 Klepto/surface-seizing, scrounger Nearshore Auks
Lesser black-backed gull (L. fuscus) 86 Klepto/surface-seizing, scrounger Nearshore Auks
Manx shearwater (Puffinus puffinus) 86 Pursuit-plunging, joining Nearshore Auks
Black-legged kittiwake (Rissa tridactyla) 72 Dipping, catalyst Offshore Auks
Northern gannet (Morus bassanus) 62 Plunge-diving, scooping, Offshore Cetaceans
suppressor
North Atlantic fulmar (Fulmarus glacialis) 47 Surface-pecking, joining Pelagic —
Arctic tern (Sterna paradisaea) 38 Shallow-plunging, catalyst Nearshore —
Sandwich tern (S. sandvicensis) 22 Shallow-plunging, catalyst Nearshore —
Common tern (S. hirundo) 9 Shallow-plunging, catalyst Nearshore —
European storm-petrel (Hydrobates pelagicus) 1 Dipping, joining Pelagic —
Black-headed gull (L. ridibundus) 1 Dipping, joining Nearshore —
a
Individuals foraging within MSFAs as a percentage of total observed feeding.
b
Feeding distances: nearshore, <40 km; offshore, 40–80 km; pelagic, >80 km.
 Distribution and foraging interactions 91

Box 6.2 Multispecies feeding associations (MSFAs)

Small, short-lived MSFAs are an important strategy used by numer-
ous species of seabirds to obtain prey (Camphuysen & Webb 1999).
Typically, small, social-feeding flocks of auks drive a dense ball
of fish towards the surface in a concerted effort and exploit this
resource from below (‘producers’; see Fig. 6.3). The term ‘social feed-
ing’ is used, because the auks dive and surface simultaneously and
cooperate in their attempts to drive a fish ball towards the surface.
Actively searching black-legged kittiwakes are normally the first to
discover and exploit the fish ball from above by dipping or shallow-
plunging. As long as only small, surface-feeders such as black-legged
kittiwakes are involved, even when the size of the flock increases sub-
stantially (to 10 to 20 individuals) the producers can continue feed-
ing seemingly undisturbed. When the auks simultaneously surface
for air, the activity of the black-legged kittiwakes normally ceases,
but resumes as soon as the auks dive again. Black-legged kitti-
wakes act as ‘catalysts’ or ‘initiators’ of MSFAs by attracting other
predators. Herring gulls Larus argentatus, great black-backed gulls
Larus marinus and northern gannets Morus bassanus arriving on the
scene typically act as ‘scroungers’ or ‘suppressors’ by taking over
the surface-feeding opportunities from smaller species (interspecific
interference competition). Suppressors attack the fish ball force-
fully, causing producers to swim away and the MSFA breaks down
shortly after. Catalysts normally outnumber producers by a factor of
2; for example, mean flock size (±SE) for black-legged kittiwakes was
9.7 ± 0.9 compared with 4.7 ± 0.3 for common guillemots, and
3.9 ± 0.7 for black-legged kittiwakes versus 2.4 ± 0.2 for razorbills.
A second common type of MSFA for seabirds is generated by hunt-
ing pods of dolphins or harbour porpoises.

surface (Camphuysen & Scott 2003). Inactive periods were recorded more
frequently in black-legged kittiwakes than in common guillemots (Fig. 6.4),
with the latter continuing to feed at certain phases of the tide when black-
legged kittiwakes had stopped entirely. Common guillemot feeding activ-
ity was more evenly spread over the day than that of kittiwakes. Clearly
more surveys are needed to investigate these interspecific differences
further.
92 C. J. Camphuysen et al.

Surface feeders
Joiners
Joiners
(Scroungers,
(Scroungers, (Initiators) kleptoparasites)
kleptoparasites) (Catalysts) (Suppressors)
Sea surface

Fish
ball

Diving seabirds1 or cetaceans2

(Producers, divers)

Producers - scroungers (Beachamp & Giraldeau 1996) 1Mostly

Initiators - joiners (Bayer 1983) auks
Catalysts - divers - suppressors - kleptoparasites (Hoffman et al. 1981) 2Dolphins or harbour porpoises

Fig. 6.3 Schematic representation of an MSFA with diving auks and facilitated
surface-feeders. (Redrawn from Camphuysen and Webb (1999).)

40
Kittiwakes
35 Guillemots
Numbers per 5-min observation

30

25

20

15

10

0
Inc. south Max. south Dec. south Inc. north Max. north Dec. north
Flow direction of tidal current

Fig. 6.4 Foraging black-legged kittiwakes and common guillemots (number ±

SE per 5-min observation) with changing tide during continuous surveys in the
Marr Bank area (the full tidal cycle was surveyed twice, 12–13 June 2003;
56◦ 15 N, 01◦ 30 W). On the x-axis, indications of currents running south
(ebb: increasing (Inc.), maximum (Max.) and decreasing (Dec.)) and north (flood)
during six tidal stages.
 Distribution and foraging interactions 93

Northern gannets were encountered both inshore of the shallow sea

front in mixed coastal waters and further offshore in the deeper, more
stratified regions of the central North Sea. They used contrasting foraging
techniques in the two regions but, unlike many of the other seabirds, the
shallow sea front was less important as a feeding area. In inshore areas,
northern gannets appeared to profit from MSFAs produced by prey-driving
common guillemots and razorbills, with birds alighting or making shallow,
oblique plunge-dives into the frenzy and scooping up sandeels while they
were swimming. In contrast, in offshore waters, gannets usually fed on fish
shoals that were herded towards the surface by dolphins or harbour por-
poises and made vertical, deep plunge-dives (Camphuysen 2004). Of 496
herds of cetaceans recorded in the offshore region, northern gannets tar-
geted 43.3%, a significantly higher frequency than that recorded inshore
(16.2% of 723; Gadj = 108.6, d.f. = 1, p < 0.001). Thus in inshore regions,
northern gannets relied on feeding opportunities created by other seabird
species while in offshore regions they were mainly associated with marine
mammals, predominantly cetaceans.

DISCUSSION

In terms of biomass, the endotherm component of the top-predator com-

munity in the northwest North Sea is dominated by marine mammals,
primarily cetaceans (Fig. 6.2c). Together with predatory fish and, in some
years, an industrial sandeel fishery, marine mammals are likely to be the
major consumers of sandeels in the region. The largest species, the minke
whale, increased during the study period from average densities of 0.001
km−1 surveyed in 1991–5, to 0.002 km−1 in 1997–9, and to 0.005 km−1 in
2001–3. However, the lack of dietary information and consumption rates
for minke whales makes it impossible to assess their impact with any
certainty.
Of the 50 predator species studied during the ship-based surveys
reported here, many were strictly coastal, some were far-ranging, while
others showed intermediate distribution patterns. Near the coast, where
densities of birds and seals were greater and avian species richness was
also higher, interspecific interference competition was presumably most
intense. Many of the top predators were associated with the shallow sea
front marking the transition zone between mixed coastal waters and ther-
mally stratified offshore waters. Black-legged kittiwakes, razorbills, com-
mon guillemots, harbour porpoises and minke whales were all most abun-
dant in this frontal region. Offshore of the shallow sea front, densities
94 C. J. Camphuysen et al.

of seabirds remained high although species richness declined. However,

with relatively high densities of marine mammals, notably minke whales
(Fig. 6.2), exploitation competition may have been more important in this
part of the North Sea.
Water depth throughout most of the study area is less than 60 m and
thus European shags Phalacrocorax aristotelis, razorbills, common guille-
mots, Atlantic puffins and all the marine mammals potentially have access
to the entire water column within their respective foraging ranges. Terns,
black-legged kittiwakes, northern fulmars and storm-petrels rely on the
presence of prey near the water surface, while northern gannets are unlikely
to dive deeper than 20 to 25m (Garthe et al. 2000). These differences in
foraging capabilities have implications as to how prey stocks can be uti-
lized by each predator. Piscivorous seabirds in most of the world’s oceans
exploit fish schools in multispecies flocks and the importance of these
assemblages cannot be over-emphasized (Hoffman et al. 1981, Camphuy-
sen & Webb 1999). Between 20 and 60 km off the coast, black-legged kit-
tiwakes, common guillemots and razorbills together accounted for 80% of
the seabird biomass (Fig. 6.2b). In this region, black-legged kittiwakes read-
ily joined, and profited from, small flocks of common guillemots and razor-
bills driving sandeels and other fish in balls to the surface. Schooling by
small fish does not apparently function as a deterrent to avian predators
in the same way as it does for predatory fish (Brock & Riffenburg 1959).
Most MSFAs included species that used complementary tactics when feed-
ing together (e.g. pursuit-diving, plunge-diving, dipping, scooping, surface-
pecking and aerial-pursuit; see Box 6.2). However, some of the large aerial
species tended to exclude smaller species thereby preventing further access
to the MSFA. Unexpectedly, northern gannets that joined these feeding
frenzies obtained prey by scooping items from the surface rather than by
plunge-diving. Some species – e.g. arctic terns Sterna paradisaea and Euro-
pean storm-petrels Hydrobates pelagicus – rarely, if ever, joined MSFAs; how-
ever, for at least eight other surface-feeding species, MSFAs must have con-
tributed significantly to their daily prey intake (Table 6.1). In the case of
black-legged kittiwakes, at-sea surveys suggested that birds avoided forag-
ing in MSFAs near to the coast where they were more likely to be adversely
affected by Larus gulls and where kleptoparasites such as skuas were most
abundant.
Changes in numbers of many North Sea seabirds over the last 15 to
20 years have varied from long-term increases – e.g. in Atlantic puffins,
common guillemots, razorbills and northern gannets – to declines, e.g. in
black-legged kittiwakes, terns and European shags (Mitchell et al. 2004).
Interestingly, while the reproductive success of sandeel specialists such as
 Distribution and foraging interactions 95

black-legged kittiwakes and shags in eastern Britain fluctuated in parallel (rS

= 0.78, n = 14, p < 0.001), their foraging habits and at-sea distribution
differ radically. In contrast, while distributions of black-legged kittiwakes
and common guillemots in this area appear to overlap, their breeding suc-
cess was not correlated (rS = −0.05, n = 14, not significant). Effects of
reduced prey availability on breeding success are often more pronounced
in surface-feeding seabirds such as black-legged kittiwakes and terns (Mon-
aghan et al., 1992, Rindorf et al., 2000). These findings have led to sug-
gestions that these species are most sensitive to changes in prey availabil-
ity, particularly sandeels (Furness & Tasker 2000). Our survey work has
emphasized the importance of the shallow sea fronts for black-legged kit-
tiwake foraging and indicates that it also forms an outer barrier for birds
breeding down the east coast of Britain (see also Daunt et al. (Chapter 12
in this volume)). In addition, combining information on at-sea distribu-
tion and activity with oceanographic data has highlighted the potentially
complex interplay between seabird breeding success, feeding location and
interspecific competition.
Given the increasing pressures on the North Sea ecosystem from both
fisheries and climate change (Edwards & Richardson 2004, Huntington
et al. 2004), using top predators to monitor ecosystem health is an attrac-
tive concept (Boyd & Murray 2001). However, as the results presented here
clearly indicate, we are still a long way from having all the background
knowledge required for such an approach. Only through multi-disciplinary
projects such as those described here, will we start to understand the func-
tional links between marine predators, their prey and the marine climate –
and thus move towards ecosystem-based fisheries management.

ACKNOWLEDGEMENTS

The work was funded by the European Commission projects ‘Interactions

Between the Marine Environment, Predators and Prey: Implications for
Sustainable Sandeel Fisheries (IMPRESS; QRRS 2000-30864)’ and Mod-
elling the Impact of Fisheries on Seabirds (MIFOS; CFP 96-079)’. We thank
numerous co-observers and the crews of RV Tridens and RV Pelagia for help
in the field. An anonymous referee kindly commented on a draft version of
this chapter.

REFERENCES
Bailey, R. S., Furness, R. W., Gauld, J. A. & Kunzlik, P. A. (1991). Recent changes in
the population of the sandeel (Ammodytes marinus Raitt) at Shetland in relation
to estimates of seabird predation. ICES Mar. Sci. Symp., 193, 209–16.
96 C. J. Camphuysen et al.

Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using

responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Brock, V. E. & Riffenburg, R. H. (1959). Fish schooling: a possible factor in
reducing predation. J. Cons. Int. Explor. Mer., 25, 307–17.
Camphuysen, C. J. (2004). Area-specific feeding behaviour of central place
foraging northern gannets Morus bassanus from the Bass Rock (North Sea).
Oral presentation at 8th International Seabird Group Conference, King’s
College Conference Centre, Aberdeen University, 2–4 April 2004, Aberdeen.
Camphuysen, C. J. & Garthe, S. (2004). Recording foraging seabirds at sea:
standardised recording and coding of foraging behaviour and multispecies
foraging associations. Atlantic Seabirds, 6, 1–32.
Camphuysen, C. J. & Scott, B. (eds.) (2003). IMPRESS Pelagia Cruise, 7–19 June
2003, Wee Bankie (North Sea). Cruise report 64 PE 212. Texel, the Netherlands:
Royal Netherlands Institute for Sea Research.
Camphuysen, C. J. & Webb, A. (1999). Multispecies feeding associations in North
Sea seabirds: jointly exploiting a patchy environment. Ardea, 87, 177–98.
Edwards, M. & Richardson, A. J. (2004). Impact of climate change on marine
pelagic phenology and trophic mismatch. Nature, 430, 881–4.
Furness, R. W. (1990). A preliminary assessment of the quantities of Shetland
sandeels taken by seabirds, seals, predatory fish and the industrial fishery in
1981–83. Ibis, 132, 205–17.
Furness, R. W. & Tasker, M. L. (2000). Seabird–fishery interactions: quantifying
the sensitivity of seabirds to reductions in sandeel abundance, and
identification of key areas for sensitive seabirds in the North Sea. Mar. Ecol.
Prog. Ser., 202, 354–64.
Garthe, S., Benvenuti, S. & Montevecchi, W. A. (2000). Pursuit plunging by
northern gannets (Sula bassana) feeding on capelin (Mallotus villosus). Proc.
R. Soc. Lond. B, 267, 1717–22.
Hammond, P. S. & Fedak, M. A. (eds.) (1994). Grey Seals in the North Sea and their
Interactions with Fisheries. MF 0503, Final Report to the Ministry of Agriculture,
Fisheries and Food. Cambridge, UK: Sea Mammal Research Unit.
Heath, M. R. & Brander, K. (2001). Report of the Workshop on Gadoid Stocks in the
North Sea during the 1960s and 1970s. The Fourth ICES/GLOBEC
Backwards-Facing Workshop. ICES Cooperative Research Report 244.
Copenhagen, Denmark: ICES.
Hoffman, W., Heinemann, D. & Wiens, J. A. (1981). The ecology of seabird feeding
flocks in Alaska. Auk, 98, 437–56.
Huntington, T., Frid, C., Banks, R., Scott, C. & Paramor, O. (2004). Assessment of
the Sustainability of Industrial Fisheries Producing Fish Meal and Fish Oil. Sandy,
UK: RSPB.
Lewis, S., Wanless, S., Wright, P. J. et al. (2001). Diet and breeding performance of
black-legged kittiwakes Rissa tridactyla at a North Sea colony. Mar. Ecol. Prog.
Ser., 221, 277–84.
Longhurst, A. R. (1999). Ecological Geography of the Sea. San Diego, CA: Academic
Press.
Martin, A. R. (1989). The diet of Atlantic puffin Fratercula arctica and northern
gannet Sula bassana chicks at a Shetland colony during a period of changing
prey availability. Bird Study, 36, 170–80.
 Distribution and foraging interactions 97

Mitchell, P. I., Newton, S. F., Ratcliffe, N. & Dunn, T. E. (2004). Seabird

Populations in Britain and Ireland. London: T. & A. D. Poyser.
Monaghan, P., Uttley, J. D. & Burns, M. D. (1992). Effect of changes in food
availability on reproductive effort in arctic terns Sterna paradisaea. Ardea, 80,
71–81.
Ollason, J. G. (ed.) (2000). Modelling the Impact of Fisheries on Seabirds. Final
Report CFP 96–079. Aberdeen, UK: University of Aberdeen.
Otto, L., Zimmerman, J. T. F., Furnes, G. K. et al. (1990). Review of the physical
oceanography of the North Sea. Neth. J. Sea Res., 26, 161–238.
Pingree, R. D. & Griffiths, D. K. (1978). Tidal fronts on the shelf seas around the
British Isles. J. Geophys. Res., 83, 4615–22.
Pingree, R. D., Pugh, P. R., Holligan, P. M. & Forster, G. R. (1975). Summer
phytoplankton blooms and red tides along tidal fronts in the approaches to the
English Channel. Nature, 258, 672–7.
Pollock, C. & Barton, C. (2004). Review of divers, grebes and seaduck distribution
and abundance in the SEA 5 area. Poster presentation at 8th International
Seabird Group Conference, King’s College Conference Centre, Aberdeen
University, 2–4 April 2004, Aberdeen.
Richardson, K., Heath, M. R. & Pedersen, S. M. (1986). Studies of a larval herring
(Clupea harengus L.) patch in the Buchan area, III. Phytoplankton distribution
and primary productivity in relation to hydrographic features. Dana, 6 (Special
issue), 25–36.
Rindorf, A., Wanless, S. & Harris, M. P. (2000). Effects of changes in sandeel
availability on the reproductive output of seabirds. Mar. Ecol. Prog. Ser., 202,
241–52.
Santos, M. B., Pierce, G. J., Learmonth, J. A. et al. (2004). Variability in the diet of
harbour porpoises (Phocoena phocoena) in Scottish waters 1992–2003. Mar.
Mamm. Sci., 20, 1–27.
Simpson, J. H. (1981). The shelf-sea fronts: implications of their existence and
behaviour. Phil. Trans. R. Soc. Lond. A, 302, 531–46.
Tasker, M. L., Jones, P. H., Dixon, T. J. & Blake, B. F. (1984). Counting seabirds at
sea from ships: a review of methods employed and a suggestion for a
standardized approach. Auk, 101, 567–77.
 7
Spatial and temporal variation in the diets of
polar bears across the Canadian Arctic: indicators
of changes in prey populations and environment
S. J. IVERSON, I. STIRLING AND S. L. C. LANG

Polar bears (Ursus maritimus) are broadly distributed in the Arctic and, as
such, have the potential to provide information about changes in ecosystem
structure and functioning over broad scales in time and space. Yet, because
they are so wide-ranging and difficult to observe, there are few quantita-
tive data on polar bear diets or on the ecological (e.g. climate change) and
demographic factors that influence prey selection. We used quantitative
fatty acid signature analysis of polar bear adipose tissue to estimate their
diets in the 1980s/90s across three major regions of the Canadian Arctic:
Davis Strait (n = 70), western Hudson Bay (n = 217) and the Beaufort Sea
(n = 34), using a database of the major prey species in each region (n =
292). Although polar bears consumed ringed and bearded seals throughout
their range, diets differed greatly among regions. Ringed seals accounted
for ≥98% of diet in the Beaufort Sea. In western Hudson Bay, ringed seals
accounted for about 80% of intake in the early 1990s, indicating the impor-
tance of foraging in ice-covered habitat. However, ringed seal consumption
declined throughout the 1990s concurrent with progressively earlier ice
breakup, while the proportions of bearded and harbour seals increased, sug-
gesting reduced reliance on ice. Throughout Davis Strait, harp seals com-
prised 50% of bears’ diets, consistent with the increase in the harp seal
population in this region. Off southern Labrador near the whelping patch,
harp seals accounted for 90% of diets. Hooded seals made up the highest
proportion of bear diets in northern Davis Strait, near their major north-
ern whelping patch. Our results demonstrate that polar bears have a high

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Spatial and temporal variation in the diets of polar bears 99

degree of plasticity in response to changing environments and prey popu-

lations, which suggests that they may be excellent indicators of ecosystem
changes.
Apex predators occupy a special niche in their ecosystems, playing an
important role in ecosystem structure and functioning, but also being
affected by changes at lower trophic levels (e.g. Katona & Whitehead 1988,
Estes 1995, Bowen 1997, Terborgh et al. 1999). Because top predators are
often large, long-lived and geographically wide-ranging, characteristics of
their populations can potentially serve to integrate the cumulative effects of
ecosystem change over a range of spatial and temporal scales. In the Arctic,
polar bears represent the highest trophic level and their circumpolar dis-
tribution is determined by the distribution of their preferred habitat, the
annual ice over the continental shelf. They depend on ice for hunting and
feeding on the seals that use ice as a platform for parturition and lactation,
and for hauling out to rest and moult. Those bears that live on the pack
ice all year round, such as in the Beaufort Sea, move north with the reced-
ing floe edge in summer and south again in winter (Amstrup et al. 2000).
The southernmost populations live year-round in the Hudson and James
bays, Canada, where ice is completely absent for at least 4 months dur-
ing summer and autumn each year, and all bears are forced ashore to fast
until freeze-up, while pregnant females fast for 8 months (Stirling et al.
1977, Ramsay & Stirling 1988). Thus, the presence of sea-ice is critical to
polar bears and changes in its distribution and duration will have a pro-
found impact on their foraging patterns and population ecology (Stirling &
Derocher 1993, Stirling et al. 1999).
Although it is commonly held that polar bears feed predominantly on
ringed seals (Phoca hispida) throughout their range (Stirling & Archibald
1977, Smith 1980), bears also occur in areas where ringed seals are
less common and where populations of other potential prey species have
increased over recent decades. These other prey species could have a signif-
icant influence on the distribution, movements, reproductive success and
dynamics of polar bear populations. Thus, as very widely distributed top
predators in the Arctic, long-term monitoring of the foraging ecology and
diets of polar bears may provide insight into temporal and spatial changes
in the Arctic ecosystem at local, regional and circumpolar scales.
Despite their potential importance, polar bear diets are difficult to study
and estimate quantitatively because of the extreme spatial heterogeneity and
dramatic seasonal variation in prey distribution in the Arctic and the diffi-
culty in directly observing predation. In Canada, there are 14 designated
subpopulations of polar bears that appear to be relatively discrete based on
movements and microsatellite analyses (Paetkau et al. 1999, Taylor & Lee
100 S. J. Iverson et al.

1995), however some individuals may travel distances of up to 6000 km

per year (Amstrup et al. 2000). To date, our understanding of the forag-
ing ecology of polar bears has come predominantly from carcasses of seals
killed by bears and direct observations, primarily from one field site in the
eastern high Arctic. Most of these studies have been conducted between late
March and early July prior to ice breakup, yet bears’ hunting behaviour on
the coastal landfast ice and adjacent floes may not be representative of the
entire year. Thus, our understanding of polar bear diets is incomplete and
new approaches are needed.
Quantitative fatty acid signature analysis (QFASA; Iverson et al. 2004)
is one such approach. This method is based upon the observation that
the characteristic fatty acid (FA) patterns of prey species are predictably
reflected in the adipose fat reserves of monogastric predators. Polar bears
should be ideally suited to the application of QFASA, given the generally
high fat content of their diets (Stirling, 1974, Smith, 1980) – which limits
biosynthesis of FAs – and because their large depots of body fat are accumu-
lated over months, providing an integrated view of the diet. Polar bears also
are thought to eat relatively few prey species (about two to eight species),
depending on their geographical region (Stirling & Øritsland 1995), which
simplifies diet estimation. We used QFASA to ask how well polar bear diets
reflect relative abundance or changes in prey populations across the Cana-
dian Arctic and thus whether bears might be used as indicators of ecosystem
change.

SAMPLE COLLECTION

We obtained 321 adipose tissue samples from 295 bears (some bears were
sampled in more than one year) across three regions of the Canadian Arc-
tic, representing four subpopulations (Fig. 7.1). Subcutaneous adipose tis-
sue samples from muscle to skin were obtained from the rump either from
biopsies (6-mm biopsy punch) from anaesthetized bears – in association
with tagging and monitoring programmes of the Canadian Wildlife Ser-
vice – or from bears harvested by Inuit hunters. FA composition of subcu-
taneous adipose tissue does not differ significantly among body locations
within individual bears (Thiemann et al. in press).
Given that polar bears consume primarily the blubber of seals and that
the blubber will represent the majority of FAs stored in a seal even if
the entire carcass was consumed, we assumed that the blubber of seals
was representative of prey FAs. We obtained full-depth blubber samples
by live biopsy or from subsistence harvests from 292 individuals (across
age classes) of six species: bearded (Erignathus barbatus), harbour (Phoca
 Spatial and temporal variation in the diets of polar bears 101

80°00' N

KB
NW 75°00' N

NB
n = 17
SB Amundsen LS
70°00' N

Gulf N VM BB
S
n = 17 65°00' N

MC GB
Cumberland Sound
FB n = 28
Frobisher Bay
60°00' N

n = 24
DS 55°00' N
Northern Labrador
n = 11
Churchill WHn =217 Southern
Labrador
n=7
SH

Fig. 7.1 Currently designated Canadian polar bear subpopulations.

Abbreviations are: BB, Baffin Bay; DS, Davis Strait; FB, Foxe Basin; GB, Gulf of
Boothia; KB, Kane Basin; LS, Lancaster Sound; MC, M’Clintock Channel; NB,
Northern Beaufort Sea; NW, Norwegian Bay; SB, Southern Beaufort Sea; SH,
Southern Hudson Bay; VM, Viscount Melville Sound; WH, Western Hudson
Bay. Circles indicate the areas and associated sample sizes of polar bears
sampled in the present study: NB, SB, WH and four regions within the DS
population. Most (95%) of the samples were taken throughout the 1990s,
although a few were available from the early 1970s and 1980s. In NB, SB and
DS, bears were mostly sampled on the ice in late winter and spring. In WH,
most bears were sampled on land after ice breakup.

vitulina), harp (Phoca groenlandica), hooded (Cystophora cristata) and ringed

seals and walrus (Odobenus rosmarus). Bearded and ringed seals were sam-
pled from the same regions as the bears. Given their regional distribution,
walrus, and harp and hooded seals were sampled from the Davis Strait
(DS) designation (Fig. 7.1) and harbour seals from the western Hudson
Bay (WH). Although polar bears are known to occasionally feed on belu-
gas (Delphinapterus leucas) and narwhals (Monodon monoceros), as well as
other prey (e.g. Smith & Sjare 1990, Derocher et al. 1993, 2000), we felt we
had obtained the species most likely to comprise the majority of diets of the
bears from the populations included in this study.

VARIATION IN FA COMPOSITION OF POLAR BEARS

AND PREY

Lipid was quantitatively extracted from samples according to Iverson et al.

(2001a) and FA composition was analysed according to Budge et al. (2002)
102 S. J. Iverson et al.

26 (a)
Prey
24
22 Bearded seal (n = 39)
20 Harbour seal (n = 7)
18
Harp seal (n = 133)
Hooded seal (n = 17)
16
Ringed seal (n = 86)
14
Walrus (n = 10)
12
10
8
Mass per cent of total fatty acids

6
4
2
0
14:0 16:0 16:1 18:0 18:1 18:1 18:2 20:1 20:1 20:4 20:5 22:1 22:1 22:5 22:6
n-7 n-9 n-7 n-6 n-11 n-9 n-6 n-3 n-11 n-9 n-3 n-3

26 (b)
24 Polar bears
22 SB (n = 17)
20 NB (n = 17)
18 WH (n = 217)
16 DS south (n = 18)
14
DS north (n = 52)
12
10
8
6
4
2
0
14:0 16:0 16:1 18:0 18:1 18:1 18:2 20:1 20:1 20:4 20:5 22:1 22:1 22:5 22:6
n-7 n-9 n-7 n-6 n-11 n-9 n-6 n-3 n-11 n-9 n-3 n-3

Fig. 7.2 Selected FAs (mean + sem; 15 of the 67 FAs identified and quantified,
representing 87% to 88% of all FAs in both seals and polar bears) that were most
abundant and/or exhibited the greatest average variance across all groups: (a) in
prey species averaged across all regions of the Canadian Arctic sampled and (b)
in polar bears within the four sampled subpopulations. Given the large range in
latitude of bears sampled in the DS population, these were separated into
northern and southern areas ( for abbreviations and areas see Fig. 7.1).

and Iverson et al. (2002). All seals and bears contained the same major
FAs, however, the FA composition varied markedly among prey species and
among bears across locations (Fig. 7.2a and b). Many of the key dietary FAs
such as 20:1n-9, 20:5n-3, 22:1n-11 and 22:6n-3, tended to characterize both
prey and bears in different regions, but means of individual FAs are a lim-
ited way to view data. Discriminant function analyses test predicted group
 Spatial and temporal variation in the diets of polar bears 103

membership and classification success, but also serve to illustrate spatial

relationships between groups. Prey were generally accurately identified to
species by their FA signature, but there was also indication of variation
within species across the Arctic (Fig. 7.3a). Bearded and ringed seals were
each classified to species with 93% to 97% accuracy, and to geographical
location within species with 75% to 85% accuracy, indicating some regional
variation in food webs and diets of those species. Harp and hooded seals
were each classified with about 80% accuracy, but exhibited the greatest
overlap; when misclassified, it was generally harp seal wrongly classified as
hooded seal and vice versa. Walrus differed most from other species and
were classified with 100% success. Hierarchical cluster analyses (in which
a single average for each group is tested therefore allowing a larger subset
of FAs to be used) supported these results, with species being the major
grouping factor – within which there was regional variation. Together these
results are consistent with general differences in the diets of these prey
species (e.g. Bowen & Siniff 1999).
Polar bears exhibited marked variation in FA signatures across regions
of the Arctic (Fig. 7.2b), indicating regional differences in diet. Using dis-
criminant function analysis, bears were 100% correctly classified to their
subpopulations (Fig. 7.3b). Within DS, there was clear indication of dif-
ferences in diets between northern and southern groups with 100% and
94.4% correct classification, respectively. Nevertheless, some overlap was
apparent.

QFASA MODELLING AND POLAR BEAR DIETS

We used the QFASA model developed by Iverson et al. (2004) to perform

simulation studies of diets and to estimate diet composition of polar bears
(Box 7.1). Simulation studies were first performed to confirm the reliabil-
ity with which prey species could be differentiated in the model estimates.
All simulations indicated that prey species were well estimated in speci-
fied diets, providing confidence that these prey species can be reasonably
estimated in the diets of sampled bears. For the purposes of the current
study, we did not attempt to separate age classes of seal prey in simu-
lations or diets, as variation was far greater between than within species
(e.g. Fig. 7.3a, Box 7.1).
The diets of polar bears were estimated using QFASA as modified for
polar bears (Box 7.2). Because of the large spatial scale over which we sam-
pled bears (North Atlantic to Beaufort Sea), we used location-specific prey
assemblages and corresponding FA signatures. Harbour seals and walrus
104 S. J. Iverson et al.

6
(a) Prey
Ringed
2nd discriminant function 4
Harbour BS
WH
WH
2 DS

WH
BS Harp
0 DSn
DS
Bearded
DSo
2 DS
Walrus
Hooded
4

6
10 8 6 4 2 0 2 4
1st discriminant function

6
(b) Bears
4
2nd discriminant function

DS

2 N
WH s

6 BS

8
6 4 2 0 2 4 6 8 10
1st discriminant function
Fig. 7.3 Results of discriminant function analyses of seals and polar bears
across the Canadian Arctic. Due to small sample sizes for some prey species (e.g.
n = 7) and bears in some locations (e.g. n = 17, see Fig. 7.2), we used a reduced
set of FAs (one minus size of the smallest group, respectively) for discriminant
analyses, to offer some assurance that covariance matrices were homogeneous
(Stevens 1986). Here we used FAs that were most abundant and/or exhibited the
greatest average variance across all groups, and analyses were performed
according to Iverson et al. (2002). (a) Plot of the group centroids (within-group
 Spatial and temporal variation in the diets of polar bears 105

do not generally occur in the areas of bears sampled in the northern and
southern Beaufort (NB and SB, Fig. 7.1). Thus these bears were modelled
only on ringed and bearded seals (n = 24 prey) collected in the Beaufort
Sea. Hooded seals do not occur in WH and harp seals or walrus are avail-
able only occasionally. Nevertheless, we modelled WH bears on ringed,
bearded and harbour seals (obtained in WH), and also included harp seals
and walrus (obtained in DS, Baffin Bay (BB) and Foxe Basin (FB), Fig. 7.1;
n = 120 prey). We were not able to obtain samples of harbour seals in DS
given their rarity there, but all other species are potentially available to bears
in this area. Thus, DS bears were modelled using ringed, bearded, harp
and hooded seals and walrus collected in the DS area (50◦ 00 to 70◦ 0 N,
Fig. 7.1; n = 231 prey).
Consistent with differences in FA signatures of polar bears (Figs 7.2b
and 7.3b), their estimated diets differed dramatically across their geographic
range (Fig. 7.6). Although ringed and bearded seals occurred in bear diets
throughout the Canadian Arctic, their relative importance differed greatly.
In both NB and SB, ringed seals comprised about 95% of the FA signa-
tures of polar bear adipose tissue, with the remainder being made up by
bearded seals (Fig 7.6a). However, on average in WH, ringed seal con-
sumption accounted for only 56% of FA signatures, followed by about 38%
bearded and 5% harbour seal; trace levels (<1%) of harp seal also appeared.
In the DS subpopulation, ringed and bearded seals accounted for 18% and
26% of signatures, respectively, whereas harp seals (49%) dominated; some
hooded seals (12%) were also present. While both harp and hooded seals
were reasonably well differentiated in simulations, there was evidence for
some degree of overlap in signatures (Box 7.1; Fig. 7.3a). Thus, the precise
proportions estimated for these two species are somewhat less certain. Wal-
rus appeared at about 1% of diets in DS overall (Fig. 7.6a).

←
Fig. 7.3 (cont.) mean for each discriminant function) for the first and second
discriminant functions for prey species by region. Ellipses represent 95% of the
data points for each species. The first two functions accounted for 86.9% of the
variance among the 11 groups tested (n = 292; Wilk’s λ < 0.001). Species overall
were separated with 86.6% accuracy. For harp seals, DSn and DSo represent
individuals sampled in DS nearshore and DS offshore, respectively. (b) Plot of
the discriminant scores for each individual polar bear (n = 257, removing
repeat-sampled bears and cubs), as well as the group centroids, for the first and
second discriminant functions. NB and SB bears were combined as BS (Beaufort
Sea), due to close proximity of sampling. The first two functions accounted for
95.0% of the variance (Wilk’s λ < 0.001) and individuals were grouped to major
region (Fig. 7.1) with 99.6% accuracy overall; the only misclassifications were
between northern and southern DS.
106 S. J. Iverson et al.

Box 7.1 FA signatures, QFASA and diet simulations

FAs are the main constituent of most lipids. During digestion, FAs
are released from ingested lipid molecules (e.g. triacylglycerols), but
unlike other nutrients – such as proteins – which are readily broken
down, FAs are generally not degraded. The FAs of carbon chain-
length 14 or greater pass into the circulation intact and the frac-
tion not required for immediate metabolic needs is taken up and
deposited in adipose tissue in a predictable way. Since a relatively
limited number of FAs can be biosynthesized by animals, it is pos-
sible to distinguish dietary versus non-dietary components. Numer-
ous studies have demonstrated qualitatively that specific FAs and
patterns of FAs are passed from prey to predator both near the bot-
tom and top of terrestrial and marine food webs (reviewed in Iver-
son et al. (2004)). QFASA is based on the notion that most prey have
characteristic FA signatures (i.e. the quantitative distribution of all
FAs measured; e.g. Budge et al. 2002, Iverson et al. 2002), that these
signatures are deposited in predator adipose tissue in a predictable
way and that by comparing FA signatures of all potential prey to
those of predator adipose tissue using a statistical computer model
(Iverson et al. 2004), one can determine what was eaten. Briefly,
QFASA asks what mix of prey species’ FA signatures comes closest
to matching that of the predator.
Following this approach, the first requirement for estimating
predator diets is a representative database of potential prey species
and an understanding of whether those species can be reliably dis-
tinguished from one another by their FA signature. While various
univariate and multivariate statistical techniques can be used for
such purposes, a powerful means with which to test the ability of
the QFASA model to estimate diet – based solely on differentiating
and quantifying prey species by their FA signatures – is to perform
simulation studies (Iverson et al. 2004). The basic procedures are as
follows: a diet composition of a polar bear was specified (e.g. 80%
ringed seal, 20% bearded seal). For each prey species, the individ-
ual samples were split into two equal sets: a simulation set and a
modelling set. A ‘pseudo bear’ was constructed by sampling from
the simulation set in the proportions specified by the simulated diet.
The ‘diet’ of this pseudo bear was then modelled using the modelling
 Spatial and temporal variation in the diets of polar bears 107

100
90 WH DS
% of FA signature
80
70 a
60
50 a
40
30
20 a a a
10 a
0
a a
Ringed

Walrus

Ringed

Walrus
Harp

Harp

Hooded
Bearded

Bearded
Harbour

Fig. 7.4

set of those prey species plus all other prey in the region’s database.
This creation of the pseudo bear and subsequent modelling was then
repeated 1000 times and gave an indication of mean reliability of
estimates and noise around those estimates.
Figure 7.4 shows the results of the simulation studies for WH
and DS presented as box plots, showing the 25th, median, and 75th
percentiles of the 1000 diet estimates as horizontal bars and with
dots representing outliers. The diet composition specified is repre-
sented in plots as ‘a’ and was designated as follows. For WH, com-
position was designated as 75% ringed seal, 20% bearded seal and
5% harbour seal; for DS, it was 20% ringed seal, 20% bearded seal,
50% harp seal, 8% hooded seal and 2% walrus. The simulations
performed for each region demonstrated the reliability with which
we could differentiate prey species of polar bears in the QFASA
model. In WH, ringed, bearded and harbour seals were well esti-
mated in diets, with only trace amounts of noise from the appearance
of harp seals or walrus. When harp seal and walrus were removed
from the simulations, results were similar except that the propor-
tions of all prey items were precisely predicted as specified. In DS,
species were again well estimated with some overlap between harp
and hooded seals. Ringed seals were only slightly overestimated
and bearded seals only slightly underestimated. Results for the
Beaufort Sea (not shown) also showed precise estimates of specified
diets.
108 S. J. Iverson et al.

Box 7.2 The QFASA model and calibration as applied to

polar bears
Although dietary FAs directly influence predator lipid stores, some
metabolism of FAs does occur within the predator, such that the
composition of predator tissue will not exactly match that of their
prey. Thus, an understanding of and accounting for, the process by
which ingested FAs are metabolized and deposited in tissues of the
predator is fundamental to the use of FAs in food-web studies. An
integral part of the QFASA model is the use of calibration coeffi-
cients to account for predator lipid metabolism by weighting indi-
vidual FAs (Iverson et al. 2004). Previous coefficients have been cal-
culated for blubber of juvenile and adult phocid1 seals fed fish diets
and for suckling phocid pups consuming high-fat milk diets. Since
the blubber of pinnipeds is more structured than that of simple adi-
pose tissue (Iverson 2002), we felt it important to develop a more
appropriate model species for polar bear adipose tissue. We used
data obtained from feeding and fattening mink (Mustela vison) kits
(n = 18) on a long-term homogenous diet supplemented with
marine oil.
Figure 7.5 shows the calibration coefficients for mink (mean ±
SEM of the 10% trimmed means calculated within each individual;
note, in most cases the SE is too small to see) estimated for all 67 FAs
quantified (calculated from Layton (1998) and Iverson et al. (2004)).
The 15 most abundant FAs found in both polar bears and seals are
labelled. The 1:1 line is presented which denotes the deviation of a
given FA in a predator from that consumed in its diet. The calibra-
tion coefficients calculated for mink were generally similar to those
for phocid pups on high-fat diets – although with some exceptions –
and were consistent with our current understanding of how specific
FAs are metabolized (Cooper et al. 2005).
The model proceeds by applying the calibration coefficients to
the FAs of the predator’s stores. It then takes an average, or series
of averages, of each prey FA signature and asks what mix of prey
signatures comes closest to matching a given calibrated predator.
Finally, a statistical distance is calculated between the real predator
and the model’s estimate of that predator and chooses the weighting
that minimizes the statistical distance to represent the best estimate
of diet (see Iverson et al. (2004) for details).
 Spatial and temporal variation in the diets of polar bears 109

Calibration coefficient 4
3
2
1
0

20:1n11
16:0

18:0
14:0

16:1n7

18:1n9
18:1n7

18:2n6

20:1n9

20:4n6

22:1n11

22:6n3
20:5n3
22:1n9

22:5n3
Fatty Acids 1–67

Fig. 7.5

We modelled bears using the two subsets of FAs outlined in Iver-

son et al. (2004) (‘dietary’, which includes only those 33 FAs that
arise solely or mostly from dietary origin; and ‘extended dietary’,
which includes an additional 8 FAs that are abundant in diet but
are also biosynthesized), and using several sets of calibration coeffi-
cients (mink, phocid pup and no calibration). The results of mod-
elling scenarios using the two FA subsets and the different cali-
bration coefficients were generally comparable within major geo-
graphic region, although this depended somewhat on the complex-
ity of the prey database. In the Beaufort Sea (BS), diet estimates of
bears were nearly identical using either dietary or extended FA sub-
sets and using mink, phocid pup or even no calibration; probably
because the two prey species available differed substantially in their
FA signatures. However, in both WH and DS, modelling with the
dietary FA subset and mink calibration provided the most consis-
tent results: we evaluated this by modelling bears in each area with
additional prey added to the model from other areas that we knew
could not occur in the diets due to their geographic range. We found
little, if any, false positives when using dietary FAs and mink cali-
bration. Given that there were a-priori reasons to suggest that mink
calibration would be most appropriate for bears, we report only these
results.
1 Inclusive of all true seals and exclusive of sea lions, fur seals and walruses.

The above diet estimates represent the total mass contribution of each
prey species’ blubber FAs to polar bear diets. However, these species dif-
fer greatly in body size. Although by no means definitive, we used one
simple approach to illustrate the relative difference in diet estimates if one
accounts for these differences in species body mass and thus total blubber
110 S. J. Iverson et al.

Polar bear diets

(a) NB SB WH DS
100

90
% of Fatty acid signature

80

70

60

50

40

30

20

10

0
Ringed

Ringed

Ringed

Ringed
Bearded

Hooded
Bearded

Bearded

Bearded
Walrus

Walrus
Harp

Harp
Harbour

(b) 100
90

80

70
% no. of seals

60

50

40

30

20

10

0
Ringed

Ringed

Ringed

Ringed
Bearded

Bearded

Bearded

Hooded
Walrus

Bearded

Walrus
Harp
Harbour

Harp

Fig. 7.6 QFASA model estimates (mean + SEM) of polar bear diets across the
three major geographical regions and four subpopulations. (a) Data are
presented as the QFASA model output, or the proportional contribution of prey
species to polar bear adipose tissue FA signatures. (b) Data are presented as the
approximate percentage number of seals that would be consumed to account for
these signatures given the large size differences among prey species. For these
estimations, we made the coarse assumption that the entire carcass’s blubber
layer was consumed, that seals contained similar percentages of blubber and,
finally, that both juveniles and adults of each species scaled similarly by body
size. We used the following adult body mass values for scaling: ringed (65 kg),
bearded (300 kg), harbour (87 kg), harp (110 kg) and hooded seal (250 kg); walrus
(1040 kg); ringed seals were used as the divisor.
 Spatial and temporal variation in the diets of polar bears 111

90 WH polar bear diets

80
WH ice breakup
70
1994: late (24 July)
% no. of seals

60 1995–6: mid (14 July)

1998: early (24 June)
50

40

30

20

10

0
Females Males Females Males Females Males
Ringed seals Bearded seals Harbour seals

Fig. 7.7 QFASA estimates of female (n = 86) and male (n = 99) polar bear
diets (mean percentage of relative number of seals ± SEM; see Fig. 7.4) in WH
during a 5-year period in relation to timing of ice breakup (Stirling et al. 1999) in
those years. Only bears >1.5 years of age and sampled in summer after coming
ashore with the ice melt were included. Diets of females and males differed in all
components (p ≤ 0.01), diets decreased in levels of ringed seals and increased in
harbour seals in both females and males (p ≤ 0.01), and diets were variable in
levels of bearded seals (P = 0.236; two-way ANOVA). Results were the same
when expressed as a percentage of FA signature or as converted to a proxy of
relative number of seals (shown).

available (Fig. 7.6b). In this scenario, diets translated into somewhat

different percentages of the relative number of seals taken by bears. Ringed
seals represented the dominant seal taken in the Beaufort Sea (almost
100%) and WH (80%). In contrast, harp seals constituted 60% of seals
taken in DS. Walrus fell to non-detectable levels in both WH and DS overall
(Fig. 7.6b).
Changes in diet were also associated with environmental variation. Polar
bears in WH were sampled in 1994, 1995, 1996 and 1998 – years of pro-
gressively earlier ice breakup (Stirling et al. 1999). Diets in the region dif-
fered significantly between female and male bears in all years, but also
changed in relation to annual timing of ice breakup (Fig. 7.7). Females con-
sumed more ringed seals and fewer bearded seals than did males. However,
between the year of latest ice breakup (1994) and that of the earliest ice
breakup (1 month earlier, 1998), ringed seals declined in the diets of both
sexes, with a corresponding increase in harbour seals and variably increas-
ing numbers of bearded seals (Fig. 7.7).
112 S. J. Iverson et al.

DS polar bear diets

65° 90' N
Ringed Cumberland
Sound
Bearded
Harp
Hooded
Walrus

63° 10' N
Ringed Frobisher
Bearded Bay

Harp
Hooded
Walrus

58° 40' N
Ringed
Northern
Bearded Labrador
Harp
Hooded
Walrus

52° 50' N
Ringed Southern
Bearded Labrador

Harp
Hooded
Walrus
0 10 20 30 40 50 60 70 80 90 100
% no. of seals

Fig. 7.8 QFASA estimates of polar bear diets (mean percentage of the relative
number of seals + SEM; see Fig. 7.6) across the wide range of latitudes of bears
sampled within the DS population (see Fig. 7.1). Bears in southern Labrador
consumed more harp seals and fewer (none) ringed seals than in all other areas
(p < 0.01), and consumed fewer bearded seals than in all areas except
Cumberland Sound (p < 0.01). The greatest numbers of hooded seals were
consumed by bears sampled near Frobisher Bay (p < 0.05). Walrus consumption
did not differ with area (ANOVA with Fisher’s PLSD post-hoc tests).
 Spatial and temporal variation in the diets of polar bears 113

Polar bear diets also varied spatially within a subpopulation as a

function of latitude. Diets differed notably in bears from the southern-
and northernmost areas of the DS subpopulation (Fig. 7.8). Harp seals
accounted for 90% of seals consumed by the seven bears sampled in south-
ern Labrador, whereas they comprised a lower and relatively constant pro-
portion of the seals consumed further north. In contrast, ringed seals were
absent from diets of southern Labrador bears but gradually increased in
diets to the north, as did bearded seals. Hooded seals were consumed most
in the areas near Frobisher Bay. Walrus were estimated as appearing in diets
of only eight individuals from northern areas of DS: five males and three
females. In three bears, walrus were estimated to contribute a large amount
to FA signatures (11% to 24%), although this translated into only 1% to 5%
of the seals consumed.

POLAR BEARS AS INDICATORS OF ARCTIC ECOSYSTEMS

AT REGIONAL AND CONTINENTAL SCALES

The results of our study provide new insight into the foraging ecology of
polar bears across their Canadian range. Although FAs have been used
previously to make inferences about changes in lower trophic levels in an
ecosystem (Iverson et al. 1997, 2001b), this is the first time FAs have been
used to estimate quantitatively the diets of a top predator in relation to spa-
tial and temporal changes in prey availability.
Previous studies, based on direct observation, have concluded that
ringed seals are the dominant prey of polar bears in the Arctic, followed by
bearded seals (e.g. Stirling 2002). However, climate and access have lim-
ited most studies of polar bear foraging to late winter and early spring, and
to areas of landfast sea-ice and immediately adjacent pack-ice where ringed
seals are most abundant, thus potentially biasing our concept of their impor-
tance in the diet. Nevertheless, ringed seals clearly remained an important
component of polar bear diets in this study (Fig. 7.6). However, their dom-
inance (and even presence) in diets across spatial and temporal scales dif-
fered dramatically. In the bears sampled in NB and SB in the early 1990s,
ringed seals were consumed almost exclusively, consistent with their distri-
bution and abundance relative to bearded seals in those areas (e.g. Stirling
2002). Similarly, ringed seals were the dominant seal consumed in WH,
but this varied substantially by individual, sex and year (e.g. Fig. 7.7). Male
bears fed more on bearded seals than did females, which is consistent with
expectations based on their larger body size and more frequent (albeit lim-
ited) observations of male bears seen over bearded seal kills than females
114 S. J. Iverson et al.

(Stirling & Derocher 1990). This has interesting implications for evaluating
male foraging and consumption. A bearded seal is over four times heavier
than a ringed seal, and thus a bear would only need to catch a single bearded
seal to equal the blubber intake of four to five ringed seals. Future determi-
nation of age classes of seals taken will also influence these conclusions.
Despite potential sex differences in foraging tactics, both sexes responded
to environmental variability in similar ways and thus both served as indica-
tors of short-term ecosystem change.
During the 1990s in WH, the trend towards progressively earlier sea-
ice breakup dates was accompanied by significant decreases in ringed seals
in polar bear diets. Through the same period, Holst et al. (1999) and Stir-
ling (2004) documented low apparent survival of ringed seal pups in 1998–
2000. It is not clear whether the increase in the proportion of bearded and
harbour seals in the diet of WH bears through the 1990s reflects a decline
in the availability of ringed seals, an increase in the population size and
availability of bearded and harbour seals, or both. However, all species are
mainly only available to bears on the ice so these changes in diet, espe-
cially reduction in ringed seals, complement evidence that during the same
period bears came ashore earlier and in progressively poorer condition, with
a decline in both physical and reproductive characteristics (Stirling et al.
1999).
Besides short-term temporal changes, diet composition of apex preda-
tors can characterize, and signal a shift in, the abundance of lower trophic
levels. In WH, bear diets reflected their prey field as an assemblage dom-
inated by ringed and bearded seals, with increasing dependence on both
harbour and bearded seals, coincident with climate warming as predicted
by Stirling and Derocher (1993). Little or no harp seal and walrus occurred
in diets, consistent with their known geographic distributions, further sup-
porting the notion that the FAs of polar bears reflect the prey available. In
DS, bear diets reflected a longer-term trend in prey abundances. Although
it was previously known, mostly from chance encounters, that polar bears
hunted harp seals, we have for the first time shown that harp seals are the
dominant prey in this region, along with some intake of hooded seals. This
coincides with the large and well-documented increases that have occurred
in these seal populations since the early 1970s (NOAA 1999, DFO 2000)
and with apparently increasing polar bear numbers observed in this area
(I. Stirling, unpublished data, 2005). Harp seals were most abundant in
the diets of bears in southern Labrador, which is closest to the harp seal
whelping patch, while hooded seals were most abundant in diets of bears
near Frobisher Bay (63◦ 10 N), which is closest to the northern hooded
seal whelping patch (62◦ to 64◦ N). These results are consistent with both
 Spatial and temporal variation in the diets of polar bears 115

known geographic distributions and abundances of prey species, but also

with large and longer-term changes in population size. While we have
focused the present study on the major prey of polar bears, future studies
should seek to include more minor, and potentially increasing, prey species.
Overall, we conclude that polar bear diets accurately reflect changes
in prey populations and thus are useful indicators of ecosystem change.
Changes occurring in polar bear diets in WH were consistent with relatively
short-term indications of temporal changes. In contrast, diets of bears in DS
accurately reflected both long-term trends in species abundance as well as
geographic variation in relation to the availability of different species. Taken
across the entire Canadian Arctic, from NB and SB to WH and DS, the dif-
ferences in diets of polar bears reflected both the distribution and abun-
dance of species across a continental scale. Long-term changes in climate
and ecosystems are occurring. Perhaps, nowhere is this more apparent than
in the Arctic (e.g. Tynan & DeMaster 1997). We conclude that long-term
monitoring of the diets of polar bears, using the methods we have set forth,
along with data on other aspects of their population dynamics, will provide
invaluable information relevant to understanding changes in Arctic ecosys-
tems at both regional and continental scales.

ACKNOWLEDGEMENTS

We thank the following for their support of the research that made this
project possible: the Natural Sciences and Engineering Research Council
(NSERC) Canada; the Canadian Wildlife Service (CWS); the Polar Continen-
tal Shelf Project; the World Wildlife Fund and the Churchill Northern Stud-
ies Centre. We are particularly grateful to the many people from the North-
west Territories Department of Resources, Wildlife, and Economic Devel-
opment; the Nunavut Department of Sustainable Development (especially
F. Piugattuk and J. Beauchesne); the Labrador Inuit Association; the Inu-
vialuit Fisheries Joint Management Committee and the CWS for collecting
fat samples from subsistence harvesting programmes and ongoing research
projects on both polar bears and seals. We also thank G. Stenson for pro-
viding most of the harp and hooded seal samples. J. Lasner and C. Beck
assisted with some sample analyses and W. Blanchard assisted with the
QFASA modelling. We thank W. D. Bowen, I. L. Boyd and A. E. Derocher
for providing helpful comments on an earlier version of the manuscript.

REFERENCES
Amstrup, S. C., Durner, G. M., Stirling, I., Lunn, N. & Messier, F. (2000).
Movements and distribution of polar bears in the Beaufort Sea. Can. J. Zool.,
78, 948–66.
116 S. J. Iverson et al.

Bowen, W. D. (1997). Role of marine mammals in aquatic ecosystems. Mar. Ecol.

Prog. Ser., 158, 267–74.
Bowen, W. D. & Siniff, D. (1999). Distribution, population biology, and feeding
ecology of marine mammals.: In Biology of Marine Mammals, eds. J. E.
Reynolds, III & S. A. Rommel. Washington, DC: Smithsonian Institution
Press, pp. 423–84.
Budge, S. M., Iverson, S. J., Bowen, W. D. & Ackman, R. G. (2002). Among- and
within-species variation in fatty acid signatures of marine fish and
invertebrates on the Scotian Shelf, Georges Bank and southern Gulf of St.
Lawrence. Can. J. Fish. Aquat. Sci., 59, 886–98.
Cooper, M. H., Iverson, S. J. & Heras, H. (2005). Dynamics of blood chylomicron
fatty acids in a marine carnivore: implications for lipid metabolism and
quantitative estimation of predator diets. J. Comp. Physiol., 175, 133–45.
Derocher, A. E., Andriashek, D. & Stirling, I. (1993). Terrestrial foraging by
polar bears during the ice-free period in western Hudson Bay. Arctic, 46, 251–4.
Derocher, A. E., Wiig, Ø. & Bangjord, G. (2000). Predation of Svalbard reindeer by
polar bears. Polar Biol., 23, 675–8.
DFO (Department of Fisheries and Oceans) (2000). Northwest Atlantic Harp Seals.
DFO Science Stock Status Report E1-01. Dartmouth, Nova Scotia: Department
of Fisheries and Oceans.
Estes, J. A. (1995). Top level carnivores and ecosystem effects: questions and
approaches. In Linking Species and Ecosystems, eds. C. G. Jones & J. H. Lawton.
Toronto, Canada: Chapman and Hall, pp. 151–8.
Holst, M., Stirling, I. & Calvert, W. (1999). Age structure and reproductive rates of
ringed seals (Phoca hispida) on the northwestern coast of Hudson Bay in 1991
and 1992. Mar. Mamm. Sci., 15, 1357–64.
Iverson, S. J. (2002). Blubber. In Encyclopedia of Marine Mammals, eds. W. F. Perrin,
B. Wursig & H. G. M. Thewissen. San Diego, CA: Academic Press, pp. 107–12.
Iverson, S. J., Arnould, J. P. Y. & Boyd, I. L. (1997). Milk fatty acid signatures
indicate both major and minor shifts in diet of lactating Antarctic fur seals.
Can. J. Zool., 75, 188–97.
Iverson, S. J., Lang, S. L. C. & Cooper, M. (2001a). Comparison of the Bligh and
Dyer and Folch methods for total lipid determination in a broad range of
marine tissue. Lipids, 36, 1283–7.
Iverson, S. J., MacDonald, J. & Smith, L. K. (2001b). Changes in diet of
free-ranging black bears in years of contrasting food availability revealed
through milk fatty acids. Can. J. Zool., 79, 2268–79.
Iverson, S. J., Frost, K. J. & Lang, S. (2002). Fat content and fatty acid composition
of forage fish and invertebrates in Prince William Sound, Alaska: factors
contributing to among and within species variability. Mar. Ecol. Prog. Ser., 241,
161–81.
Iverson, S. J., Field, C., Bowen, W. D. & Blanchard, W. (2004). Quantitative fatty
acid signature analysis: a new method of estimating predator diets. Ecol.
Monogr., 74, 211–35.
Katona, S. & Whitehead, H. (1988). Are cetacea ecologically important? Oceanogr.
Mar. Biol. Annu. Rev., 26, 553–68.
Layton, H. (1998). Development of digestive capabilities and improvement of diet
utilization by mink pre- and post-weaning with emphasis on gastric lipase.
M.Sc. thesis, Nova Scotia Agricultural College, Truro.
 Spatial and temporal variation in the diets of polar bears 117

NOAA (National Oceanic and Atmospheric Administration) (1999). US Atlantic

and Gulf of Mexico Marine Mammal Stock Assessments. NOAA Technical
Memorandum NMFS-NE-153. Washington, DC: NOAA.
Paetkau, D., Amstrup, S. C., Born, E. W. et al. (1999). Genetic structure of the
world’s polar bear populations. Mol. Ecol., 8, 1571–85.
Ramsay, M. A. & Stirling, I. (1988). Reproductive biology and ecology of female
polar bears (Ursus maritimus). J. Zool. Lond., 214, 601–34.
Smith, T. G. (1980). Polar bear predation of ringed and bearded seals in the
land-fast sea ice habitat. Can. J. Zool., 58, 2201–9.
Smith, T. G. & Sjare, B. (1990). Predation of belugas and narwhals by polar bears
in nearshore areas of the Canadian High Arctic. Arctic, 43, 99–102.
Stevens, J. (1986). Applied Multivariate Statistics for the Social Sciences. Mahwah, NJ:
Lawrence Erlbaum Associates.
Stirling, I. (1974). Midsummer observations on the behavior of wild polar bears
(Ursus maritimus). Can. J. Zool., 52, 1191–8.
(2002). Polar bears and seals in the Eastern Beaufort Sea and Amundsen Gulf: a
synthesis of population trends and ecological relationships over three decades.
Arctic, 55 (Suppl. 1), 59–76.
(2004). Reproductive rates of ringed seals and survival of pups in Northwestern
Hudson Bay, Canada, 1991–2000. Polar Biol., 28, (DOI:
10.1007/s00300-004-0700-7).
Stirling, I. & Archibald, W. R. (1977). Aspects of predation of seals by polar bears.
J. Fish. Res. Board Can., 34, 1126–9.
Stirling, I. & Derocher, A. E. (1990). Factors affecting the evolution of the modern
bears. Int. Conf. Bear Res. Managmt, 8, 189–205.
(1993). Possible impacts of climatic warming on polar bears. Arctic, 46, 240–5.
Stirling, I. & Øritsland, N. A. (1995). Relationships between estimates of ringed
seal and polar bear populations in the Canadian Arctic. Can. J. Fish. Aquat. Sci.,
52, 2594–612.
Stirling, I., Jonkel, C., Smith, P., Robertson, R. & Cross, D. (1977). The Ecology of
the Polar Bear (Ursus maritimus) Along the Western Coast of Hudson Bay.
Canadian Wildlife Service Occasional Paper 33. Ontario, Canada: Canadian
Wildlife Service.
Stirling, I., Lunn, N. J. & Iacozza, J. (1999). Long-term trends in the population
ecology of polar bears in western Hudson Bay in relation to climatic change.
Arctic, 52, 294–306.
Taylor, M. & Lee, J. (1995). Distribution and abundance of Canadian polar bear
populations: a management perspective. Arctic, 48, 147–54.
Terborgh, J., Estes, J. A., Paquet, P. et al. (1999). The role of top carnivores in
regulating terrestrial ecosystems. In Continental Conservation: Scientific
Foundations of Regional Reserve Networks, eds. M. E. Soule & J. Terborgh.
Washington, DC: Island Press, pp. 39–64.
Thiemann, G. W., Iverson, S. J. & Stirling, I. Seasonal, sexual, and anatomical
variability in the adipose tissue composition of polar, bears (Ursus maritimus).
J. Zool. Lond., in press.
Tynan, C. T. & DeMaster, D. P. (1997). Observations and predictions of Arctic
climatic change: potential effects on marine mammals. Arctic, 50, 308–22.
 8
Biophysical influences on seabird trophic
assessments
W . A . M O N T E V E C C H I , S . G A R T H E A N D G . K . D AV O R E N

The foraging behaviour and ecology of top predators are expressions

of trophic and ecosystem dynamics. Oceanographic fluctuations as well
as biological interactions affect exothermic species and, through them,
influence their endothermic predators. Planktivorous and piscivorous,
surface-feeding and diving seabirds exhibit varying constraint, flexibility,
specialization and opportunism in their responses to prey and environ-
mental conditions. Responses can be direct in terms of foraging behaviour,
prey capture and diet; or indirect in terms of egg and chick produc-
tion, growth, breeding success, recruitment and population change. Pro-
tracted indirect effects lag behind and buffer environmental change with
behaviour and life-history attributes. Focal forage species that fuel large
vertebrate food webs exhibit extreme fluctuations in abundance, being
highly sensitive to biophysical perturbations, including fishing. Changes
in their biology often shift ecosystems to alternative states, yet forage
species are understudied. Indications about forage species derived from
seabirds can be broadly informative. Synoptic meso-scale studies that link
colony measurements to vessel surveys of prey and predators within avian
foraging ranges provide an approach for assessing predator responses
to variation in prey fields and oceanography. Tracking free-ranging for-
agers with animal-borne data loggers (which record temperature, pres-
sure, activity and position) details behavioural solutions to current con-
ditions. These foraging tactics of individual predators are mechanisms
of the social and population responses that we measure, estimate and
model. Physical data from loggers, vessels and satellites can be combined to

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Biophysical influences on seabird trophic assessments 119

define thermal habitats and ‘hotspots’ used by predators and prey. Regional
and global comparisons aid in understanding changing meso-scale pro-
cesses and ocean-scale patterns for effective ocean conservation and
management.
Top-predator responses to changes in prey and environmental con-
ditions pose pressing questions about food webs, ecosystem dynamics,
climate change and conservation. In marine systems, physical processes
impose more pervasive influences on animals than in fresh water. Physi-
cal events as well as biological interactions determine the distributions and
abundance of exothermic animals which in turn drive the distributions and
activities of their endothermic predators (Springer et al. 1984).
Owing to mobility, high metabolic rates and limited lipid storage,
many seabirds maintain relatively close spatial and temporal associations
with prey (Russell et al. 1992, Davoren et al. 2002). Comparisons of
short- and long-term behavioural, dietary, distributional, production and
population responses of surface-feeding and diving seabirds that feed at
different trophic levels reflect changes in prey and ecosystem conditions
(Montevecchi 1993, Springer & Speckman 1997). Surface-feeders often
have more extensive foraging ranges and are more sensitive to surface-
water effects on prey distributions than divers. Feeding at lower trophic lev-
els, planktivores are often more affected by oceanographic fluctuations than
piscivores.
Seabird–prey interactions are most spatially constrained, yet highly
energy-demanding, during breeding seasons when birds forage from land-
based colonies (fixed-place foraging). During these periods, birds are highly
attuned to prey conditions and often develop traditional colony-specific for-
aging areas (e.g. Cairns 1989, Grémillet et al. 2004). Over decades and cen-
turies, colonies tend to be in areas of predictable prey availability and, conse-
quently, changes in breeding distributions and populations can be related to
long-term oceanographic change (Montevecchi & Myers 1997, Ainley 2002,
Emslie et al. 2003).
Seabird foraging ecology has been studied indirectly by sampling the
diets of chicks and adults at colonies and by surveying distributions and den-
sities from vessels. Diet samples at colonies reveal what birds have preyed
on but not where it came from or its distribution or relative abundance
compared with other prey. Surveys at sea associate avian occurrences with
oceanography and prey fields. Synoptic integration of these approaches is
significantly more informative about where predators captured prey and
options that they might have had in the process. Yet, to determine mech-
anisms of predator responses, direct studies of the foraging tactics of
120 W. A. Montevecchi et al.

individuals are needed to fully integrate colony and oceanographic research

(Croll et al. 1998, Hooker et al. 2002).
Studies using animal-borne data loggers and satellite tags (which record
temperature, pressure, activity and position) attached to free-ranging ani-
mals permit assessments of how foraging tactics vary with changing prey
fields and oceanography (Jouventin & Weimerskirch 1990, Ancel et al.
1992, Benvenuti et al. 1998, Garthe et al. 2000). Physical and positional
measurements (Wilson et al. 2002) in association with diet sampling and
vessel and satellite surveys of prey and oceanography can detail habitat use
and predator–prey overlap.
Here, we briefly overview the dietary and foraging responses of seabirds
to changing prey and ocean conditions over multiple spatial and temporal
scales. We consider multispecies approaches and emphasize oceanographic
factors that modulate trophic interactions by influencing prey distributions
and varying the scales of prey patchiness that endotherms have to negoti-
ate (Boyd 1996). Examples from the northwest Atlantic are used to illustrate
the influences of a regime shift on a focal forage species (capelin) and hence
on top predators. Information about predator responses to changing condi-
tions is maximized in synoptic meso-scale colony and vessel research inte-
grated with studies of free-ranging predators carrying data-storage loggers.
Such investigation helps to define the marine habitats of predators and prey,
their shifting overlap and how these change with oceanography and climate.

INDICATOR RESPONSES

Response variables include behavioural (e.g. foraging effort) and longer-

term, behaviourally mediated responses (e.g. recruitment). Protracted indi-
rect responses are buffered by behavioural and life-history adaptations
(Burger & Piatt 1990, Montevecchi & Berruti 1991) and lag behind envi-
ronmental change (Cairns 1987, Thompson & Ollason 2001).
Predator responses often reflect broader environmental conditions and
are heuristically valuable. For example, breeding failure is hypothesized to
indicate changes in prey bases (Montevecchi 1993, Reid & Croxall 2000),
mediated by ocean physics that constrain the distributions of exotherms
or by fishing (Montevecchi 2001). Predators either negotiate these stochas-
tic relationships successfully or die, resulting in powerful selection for
life-history strategies that reflect spatial and temporal uncertainty in food
supplies. Long-term trends in multispecies indices are most informative
(Anderson & Piatt 1999), and non-linear functional responses to prey abun-
dance direct attention to extreme rather than gradual prey fluctuations
(Boyd & Murray 2001).
 Biophysical influences on seabird trophic assessments 121

Fig. 8.1 Example of a simple multispecies assessment of food webs and

ecosystem conditions in the northwest Atlantic. For food-web data relating to this
figure see Table 8.1.

Multispecies assessments

Incorporation of planktivores, benthic feeders, forage-fish specialists, gen-

eralists and predators of large pelagic prey in multispecies studies involves
considerable effort but provides substantially more information than single-
species approaches and is essential for understanding food-web dynamics.
Planktivores and piscivores, and surface-feeders and divers, often respond
differently to environmental perturbations and changes (Baird 1990,
Barrett 1996, Kitaysky & Golubova 2000, Jones et al. 2002). Surface-
feeders are highly sensitive to sea-surface temperatures which influence
vertical prey distributions. In contrast, owing to physiological and ener-
getic constraints associated with having both flying and diving capabili-
ties (Birt-Friesen et al. 1989), pursuit-divers are generally better able to
cope with vertical changes in prey distributions but less able to cope
with extensive horizontal scales. Differences in foraging and breeding suc-
cess between surface-feeders and divers have mainly been attributed to
deeper distributions of forage fish (Barrett 1996, Regehr & Montevecchi
1997). A selection of three avian species that are being used to probe
food webs in the northwest Atlantic is depicted in Fig. 8.1 and details
given in Table 8.1. These seabirds span the ranges of body sizes
(energetics), populations, foraging modes and trophic levels exploited by
122 W. A. Montevecchi et al.

Table 8.1. Food-web relationships for Fig. 8.1

Order of magnitude
Index
species Mass (g) Population Max. dive Trophic level

Gannet g × 103 104 20 m Large pelagics

Guillemot g × 102 105 100 m Forage fish
Storm-petrel g × 10 106 0m Mytophids,
crustacean

avian predators in the region and were selected to maximize the efficient
assessment of prey and environmental conditions using top predators.

OCEANOGRAPHIC INFLUENCES ON PREY ABUNDANCE

AND AVAILABILITY

Variations in hydrography and fishing frequently induce major fluctuations

among small pelagic species (e.g. Cury et al. 2000, Chavez et al. 2004). Yet
even large fluctuations are at times difficult to detect, with smaller ones
being even less evident. Oceanography also drives the horizontal and verti-
cal distributions and patchiness of exothermic animals and their availability
to predators. Spatial and temporal mismatches between prey and preda-
tors carry profound consequences for upper trophic levels. How species
respond differently to this variation depends on their foraging modes and
constraints (Fig. 8.2a), with prey availability often exerting a greater effect on
predator catch per unit effort than prey abundance (Montevecchi & Berruti
1991, Gjerdrum et al. 2003). To cope with oceanographic stochasticity,
seabirds often exploit predictable prey patches (Skov et al. 2000, Davoren
et al. 2003), yet functional responses and stock recruitment indicators are
difficult to define when changes in predator responses and diets can reflect
changes in prey abundance, distribution and accessibility or some combi-
nation of these (Fig. 8.2).

REGIME SHIFTS

Surprisingly, some of the most profound oceanographic changes also take

considerable time to document, may be impossible to predict and hence
have remained relatively intractable to scientific scrutiny. Low-frequency,
high-amplitude oceanographic events, referred to as ‘regime shifts’, involve
sudden pervasive long-term changes in ecosystem state (Collie et al. 2004,
 Biophysical influences on seabird trophic assessments 123

(a)
Gulls Surface feeders

Colony

Puffins Murres
Capelin

Gulls Surface feeders

(b) Colony

Puffins Murres

Capelin

Gulls Surface feeders

Colony
(c)
Puffins Murres
Capelin

Fig. 8.2 Circumstances in which forage fish are: (a) available to surface-feeders
(arrows represent horizontal ranges) and divers (vertical ranges are indicated by
the position of the ellipsoid beneath the sea level); (b) unavailable to surface-
feeders due to a change in vertical distribution; (c) available to surface-feeders
and divers despite a change in abundance represented by the size of the dashed
ellipsoids.

Cury & Shannon 2004). Definitions of regime shift converge on a few com-
monalities, i.e. rapid onset, physically forced, ‘bottom-up’ perturbations that
initiate decadal or longer systemic food-web and ecological changes (Steele
1996, 1998, Hare & Mantua 2000). Shifts can also be forced from the ‘top’
by over-fishing, and ecosystems with food webs dominated by mid-trophic-
level forage fishes are especially sensitive to such effects (Cury et al. 2000).
The persistence of regime shifts is due in part to lagged biological responses
to physical perturbation. Following a regime shift, a system must be ‘forced’
again to change states, being unlikely to return to its former condition.
124 W. A. Montevecchi et al.

Alternative interpretations of regime shifts (drawn primarily from closed

freshwater systems) focus on multiple causality based on gradual internal
processes as well as sudden external forcers (Scheffer et al. 2001).
Owing to lack of predictability, analysis of marine regime shifts has been
retrospective and has focused on precipitating events. Because natural dis-
turbance is an integral aspect of ecosystem process, suppression is not a
fruitful management strategy (Holling & Meffe 1996).Yet, human-induced
disturbances and their environmental interactions are amenable to scien-
tific experimentation and adaptive management with the objective of main-
taining population and system resilience (Scheffer et al. 2001).
Seabird diets, distributions and populations have been greatly influ-
enced by regime shifts (Anderson & Piatt 1999, Kitaysky & Golubova 2000).
Interestingly, signals from seabirds can provide insights into the occur-
rences, consequences and predictability of pervasive ecosystem changes.

Food-web shift in the northwest Atlantic

Seabird research has revealed a number of consequences of a recent

cold-water event in the northwest Atlantic. Long-term studies gave ini-
tial indications of a major shift in pelagic food webs during the 1990s
(Montevecchi & Myers 1995, 1996). A multi-decadal shift from warm-
to cold-water prey taken by gannets generated the hypothesis that large,
migratory pelagic species ceased moving into the region after a major
perturbation in cold surface water in 1991 (Drinkwater 1996; Fig. 8.3).
Contrasting breeding success by divers (auks) and failure by surface-feeders
(gulls; Montevecchi 1996, Regehr & Montevecchi 1997) generated the
hypothesis that cold surface water kept capelin at depth, inaccessible to
surface-feeders (see Fig. 8.2). Subsequent demonstration of a significant
downward shift in the vertical distribution of capelin (Mowbray 2003) sup-
ported the hypothesis. Parental prey deliveries at the world’s largest com-
mon guillemot colony revealed that the capelin condition decreased through
the 1990s, even though surface waters had warmed to pre-perturbation lev-
els by 1996 (Davoren & Montevecchi 2003).

FOCAL FORAGE SPECIES

Large vertebrates in many eastern boundary currents and other ecosystems

depend heavily on focal forage fishes or crustaceans (Springer & Speckman
1997). Examples of forage species include: anchovies and sardines in east-
ern boundary currents; sandeels in the North Sea, Shetlands and Pacific;
 Biophysical influences on seabird trophic assessments 125

100

Cold
80
Percentage of landings

Warm

60

40

20

0
1978 1983 1988 1993 1998 2003
Year

Fig. 8.3 Shift from landings of warm-water prey (e.g. mackerel, short-finned
squid) to cold-water prey (capelin, Atlantic salmon) by breeding northern
gannets at Funk Island in the northwest Atlantic.

sprats in the Baltic Sea; capelin in the Barents Sea and northwest Atlantic;
pollack in Alaskan waters; krill in the Antarctic. Biophysical influences on
forage species generate systemic changes that permeate pelagic food webs
(Anderson & Piatt 1999, Cury et al. 2000) in the Baltic Sea. Unfortunately,
forage species are often not well studied, either because they are not com-
mercially pursued or because they are undervalued as a raw material by
industrial fisheries (Aikman 1997). Seabird research involving focal for-
age species provides insight and predictive capability about their population
resiliency and about their roles in potential trophic shifts (Hatch & Sanger
1992, Springer & Speckman 1997, Barrett 2002, Davoren & Montevecchi
2003, Miller & Sydeman 2004, Wanless et al. 2004).

THERMAL HABITATS

Thermal oceanographic changes exert powerful influences on exotherms

and, via prey availability, determine intersections between predators and
prey. Study of the habitat associations of marine animals offers an effec-
tive way to integrate environmental context into trophic investigations and
models. Physical data from free-ranging predators aid in assessments of the
thermal habitat of both predators and prey. Spatial and temporal concentra-
tions of prey have broad implications for top-predator foraging strategies
as well as for conservation and management. Thermal and other physical
and biophysical features that tend to concentrate prey cause the foraging
distributions of many predators to aggregate and overlap. These ‘hotspots’
are sites where energy transfer among trophic levels is maximized.
126 W. A. Montevecchi et al.

SYNOPTIC COLONY- AND VESSEL-BASED RESEARCH

Considering the research options currently available, the most comprehen-

sive way of investigating the biophysical complexity of food webs involves
engagement of meso-scale, multi-disciplinary oceanographic programmes
and modelling (Yodzis 1994, Scott et al. (Chapter 4 in this volume)). Inte-
grating multi-disciplinary studies of diets and foraging absences at colonies
with synoptic vessel surveys of prey and birds within avian foraging ranges
provides the best opportunity to obtain predator and functional responses
(see Asseburg et al. (Chapter 18 in this volume)) to changing food and ocean
conditions. By linking these approaches with studies of predators carrying
data storage tags (temperature, pressure (dive depths), activity and position),
decision-making and foraging tactics can be identified. These tactics are the
behavioural mechanisms of predator–prey interactions, and they generate
the higher-level population patterns and responses that we measure, esti-
mate and model (May et al. 1979).

MAXIMIZING INFORMATION FROM TOP PREDATORS

Long-term biophysical studies are essential for comprehensive assessment

of the uncertainty inherent in ecological and evolutionary processes. Studies
of top predators can be designed to probe and anticipate food-web dynam-
ics and to improve understanding of the biological effects of oceanographic
change (Cairns 1987, Montevecchi 1993, Boyd & Murray 2001, Barrett
2002, Jones et al. 2002, Miller & Sydeman 2004, Wanless et al. 2004). The
more that this research can be integrated into multi-disciplinary oceano-
graphic programmes with emphasis on biophysical interactions, the more
informative it will be.
Research directed at top predators at the limits of oceanographic
domains, where species are presumably at or near limits of their physio-
logical and behavioural tolerances, will help maximize information about
ocean-climate change. The Sea of Okhotsk in the western Pacific, the north-
west Atlantic where Arctic waters make their lowest latitudinal penetrations
and north Norway where boreal waters move inside the Arctic Circle are
representative sites where such studies could be conducted.
Biological oceanographic comparisons require access to control or ref-
erence conditions, as free as possible of human-induced influence, in order
to provide environmental baselines with which to gauge change. These
sites are becoming increasingly rare, but are nonetheless globally available.
Long-term studies of top predators in such sites and documentations of
 Biophysical influences on seabird trophic assessments 127

‘hotspots’, especially when they are within marine protected areas (Schmidt
1997), will help provide the archival oceanographic information required
to place environmental changes in an ecological context and will facilitate
effective conservation and management.

ACKNOWLEDGEMENTS

We thank the following : Ian Boyd for the opportunity to participate

in the interesting symposium that provided the forum for this volume;
Deborah Body of the Zoological Society of London for logistic support;
Chantelle Burke for figure preparation; the Natural Sciences and Engineer-
ing Research Council of Canada, and Fisheries and Oceans Canada, for
supporting our long-term research on trophic interactions in the northwest
Atlantic.

REFERENCES
Aikman, P. (1997). Industrial ‘Hoover’ Fishing: A Policy Vacuum. Amsterdam, the
Netherlands: Greenpeace.
Ainley, D. (2002). Adélie Penguin: Bellweather of Climate Change. New York:
Columbia University Press.
Ancel, A., Kooyman, G. L., Ponganis, P. J. et al. (1992). Foraging behaviour of
emperor penguins as a resource detector. Nature, 360, 336–9.
Anderson, P. J. & Piatt, J. F. (1999). Community reorganization in the Gulf of
Alaska following ocean climate regime shift. Mar. Ecol. Prog. Ser., 189, 117–23.
Baird, P. H. (1990). Influence of abiotic factors and prey distribution on diet and
reproductive success of three seabird speices in Alaska. Ornis Scand., 21,
224–35.
Barrett, R. T. (1996). Prey harvest, chick growth, and production of three seabird
species on Bleiksøy, North Norway during years of variable food availability. In
Studies of High-Latitude Seabirds. 4. Trophic Relationships and Energetics of
Endotherms in Cold Ocean Systems, ed. W. A. Montevecchi. Canadian Wildlife
Service Occasional Paper 91. Ottowa, Ontario, Canada: Canadian Wildlife
Service, pp. 20–26.
(2002). Puffin and guillemot chick food and growth as indicators of fish stocks
in the Barents Sea. Mar. Ecol. Prog. Ser., 230, 275–87.
Benvenuti, S., Bonadonna, F., Dall’Antonia, L. & Gudmundsson, G. A. (1998).
Foraging flights of breeding thick-billed murres carrying direction recorders.
Auk, 115, 57–66.
Birt-Friesen, V., Montevecchi, W. A., Cairns, D. K. & Macko, S. A. (1989). Activity
specific metabolic rates of free living seabirds with emphasis on Northern
Gannets Sula bassanus. Ecology, 70, 357–67.
Boyd, I. L. (1996). Temporal scales of foraging in a marine predator. Ecology, 77,
426–34.
Boyd, I. L. & Murray, W. (2001). Monitoring marine ecosystems with responses of
upper trophic level predators. J. Anim. Ecol., 70, 747–60.
128 W. A. Montevecchi et al.

Burger, A. E. & Piatt, J. F. (1990). Flexible time-budgets in common murres:

buffers against variable prey abundance. Stud. Avian Biol., 14, 71–83.
Cairns, D. K. (1987). Seabirds as indicators of marine food supplies. Biol.
Oceanogr., 5, 261–71.
(1989). Regulation of seabird colony size: a hinterland model. Am. Nat., 134,
141–6.
Chavez, F. P., Ryan, J., Lluch-Cota, S. E. & Ñiquen, C. M. (2004). From anchovies
to sardines and back: multi-decadal change in the Pacific Ocean. Science, 299,
217–21.
Collie, J. S., Richardson, K. & Steele, J. H. (2004). Regime shifts: can ecological
theory illuminate mechanisms? Prog. Oceanogr., 20, 281–302.
Croll, D. A., Tershy, B. R., Hewitt, R. P. et al. (1998). An integrated approach to the
foraging ecology of marine birds and mammals. Deep-Sea Res. Part II, 45,
1353–71.
Cury, P. & Shannon, L. (2004). Regime shifts in upwelling ecosystems: observed
changes and possible mechanisms in the northern and southern Benquela.
Prog. Oceanogr., 60, 223–43.
Cury, P., Bakun, A., CrawFord, R. J. M. et al. (2000). Small pelagics in upwelling
systems: patterns of interaction and structural changes in ‘wasp-waist’
ecosystems. ICES J. Mar. Sci., 57, 603–18.
Davoren, G. K. & Montevecchi, W. A. (2003). Signals from seabirds indicate
changing biology of capelin. Mar. Ecol. Prog. Ser., 258, 253–61.
Davoren, G. K., Montevecchi, W. A. & Anderson, J. (2002). Scale-dependent
associations of predators and prey: constraints imposed by flightlessness of
murres. Mar. Ecol. Prog. Ser., 245, 259–72.
(2003). Search strategies of a pursuit-diving seabird. Ecol. Monogr., 73, 463–81.
Drinkwater, K. (1996). Atmospheric and oceanic variability in the northwest
Atlantic during the 1980s and early 1990s. J. Northw. Atl. Fish. Sci., 18, 77–97.
Emslie, S. D., Berkman, P. A., Ainley, D. G., Coats, L. & Pollito, M. (2003). Late
Holocene initiation of ice-free ecosystems in the southern Ross Sea, Australia.
Mar. Ecol. Prog. Ser., 262, 19–25.
Garthe, S., Benvenuti, S. & Montevecchi, W. A. (2000). Pursuit-plunging by
gannets feeding on capelin. Proc. R. Soc. Lond. B, 267, 1717–22.
Gjerdrum, C., Vallee, A. M., St Clair, C. C. et al. (2003). Tufted puffin reproduction
reveals ocean climate variability. Proc. Natl Acad. Sci. U. S. A., 100, 9377–82.
Grémillet D., Dell’Omo, G., Ryan, P. G. et al. (2004). Offshore diplomacy, or how
seabirds mitigate intra-specific competition: a case study based on GPS
tracking of cape gannets from neighbouring colonies. Mar. Ecol. Prog. Ser.,
268, 265–79.
Hare, S. & Mantua, N. (2000). Empirical evidence of North Pacific regime shifts in
1977 and 1989. Prog. Oceanogr., 47, 103–45.
Hatch, S. A. & Sanger, G. A. (1992). Puffins as predators on juvenile pollock and
forage fish in the Gulf of Alaska. Mar. Ecol. Prog. Ser., 80, 1–14.
Holling, C. & Meffe, G. (1996). Command, control and the pathology of natural
resource management. Conserv. Biol., 10, 328–37.
Hooker, S. K., Boyd, I. L., Jessop, M. et al. (2002). Monitoring the prey-field of
marine predators: combining digital imaging with datalogging tags. Mar.
Mamm. Sci., 18, 680–7.
 Biophysical influences on seabird trophic assessments 129

Jones, I. L., Hunter, F. M. & Robertson, G. J. (2002). Annual adult survival of Least
Auklets (Aves, Alcidae) varies with large-scale climatic conditions of the North
Pacific Ocean. Oecologia, 133, 38–44.
Jouventin, J. & Weimerskirch, H. (1990). Satellite tracking of wandering
albatrosses. Nature, 343, 746–8.
Kitaysky, A. S. & Golubova, E. G. (2000). Climate change causes contrasting
trends in reproductive performance of planktivorous and piscivorous alcids.
J. Anim. Ecol., 69, 248–69.
May, R., Beddington, J. R., Clark, C. W., Holt, S. J. & Laws, R. M. (1979).
Management of multispecies fisheries. Science, 205, 267–77.
Miller, A. K. & Sydeman, W. J. (2004). Rockfish response to low-frequency
ocean climate change as revealed by the diet of a marine bird over multiple
time scales. Mar. Ecol. Prog. Ser., 281, 207–16.
Montevecchi, W. A. (1993). Birds as indicators of change in marine prey stocks. In
Birds as Monitors of Environmental Change, eds. R. W. Furness & J. Greenwood.
London: Chapman and Hall, pp. 217–66.
(ed.) (1996). Studies of High-Latitude Seabirds. 4. Trophic Relationships and
Energetics of Endotherms in Cold Ocean Systems. Canadian Wildlife Service
Occasional Paper 91. Ontario, Canada: Canadian Wildlife Service.
(2001). Interactions between fisheries and seabirds. In Biology of Marine Birds,
eds. E. A. Schrieber & J. Burger. Boca Raton, FL: CRC Press, pp. 527–57.
Montevecchi, W. A. & Berruti, A. (1991). Avian indication of pelagic fishery
conditions in the southeast and northwest Atlantic. Acta Int. Ornithol. Cong.,
20, 2246–56.
Montevecchi, W. A. & Myers, R. A. (1995). Seabird prey harvests reflect pelagic fish
and squid abundance on multiple spatial and temporal scales. Mar. Ecol. Prog.
Ser., 117, 1–9.
(1996). Dietary changes of seabirds reflect shifts in pelagic food webs. Sarsia, 80,
313–22.
(1997). Centurial and decadal oceanographic influences on changes in northern
gannet populations and diets in the Northwest Atlantic: implications for
climate change. ICES J. Mar. Sci., 54, 608–14.
Mowbray, F. (2003). Changes in the vertical distribution of capelin off
Newfoundland. ICES J. Mar. Sci., 59, 942–9.
Regehr, H. & Montevecchi, W. A. (1997). Effects of food shortage and predation on
breeding failure of kittiwakes. Mar. Ecol. Prog. Ser., 155, 249–60.
Reid, K. & Croxall, J. P. (2000). Environmental response of upper trophic-level
predators reveals system change in an Antarctic ecosystem. Proc. R. Soc. Lond.
B, 268, 377–84.
Russell, R., Hunt, G. L., Coyle, K. & Cooney, K. (1992). Foraging in a fractal
environment: spatial patterns in a marine predator–prey system. Landscape
Ecol., 7, 195–209.
Scheffer, M., Carpenter, S., Foley, J. A., Folke, C. & Walker, B. (2001). Catastrophic
shifts in ecosystems. Nature, 413, 591–6.
Schmidt, K. (1997). ‘No take’ zones spark fisheries debate. Science, 277, 489–91.
Skov, H., Durinck, J. & Andell, P. (2000). Associations between wintering avian
predators and schooling fish. J. Avian Biol., 31, 135–43.
130 W. A. Montevecchi et al.

Springer, A. M. & Speckman, S. G. (1997). A forage fish is what? Summary of

symposium. In Proceedings of the International Symposium on the Role of Forage
Fishes in Marine Ecosystems. Alaska Sea Grant College Program Publication
AK-SG-97–01. Fairbanks, Alaska: University of Alaska Fairbanks, pp. 773–805.
Springer, A. M., Roseneau, D. G., Murphy, E. C. & Springer, M. I. (1984).
Environmental controls of marine food webs: food habits of seabirds in the
eastern Chukchi Sea. Can. J. Fish. Aquat. Sci., 41, 1202–15.
Steele, J. (1996). Regime shifts in fisheries management. Fish., Res., 25, 19–23.
(1998). Regime shifts in marine ecosystems. Ecological Applic., 8, S33–6.
Thompson, P. M. & Ollason, J. C. (2001). Lagged effects of ocean climate change
on fulmar population dynamics. Nature, 413, 417–20.
Wanless, S., Wright, P. J., Harris, M. P. & Elston, D. A. (2004). Evidence for a
decrease in size of lesser sandeels Ammodytes marinus in a North Sea
aggregation over a 30-yr period. Mar. Ecol. Prog. Ser., 279, 237–46.
Wilson, R. P., Grémillet, D., Syder, D. et al. (2002). Remote-sensing systems and
seabirds: use, abuse and potential for measuring environmental variables. Mar.
Ecol. Prog. Ser., 228, 241–61.
Yodzis, P. (1994). Predator–prey theory and management of multispecies fisheries.
Ecological Applic., 4, 51–8.
 9
Consequences of prey distribution for the
foraging behaviour of top predators
I . J . S T A N I L A N D , P . T R AT H A N A N D A . R . M A R T I N

Prey distribution and dynamics have a strong effect on the foraging

behaviour of marine predators. Increasingly sophisticated logging devices
allow us to measure the behaviour of these predators while they are forag-
ing. In this chapter we show how these behavioural measures can be used
to provide insight into the dispersion and patchiness of krill using Antarctic
fur seals and macaroni penguins as example predators. We illustrate how
a sound understanding of their ecology and the constraints on their forag-
ing is needed to interpret these data. Examples are provided of how simple
measures such as the organization of diving and dive depths have been
used to detect differences in the distribution of krill both temporally and
geographically.
The food web of the Southern Ocean is dominated by Antarctic krill
Euphausia superba. Krill occur in loose aggregations and dense swarms, both
of which may extend in size from a few metres to several tens of kilometres
across (Miller & Hampton 1989). Krill abundance is also subject to large
seasonal and annual fluctuations (Brierley et al. 1999). Despite the high
levels of spatial and temporal variability, in some areas krill occur in pre-
dictable quantities such that large populations of top predators – including
penguins, seals, whales and commercially harvested fish species – rely on
it as their main food source.
The highly productive oceanic region around South Georgia supports
numerous predators that depend on krill as their principal prey. This
includes an estimated 2.75 million macaroni penguins (Trathan et al. 1998)
and over 3 million Antarctic fur seals (Barlow et al. 2002). These two species

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
132 I. J. Staniland et al.

are among the most numerous avian and mammalian krill-dependent

predators at South Georgia (Boyd 2002). Because of their reliance upon
krill, these predators provide an insight into long- and short-term variabil-
ity in the marine environment at South Georgia (Croxall (Chapter 11 in this
volume)). Simple analyses of their diet, abundance and distribution can pro-
vide estimates of marine resources and the presence of key prey species over
broad geographic areas (Boyd & Murray 2001). Indices of breeding success
such as offspring production, growth rates and mortality can also be used
to detect changes in the abundance of prey in the surrounding waters (Boyd
& Murray 2001, Reid & Croxall 2001).
The increasing miniaturization and sophistication of electronic logging
devices has now allowed us to measure the activity of animals while they
forage with minimal disturbance to their natural behaviour. Through the
analysis of such bio-logging data we can infer the distribution of both preda-
tors and their prey in space and time. In this chapter we discuss the use
of such devices to provide information about South Georgia predators and
their prey.

CURRENT USE OF BEHAVIOURAL MEASURES

During the summer months at South Georgia the diet of female Antarc-
tic fur seals is dominated (88% frequency of occurrence) by krill (Reid
& Arnould 1996). With such a reliance on one prey species, changes in
the abundance of krill influences fur seal foraging (Mori & Boyd 2004)
and breeding success (Reid & Croxall 2001) at South Georgia. In years
of low krill availability females must increase their foraging effort, spend-
ing around 40% longer at sea and 25% less time ashore (McCafferty et al.
1998).
Monitoring basic behavioural responses, such as trip duration, can
therefore potentially be used as a measure of krill availability (CCAMLR
1995). However, trip duration is highly variable between individual seals,
even for those operating in the same environmental conditions. Recent
work has shown that there is a strong individual component to their
behaviour, with some seals consistently undertaking longer trips than oth-
ers (Staniland et al. 2004). Fur seals are able to adjust their behaviour at a
number of scales (not just overall trip length) in order to maximize their
rate of energy gain (Boyd 1996, Mori & Boyd 2004). As a result, they
are able to reproduce successfully under a wide range of environmental
conditions.
 Prey distribution and top-predator foraging behaviour 133

Box 9.1 Differences in energy demands

The metabolic rate of foraging macaroni penguins has been esti-
mated at 9 W kg−1 (Green et al. 2002), 50% higher than the 6 W kg−1
estimated for female fur seals (Arnould et al. 1996). However, the
larger body mass of fur seals (average 36 kg) than macaroni pen-
guins (3.6 kg) dictates a more than six-fold higher energy demand
(216 versus 32.5 J s−1 for seals and penguins respectively). Relating
this to their diving behaviour and their foraging patterns, we can
develop a picture of what each species needs to consume in terms
of krill per dive. This simplistic model (Table 9.1) does not account
for provisioning of offspring, time spent fasting ashore or assimi-
lation efficiency. However, the figures show that in order to forage
successfully fur seals must capture a considerably greater number
of krill per dive. This means that macaroni penguins can potentially
exploit krill at far lower densities than do fur seals.

PREDATOR ECOLOGY AND CONSTRAINTS

ON FORAGING

In order to fully realize the potential of predators as monitoring platforms, it

is necessary to understand their biology in some detail. During the summer
months, female Antarctic fur seals follow a predictable pattern of foraging:
spending 2 to 10 days feeding at sea, interspersed with short periods ashore.
As air-breathing aquatic mammals of small size they are restricted to for-
aging within surface waters (<200 m). Compared with other pinnipeds,
such as elephant seals (Mirounga spp.), they are not prolific divers and
departures from the water surface to depth are likely to represent individual
foraging events (Boyd 1996). Macaroni penguins also undertake foraging-
trips during their breeding season but during this time they experience dif-
ferent constraints. Macaroni penguins have a higher metabolic rate per unit
mass than fur seals. However, the larger body size of fur seals means that
they have higher overall metabolic demands. A simple calculation based on
a maintenance diet (ignoring the demands of offspring) shows that fur seals
need to gain as much as eight times the energy from each dive as do mac-
aroni penguins (Box 9.1). During breeding, macaroni penguins provision
the chick by directly regurgitating their stomach contents; this means that,
compared with fur seals, they have smaller capacity for energy delivery and
134 I. J. Staniland et al.

Table 9.1. Comparison of the energetic maintenance costs and predicted

krill consumption of Antarctic fur seals and macaroni penguins in the
different stages of their breeding season

Average Energy (kJ) needed to

foraging trip cover metabolic costs

Days Dives Per trip Per dive No. of krill per divea

Antarctic fur seal 4 1222 79 969 65.44 6.14

Macaroni penguin
Incubation 13 6540 35 340 5.40 0.51
Brood-guard 4 1105 11 092 10.04 0.94
Crèche 5 1432 13 272 9.27 0.87
Premoult 22 9172 61 147 6.67 0.63
Mean 11 4562 30 213 7.84 0.74

a Based on the consumption of gravid female krill (60-mm long and containing 5.45
kJ g−1 ).

less capacity to adapt their foraging strategy to variation in food availability,

and must therefore return more frequently to feed their offspring. In con-
trast, fur seals use fat-rich milk to provision their offspring and although
this confines provisioning to the female it allows greater energy delivery
and increased flexibility in time–energy budgets (Dall & Boyd 2002). As a
result, female fur seals can forage for longer and travel greater distances,
concentrating resources into milk that can be delivered to the pup over a
number of days ashore.
While a fur seal pup can be left unattended after only a week, macaroni
penguin young must be guarded by one parent from egg laying until at least
20 days after hatching. As a result the macaroni penguin breeding season
must be considered as comprising distinct stages: incubation, brood-guard,
crèche (when chicks are old enough to be left alone) and premoult (after the
chick has fledged and when adults forage at sea to build up reserves prior
to their annual moult). These periods place different constraints on adult
penguins and thus their foraging behaviour varies considerably throughout
the breeding season.

USING BEHAVIOURAL MEASURES TO MONITOR

PREY DISTRIBUTION

Satellite telemetry data have allowed the important foraging areas of top
predators to be described (Fig. 9.1). Clearly concentrations of foraging
predators are likely to correspond to areas of high prey availability. Indeed
 Prey distribution and top-predator foraging behaviour 135

42° W 40° W 38° W 36° W

(a)

53° S

54° S

55° S
Relative time

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

42° W 40° W 38° W 36° W

(b)

53° S

54° S

55° S
Relative time

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Fig. 9.1 Density distribution maps for (a) Antarctic fur seals and (b) macaroni
penguins foraging from Bird Island, South Georgia. The density distributions
are contour plots using linear spline interpolation based on the relative time
spent within each 0.1 km × 0.1 km area. Fur seal data represent 31 animals
throughout the breeding season, from December until March. Macaroni penguin
data represent 46 individual trips during January.

such data have been used to define small-scale management units for
Antarctic fisheries (CCAMLR, 2002).

MEASURING THE DISTRIBUTION OF PREY

IN THE WATER COLUMN

Work on fur seal foraging over the last 20 years has given an insight into the
distribution of krill in the water column. Krill swarms, like many other zoo-
plankton, undergo diel vertical migration that is known to vary depending
136 I. J. Staniland et al.

on environmental conditions (Godlewska 1996). Fur seal diving depths

throughout a 24-h period often show a distinct pattern with deeper dives
occurring at dusk, gradually shifting to shallow surface dives around the
middle of the night (Fig. 9.2a). As dawn approaches the dives again deepen
until diving virtually ceases during daylight hours. This is thought to reflect
the availability of krill within surface waters (Croxall et al. 1985). The con-
centration of diving at night is assumed to mirror the migration of krill into
surface layers during the hours of darkness. As deeper diving is presumed
to have increased costs for the seals there will be a maximum depth below
which foraging on krill is not profitable. The reduced level of daytime diving
coincides with the descent of krill to depths outside the foraging range of
fur seals.
Telemetry data from Bird Island, South Georgia have highlighted differ-
ences in the diving behaviour of Antarctic fur seals that were foraging in dif-
ferent water masses (Staniland & Boyd 2003). Seals that foraged in oceanic
waters a long way (>105 km) from the breeding beaches dived mostly at
night (75%) and to relatively shallow depths. Seals foraging within shelf and
shelf-break waters dived during both the day and night. For these shelf for-
agers the split between day and night diving was almost 50:50 with a marked
increase in the depth of daytime dives. If we assume that the observed diur-
nal pattern of fur seal diving was the result of the diel vertical migration
of krill, we would expect changes in fur seal behaviour to reflect changes
in these migration patterns both in their timing and extent. The continua-
tion of diving throughout the day and night suggests that the diel vertical
migration of krill in shelf waters was reduced and that there were exploitable
quantities of krill close to the surface during daylight hours.
Patterns in fur seal diving depths also vary through the breeding season.
Dive depths increase from the beginning of the breeding season until the
end of December and decrease thereafter (Boyd & Croxall 1996, McCafferty
et al. 1998). In these studies, the greatest dive depths corresponded to
the longest period of daylight. This fits the photoperiodic response of
krill, which are thought to migrate progressively further from the surface
between spring and mid summer and then reduce their vertical migration
towards autumn (Godlewska 1996).

PREY DISPERSAL AND THE ORGANIZATION OF DIVING

Using simple models, it is evident that fur seals need to exploit krill at higher
densities than macaroni penguins (Box 9.1). This is further highlighted by
the contrast in the organization of diving in the two species.
 Prey distribution and top-predator foraging behaviour 137

(a)
0

0
Depth (m)

50

100
00:00 00:00 00:00 00:00
(b)

0
Depth (m)

50

100
17:00 19:00 21:00
Local time

Fig. 9.2 Examples of typical dive records from an individual Antarctic fur seal
(a and b) and macaroni penguin (c and d). Organization of dives is shown at the
trip level and expanded to show four hours in detail. The unfilled and filled
rectangles represent the day–night cycle. Barred lines show dives grouped into
bouts separated by inter-bout periods (curled brackets). Dives marked with a
cross were not included in bouts.
138 I. J. Staniland et al.

(c)
0
Depth (m)

50

100
00:00 00:00

(d)

0
Depth (m)

50

100
11:00 01:00 03:00
Local time

Fig. 9.2 (cont.)

Fur seal diving is organized at a number of different time scales. At the

broad scale there is a diurnal pattern of diving with a concentration of activ-
ity at night (Fig. 9.2a). Within these intensive foraging periods dives are
grouped into bouts. These bouts are separated by longer periods when the
seal remains at the surface (Fig. 9.2b). During these inter-bout periods seals
swim actively, averaging 1.8 m s−1 (Boyd 1996).
 Prey distribution and top-predator foraging behaviour 139

Diving bouts are thought to reflect the exploitation of krill in patches.

Using mean swimming speeds it is possible to estimate the distances trav-
elled between bouts and to use the different levels of organization in fur seal
diving to infer the distribution of krill. Indeed the spacing of prey patches
indicated by inter-bout periods is at a similar scale to the spacing of krill
measured during ship-based hydro-acoustic surveys (Miller & Hampton
1989, Miller et al. 1993, Boyd 1996).
In contrast to fur seals, macaroni penguins dive more during the day
than at night (Green et al. 2003). Dives take place continually through-
out daylight hours with little consistency in the depths of successive dives
(Fig. 9.2c). Dives begin almost immediately after birds leave the colony.
Also satellite tracking data shows that birds cover large distances (e.g.
80 km) throughout the daylight foraging period (BAS, unpublished data).
Thus, given the difference in foraging behaviour, it is likely that maca-
roni penguins are not exploiting krill in the same patch structure as fur
seals.
For species such as fur seals that dive in bouts, we can use the bouts
to make both spatial and temporal predictions of prey distribution at scales
that directly affect predators. For example, seals foraging in different water
masses show differences in their diving behaviour (Box 9.2). Seals foraging
in shelf waters spent significantly longer in each bout (patch) than animals
foraging in oceanic waters. Based upon optimal foraging theory there are
two reasons for increasing patch residency time: either the quality of patches
is reduced, or the travel time between patches increases. In this case the
travel time between bouts was not different for the two areas, suggesting
that patches were of lower quality in shelf waters. This reduced quality could
be because patches had a lower energy density or were further from the
surface (and therefore were energetically more costly to exploit).
Bout times also show significant differences between years (Boyd 1996).
During a year of low krill abundance when pup growth was reduced, the
analysis of bouts showed that seals increased their time spent foraging
within patches, but the time spent travelling between patches remained con-
stant. This suggests that in the year of low krill availability it was the patch
quality (density) that was affected rather than the spacing and number of
patches.
Further evidence that diving behaviour is strongly influenced by prey
behaviour is seen in studies of individual seals. Staniland et al. (2004),
showed that there was a strong individual component to where an individ-
ual seal foraged, but the geographical region (water mass) most strongly
influenced the diving behaviour of the seal. Animals foraging in shallow
140 I. J. Staniland et al.

Box 9.2 Behavioural differences between foraging locations

Trips by 12 lactating female seals foraging from Bird Island, South
Georgia were categorized into two groups based on the locations of
their foraging. Six trips where the animals foraged in water of depth
greater than 1000 m were classified as oceanic. On the remaining
six trips, females concentrated their foraging close to the breeding
beach in shelf waters (water depth <1000 m).
Within each trip dives were classified into bouts. A starting cri-
terion of three dives occurring within a 15-min period was used to
define a bout. An iterative process was then used to decide whether
the next dive should be included in the bout. A Student’s t-test
was used to compare the surface interval between the bout and the
next dive with the mean surface interval within the bout. When the
surface interval was significantly different to the mean of the bout,
the bout was deemed to have ended (Boyd et al. 1994).
To restrict the analysis to the time when the animal was actively
foraging, only the portion of the trip between the first and last div-
ing bout was used. The mean time spent in bouts and the mean time
spent travelling between bouts were compared between the two for-
aging areas. As the times within a trip were not independent, the
mean values from each trip were used as the sampling unit in the
analysis. The distributions of bout times and travel times within a
trip were highly skewed and were log transformed before means
were calculated.
Seals foraging in shelf waters spent significantly greater time
within bouts (10.6 min) than those foraging in oceanic waters
(7.7 min) (t-test equal variance: t = −2.40, d.f. = 10, p = 0.04).
However, there was no difference in the time spent travelling
between bouts for the two foraging areas (shelf, 6.0 min; oceanic,
6.6 min) (t-test unequal variance: t = −0.54, d.f. = 6.9, p = 0.60).

water made longer, deeper dives and spent longer at the bottom of the dive
than those foraging in deep water.

CONCLUSION

Prey distribution and dynamics have a strong effect on the foraging

behaviour of marine predators. Increasingly sophisticated logging devices
 Prey distribution and top-predator foraging behaviour 141

allow us to measure the behaviour of these predators while they are for-
aging. This can provide an insight into the dispersion and patchiness of
prey populations in a way that would otherwise be impossible or at least
impractical. However, perhaps the biggest challenge is to condense the
results of such analyses to a form that is useful for management purposes.
In this context, there is a growing recognition of the need for functional
response relationships between the environment and marine predators,
and for an understanding of the behavioural mechanisms that lead to par-
ticular response functions (Mori & Boyd 2004). These are critical to using
predator information in the monitoring and management of marine ecosys-
tems. By understanding their behaviour when foraging, we can make pre-
dictions about critical prey densities at which predators are unable to main-
tain their energy intake. Thus behavioural data are critical for any under-
standing of how functional response curves relate the abundance of prey to
predator performance indices.

REFERENCES

Arnould, J. P. Y., Boyd, I. L. & Speakman, J. R. (1996). The relationship between

foraging behaviour and energy expenditure in Antarctic fur seals. J. Zool. Lond.,
239, 769–82.
Barlow, K. E., Boyd, I. L., Croxall, J. P. et al. (2002). Are penguins and seals
in competition for Antarctic krill at South Georgia? Mar. Biol., 140,
205–13.
Boyd, I. L. (1996). Temporal scales of foraging in a marine predator. Ecology, 77,
426–34.
(2002). Estimating food consumption of marine predators: Antarctic fur seals
and Macaroni penguins. J. Appl. Ecol., 39, 103–19.
Boyd, I. L. & Croxall, J. P. (1996). Dive durations in pinnipeds and seabirds. Can. J.
Zool., 74, 1696–705.
Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Boyd, I. L., Arnould, J. P. Y., Barton, T. & Croxall, J. P. (1994). Foraging behaviour
of Antarctic fur seals during periods of contrasting prey abundance. J. Anim.
Ecol., 63, 703–13.
Brierley, A. S., Watkins, J. L., Goss, C., Wilkinson, M. T. & Everson, I. (1999).
Acoustic estimates of krill density at South Georgia 1981–1998. CCAMLR Sci.,
6, 47–57.
CCAMLR (Convention on the Conservation of Antarctic Marine Living Resources)
(1995). CEMP Standard Methods. Hobart, Australia: CCAMLR.
(2002). Report of the 21st meeting of the scientific committee on ecosystem
monitoring and management. Hobart, Australia: CCAMLR.
Croxall, J. P., Everson, I., Kooyman, G. L., Ricketts, C. & Davis, R. W. (1985). Fur
seal diving behaviour in relation to vertical distribution of krill. J. Anim. Ecol.,
54, 1–8.
142 I. J. Staniland et al.

Dall, S. R. X. & Boyd, I. L. (2002). Provisioning under the risk of starvation. Evol.
Ecol. Res., 4, 883–96.
Godlewska, M. (1996). Vertical migrations of krill (Euphausia superba Dana). Pol.
Arch. Hydrobiol., 43, 9–63.
Green, J. A., Butler, P. J., Woakes, A. J. & Boyd, I. L. (2002). Energy requirements
of female Macaroni penguins breeding at South Georgia. Funct. Ecol., 16,
671–81.
(2003). Energetics of diving in Macaroni penguins. J. Exp. Biol., 206, 43–57.
McCafferty, D. J., Boyd, I. L., Walker, T. R. & Taylor, R. I. (1998). Foraging
responses of Antarctic fur seals to changes in the marine environment. Mar.
Ecol. Prog. Ser., 166, 285–99.
Miller, D. G. M. & Hampton, I. (1989). Krill aggregation characteristics: spatial
distribution patterns from hydroacoustic observations. Polar Biol., 10, 125–34.
Miller, D. G. M., Barange, M., Klindt, H.et al. (1993). Antarctic krill aggregation
characteristics from acoustic observations in the South West Atlantic Ocean.
Mar. Biol., 117, 171–83.
Mori, Y. & Boyd, I. L. (2004). The behavioral basis for nonlinear functional
responses and optimal foraging in Antarctic fur seals. Ecology, 85, 398–410.
Reid, K. & Arnould, J. P. Y. (1996). The diet of Antarctic fur seals Arctocephalus
gazella during the breeding season at South Georgia. Polar Biol., 16, 105–14.
Reid, K. & Croxall, J. P. (2001). Environmental response of upper trophic-level
predators reveals a system change in an Antarctic marine ecosystem. Proc. R.
Soc. Lond. B, 268: 377–84.
Staniland, I. J. & Boyd, I. L. (2003). Variation in the foraging location of Antarctic
fur seals (Arctocephalus gazella), the effects on diving behaviour. Mar. Mamm.
Sci., 19, 331–43.
Staniland, I. J., Boyd, I. L. & Reid, K. (2004). Comparing individual and spatial
influences on foraging behaviour in Antarctic fur seals. Mar. Ecol. Prog. Ser.,
275, 263–74.
Trathan, P. N., Croxall, J. P., Murphy, E. J. & Everson, I. (1998). Use of at-sea data
to derive potential foraging ranges of Macaroni penguins during the breeding
season. Mar. Ecol. Prog. Ser., 169, 263–75.
 10
Identifying drivers of change: did fisheries play a
role in the spread of North Atlantic fulmars?
P. M. THOMPSON

Uncertainty over the role of top-down and bottom-up forces influencing

marine top-predator populations often constrains their use as indicators
of marine-ecosystem change. This chapter reviews historic and contempo-
rary data to explore how different potential drivers have shaped abundance
trends and distribution of the North Atlantic fulmar Fulmaris glacialis. Previ-
ously, debate on the causes underlying this classic example of range expan-
sion has centred on alternative hypotheses; each championing single drivers
of change. In contrast, studies now suggest that fulmar populations have
responded to multiple drivers, each with varying influence depending both
upon the population parameters being investigated, and the scale at which
these investigations are made. These findings highlight how attempts to
identify a single driver of change may be misplaced, and efforts should
instead be made to understand how different drivers interact to influence
the dynamics of these and other marine top predators.
Many marine top predators have shown dramatic changes in abun-
dance over the last century, potentially providing useful clues to the state
of the marine environment. However, in most cases, there is uncertainty
over the drivers of observed abundance changes, constraining attempts to
predict future trends or to incorporate such information into ecosystem
management. Most fundamentally, it is often unclear whether changes are
driven by top-down processes or bottom-up influences on food supplies
(e.g. Springer et al. 2003, Trites & Donnelly 2003). Furthermore, where
there is evidence that bottom-up processes exist, changes in food availability
could result from climate-driven regime shifts, or could be a by-product of

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
144 P. M. Thompson

fisheries – either directly from discards, or indirectly due to fishery-induced

changes in prey community structure.
That there should be uncertainty over the drivers of change in top-
predator abundance is not surprising. Marine mammals and seabirds are
long-lived species, living in environments where it is difficult to study the
population ecology of both predators and their prey. Estimates of marine
top-predator abundance are generally based upon counts at breeding sites,
providing only indices of population size which can be sensitive to vari-
ability in breeding success or population structure. Estimates are also often
highly variable, and it may take many years to detect population trends.
Alongside broader-scale assessments of population trends, many finer-
scale studies have investigated variability in breeding numbers and the fac-
tors influencing key population parameters at particular colonies. This work
has highlighted the role that both broad-scale ocean climate variation, and
more local measures of food availability, may have on reproduction (e.g.
Rindorf et al. 2000, Durant et al. 2003, Gjerdrum et al. 2003) and survival
(Barbraud et al. 2000, Jones et al. 2002). In general, however, most studies
have focused on variation in reproductive parameters. While reproductive
success may be highly sensitive to changes in food availability and preda-
tion, uncertainty about other key population parameters (particularly dis-
persal and juvenile survival) makes it difficult to assess longer-term popu-
lation consequences.
Therefore, to what extent can we generalize from these finer-scale stud-
ies to understand broader-scale changes in population dynamics? For exam-
ple, do the environmental variables that influence reproductive output at
particular colonies have similar effects on survival or dispersal rates? And
even where there is a good understanding of the driving forces underlying
population changes at specific colonies, are similar factors shaping colony
dynamics in other parts of the species range?

POPULATION CHANGES IN THE NORTH

ATLANTIC FULMAR

The North Atlantic fulmar (Fulmarus glacialis) provides one of the

best-known examples of an increasing marine top-predator population
(Box 10.1). Because this species was absent from mainland coasts of the
United Kingdom, the broader-scale pattern of expansion and population
increase during the twentieth century was extremely well documented.
At the same time, finer-scale studies have investigated factors influencing
key population parameters at particular colonies (Ollason & Dunnet 1978,
 Identifying drivers of change 145

Box 10.1 Population expansion of North Atlantic fulmars

The North Atlantic fulmar is currently one of the most numer-
ous seabirds in the northern hemisphere, with an estimated pop-
ulation of 5 to 7 million pairs (Mitchell et al. 2004). Arctic pop-
ulations appear to have remained relatively stable, but boreal
populations have expanded greatly over the last 400 years. In the
seventeenth century, the boreal population was believed to occur at
just two sites: St Kilda in the Outer Hebrides and Grimsey off north-
ern Iceland. The subsequent spread through Europe has been docu-
mented by Fisher (1952), and more detailed data on the expansion of
UK colonies are available through a series of national surveys orga-
nized through the British Trust for Ornithology (Fisher 1966) and
later the Seabird Group and the Joint Nature Conservation Com-
mittee (Cramp et al. 1974, Lloyd et al. 1991, Mitchell et al. 2004).
During the nineteenth and twentieth centuries, fulmars spread from
Grimsey to the coast of mainland Iceland, and colonized the Faroe
Islands between 1816 and 1839. By 1878 they were breeding on
Foula in Shetland, and they subsequently spread south through
the United Kingdom and Ireland during the twentieth century.
Norway was colonized in about 1920, and fulmars have since col-
onized France and Germany. In the western Atlantic, the spread
has not been so dramatic, but new colonies were established in
Newfoundland, Labrador and southwest Greenland (Brown, 1970,
Stenhouse & Montevecchi 1999).
The current estimate of the UK breeding population is around
538 000 pairs (Fig. 10.1). Following a rapid increase in the first half of
the twentieth century, UK population growth has slowed. Typically,
the rapid growth of the UK population has been illustrated using
data that exclude the largest population on St Kilda (Fig. 10.1b). Data
from St Kilda are sparse and uncertain, but if one assumes that num-
bers remained relatively stable at 21 000 up to 1939, as proposed by
Fisher and Waterston (1941), Fig. 10.1c provides a more represen-
tative picture of the rate of expansion of the British and Irish pop-
ulation. Notably, the period in which the rate of increase is fastest
occurs later (1925–75) when one considers population dynamics at
this larger spatial scale.
146 P. M. Thompson

(a) (b)
1800
1800–99 100000
1900–49

Number of nests
10000
1950–2000
1000

100

1875 1901 1927 1953 1979 2005

Year

(c)
100000

Number of nests
10000

1000

100

1875 1901 1927 1953 1979 2005

Year

Fig. 10.1 Changes in the distribution and abundance of North Atlantic fulmars.
(a) Schematic distribution of boreal populations, illustrating the timing of their
range expansion. (b) Changes in the number of apparently occupied nest sites at
newly colonized breeding sites in Britain and Ireland, excluding St. Kilda. Data
from Fisher and Waterston (1941), Fisher (1966), Cramp et al. (1974) and
Mitchell et al. (2004). (c) Changes in the number of apparently occupied nest
sites at all breeding sites in Britain and Ireland, including St Kilda. Data sources
as for Fig. 10.1b, with an assumed stable population of 21 000 on St Kilda
between 1875 and 1939 (see Fisher and Waterston (1941)).

Hatch 1991, Thompson & Ollason 2001). Throughout the expansion, there
has been intense debate about the underlying causes of these changes,
particularly in relation to the potential influence of commercial fisheries
(Box 10.2). As such, this provides an interesting case study where inferences
can be integrated from both broad-scale and fine-scale ecological studies of
this species. More recently, fulmar populations in European waters have
stabilized (Mitchell et al. 2004), and investigations of earlier trends may
provide useful insights into current management issues involving both ful-
mars and other marine top predators.

INVESTIGATING THE DRIVING FORCES UNDERLYING

HISTORICAL CHANGES IN ABUNDANCE AND
DISTRIBUTION

Obtaining data to understand ecological changes that may have occurred

decades earlier presents enormous challenges. Indeed, in many cases it
 Identifying drivers of change 147

Box 10.2 What caused the spread of fulmars? An historical

perspective
Recent reviews of the causes underlying the fulmars’ expansion
focus on three competing hypotheses that were prevalent during
the 1950s and 1960s. The first, and most widely cited, is Fisher’s
(1952) suggestion that fulmars responded to increases in food avail-
ability following the expansion of whaling and commercial fish-
eries. A second hypothesis, proposed by Salomonsen (1965) and
later supported by Brown (1970), argued that fulmars had instead
responded to natural changes in food availability, coinciding with a
period of warming in the temperate North Atlantic. Finally, Wynne-
Edwards (1962) suggested that the spread resulted from natural evo-
lution following a genetically or culturally based change in dispersal
behaviour. One notable feature about all three of these ideas is that
the authors provide only the most anecdotal of evidence in support of
their own hypothesis. Instead, they each focused on presenting argu-
ments against the alternative hypotheses of the day. Wynne-Edwards
(1962) highlighted that neither the timing nor the geographical pat-
tern of new breeding colonies showed any close correlation with
major developments in whaling or trawling. For example, in the early
phase of the expansion, there was no evidence of an increase around
Greenland and the Davis Strait, despite much of the eighteenth-and-
nineteenth century whaling occurring in these areas. Similarly, dur-
ing the later expansion period, there was a tendency for colonies to
form on the west coasts of the United Kingdom and Ireland rather
than on the coasts around the more heavily fished southern North
Sea. Both Wynne-Edwards (1962) and Fisher (1952) were, however,
unanimous in their dismissal of any role of climate change. In both
cases, this possibility was ruled out because a period of warming
could not have driven a population increase in a species that has a
predominantly arctic distribution (see Fig. 10.2 however).
The common feature of these hypotheses is that they all relate
the expansion to the fulmars’ exploitation of new food resources.
Yet this focus on different bottom-up processes neglects decades
of earlier debate on the potential role of top-down processes. Ful-
mars were regularly taken by island communities around Ice-
land, the Faroes and St Kilda; providing important supplies of oil,
down and meat. Between 1900 and 1940, Icelandic government
148 P. M. Thompson

statistics recorded annual catches of 20 000 to 60 000. Estimates

of Faroese catches in the 1930s were even higher at 80 000 per year,
and those from St Kilda were in the region of 6000 to 10 000 per
year until the islanders were evacuated in 1934 (Fisher 1952). Based
on such figures, several authors argued that the fulmars’ expansion
into other parts of the United Kingdom could have resulted from
a decrease in human predation. This, in turn, was believed to have
followed the introduction of regular supply ships that reduced the St
Kildan’s dependence upon their seabird harvest. Fisher and Water-
ston (1941) argued that this was an unlikely cause of the spread, par-
ticularly because they found no clear evidence that the St Kildans
had reduced their take of birds during the early phase of expan-
sion. By the time that Fisher wrote his 1952 monograph, that belief
became almost evangelical, and the ‘St Kilda theory of the fulmar’s
spread is stated, if only to be demolished’ (Fisher 1952). Detailed
data on changes in the numbers of birds killed before the 1930s
remain unavailable, although there may be potential to explore varia-
tions in harvest through more detailed analyses of estate records (see
Harman 1997). What is certain, however, is that harvests on St Kilda,
Iceland and the Faroes had all reduced dramatically by the end of
the 1930s. St Kilda was evacuated, and legislation in Iceland and the
Faroes banned the harvest of young fulmars following their identifi-
cation as a source of psittacosis infection (Fisher 1952).

may prove impossible to test hypotheses put forward to explain historical

changes in abundance. Nevertheless, there have been a number of recent
studies that have provided new insights into this long-standing debate. Fur-
thermore, they highlight the potential for other work that could help under-
stand both historical and contemporary changes in the dynamics of these
populations.

Foraging ecology and diet composition

It has been strongly argued, and is now widely believed, that the spread
of fulmars was driven largely by increases in food availability from whal-
ing and commercial fisheries (Fisher 1952). At the time, little was known
about fulmar diet and feeding ecology but, over the last decade, there has
been more research in this area. Fulmars often scavenge around fishing
boats, but wider-scale studies indicate that their distribution is more closely
related to hydrographic features than to fisheries (Camphuysen & Garthe
 Identifying drivers of change 149

1997, Skov & Durinck 2000). Similarly, comparisons of diet across their
range suggest that fishery-derived offal and discards can form an important
part of the diet in some areas, but that birds at many other colonies tend to
forage on pelagic crustaceans and small fish (Furness & Todd 1984, Hamer
et al. 1997, Phillips et al. 1999). Nevertheless, while these prey appear to be
taken directly, their availability may also have increased indirectly as a result
of fishing pressure (Pauly et al. 1998). More generally, this work has shown
that fulmars are extremely catholic in their diet, and that diet composition
can differ markedly between years at a single site (Phillips et al. 1999). Such
inter-annual variation makes assessments of longer-term trends difficult to
evaluate, and further highlights that seasonal variation in diet is likely. The
predominance of breeding-season studies may therefore bias our under-
standing of the overall diet of these birds.
The development of bio-energetic models can also be used to assess
whether current levels of discarding could support the energetic require-
ments of different seabird populations. Even in heavily fished areas such as
the North Sea, fewer than 50% of fulmars could be fully supported by fishery
waste (Camphuysen & Garthe 1997). Together, these studies suggest that
fulmars are not, at least currently, heavily dependent upon fishery waste.
Nevertheless, the availability of discards may be important to these birds at
times when, or in areas where, natural prey are more limited (Mitchell et al.
2004); probably during the winter (when natural prey are less available),
and during the early chick-rearing period (when adults are constrained to
shorter foraging trips).

Individual based studies of reproductive success and survival

Long-term individual based studies have underpinned many finer-scale

studies of marine top-predator population ecology (Wooller et al. 1992).
One of the longest running of these studies has been of a colour-ringed
population of fulmars at a small colony on Eynhallow, Orkney (Dunnet
1991). Since 1950, studies have described the continued increase of this
population, and provided estimates of reproductive and survival rates. More
recently, these data were used to explore the relative influence of large-scale
climate variation and local measures of fisheries activity on this population.
Inter-annual variability in reproductive success and the proportion of each
cohort that recruited back to the natal colony were both related to indices
of climate variation, but not to available data on fisheries (Thompson &
Ollason 2001). However, while reproductive success was most closely
related to variations in the winter North Atlantic Oscillation (wNAO),
150 P. M. Thompson

(a) (c) 9
8
7

% recruiting
6
5
4
3
2
1
0
0.2 0.1 0.0 0.1
Temperature anomaly (°C)
(b) (d)
60 9
8
Reproductive success (%)

50 7

% recruiting
6
40 5
4
30
3
2
20
1
10 0
6 4 2 0 2 4 6 10 20 30 40 50 60
Winter North Atlantic Oscillation Annual reproductive success

Fig. 10.2 Studies of individually marked fulmars on Eynhallow (a) have shown
that reproductive success and recruitment are influenced by climate variation. (b)
Reproductive success is negatively related to the winter North Atlantic Oscillation
index. (c) Cohort recruitment rates are positively related to northern-hemisphere
temperature anomalies. (d) Cohorts experiencing higher reproductive success do
not necessarily exhibit higher levels of recruitment Annual breeding success is
calculated as the percentage of eggs laid that produce successful fledglings.
Panels b and c are redrawn from Thompson and Ollason (2001).

recruitment was strongly and negatively related to northern-hemisphere

temperature (NHT) anomalies. Thus, contrary to earlier assumptions (see
Box 10.2), these data suggest that warmer conditions may improve some
measures of productivity, even in this arctic species. More generally, the dif-
ferential effects of these two variables meant that annual estimates of repro-
ductive success and cohort recruitment rates did not co-vary (Fig. 10.2).
Causal links between these large-scale climate variables, and reproduction
and recruitment, remain unclear; however, the wNAO and NHT seem likely
either to be influencing different prey stocks or to have an impact on birds at
different times of year. Fulmars do not recruit until they are at least 7 years
old, and these bottom-up effects on reproduction and recruitment there-
fore had a lagged effect on short-term variability in de-trended estimates
of colony size (Thompson & Ollason 2001). Together, these data highlight
that natural variations in ocean climate have influenced the local dynam-
ics of this colony, but this does not necessarily mean that other factors
 Identifying drivers of change 151

are not involved. Indeed, these analyses show that there remains a strong,
unexplained, linear increase in colony size, which earlier work suggests
must have been driven by immigration from other areas (Ollason & Dunnet
1983).

Use of proxies to explore historical ecological change

One of the key constraints when interpreting recent data on diet composi-
tion is that these patterns may not be representative of earlier critical periods
during the fulmar’s expansion. Even the retrospective analysis of long-term
individual-based data is restricted to the later phases of the expansion, and
there is limited information on relevant environmental covariates. Attempts
to understand historical patterns of ecological changes have therefore often
drawn upon a wide variety of indirect proxies of abundance, diet or environ-
mental changes that may have driven such changes. In other systems, these
have included economic records of catches (Allen & Keay 2001), analyses
of hair of seals and fish scales in seabed sediments (Hodgson et al. 1998,
O’Connell & Tunnicliffe 2001) and an increasing array of molecular and
biochemical techniques for understanding variation in abundance (Roman
& Palumbi, 2003), dispersal and feeding ecology (Hobson 1993, 1999,
Smith et al. 1997). Such studies provide useful new insights into the nature
of the fulmar’s spread, and highlight the potential for similar approaches to
extend these findings in the future.
The first of these insights involve paleoecological studies, where
Montevecchi and Hufthammer (1990) described the distribution of ful-
mar bones from archaeological sites in Norway. They found evidence
of fulmars at 26 sites, extending through northern and southern Nor-
way. Most recovered bones were dated at between 1000 and 4000 years
before present, with a peak in the period 1000 to 2000 years ago. The
recent colonization of Norway in 1920 therefore appears to have been
a re-colonization, and clearly shows that recent changes in distribution
did not necessarily depend upon human-induced changes in food sup-
plies. Records of fulmars at Scottish archaeological sites are also scattered
through the literature. Fisher and Waterston (1941) briefly mention, but
then ignore, the fact that fulmar bones were found in a midden from
the west of Scotland; more recent studies have recorded fulmars from
excavations dating from the Neolithic to the early medieval (Serjeantson
1988). Further review of archaeological data from other parts of the ful-
mar’s contemporary range, together with carbon dating of the Scottish spec-
imens, would provide valuable insights into their historical distribution.
152 P. M. Thompson

In addition, stable-isotope analyses of such specimens can also provide

information on changes in diet composition. Using more recent museum
samples, Thompson et al. (1995) found reductions in both δ 15 N and
δ 13 C values in contemporary feathers compared with those from skins col-
lected between 1850 and 1950. The difference in δ 15 N was modest, repre-
senting a reduction of 25% to 33% of a trophic level, but is consistent with a
decreasing tendency for fulmars to feed upon offal from toothed whales or
large fish. However, contemporary samples were taken in a single year and
may not fully reflect the entire breadth of contemporary diet, suggesting
that further work on more recent, and older, samples would be worthwhile.
The second of these insights comes from recent molecular analyses
investigating the relationship between fulmars from five more recently
established colonies, and the possible source colonies of St Kilda and Ice-
land (Burg et al. 2003). Many early writers had argued in favour of one or
other of these sites as the most likely source colony (Gordon 1936, Fisher
1952), but mitochondrial DNA analyses provided evidence that these birds
had recruited from both St Kilda and Iceland. Furthermore, there were high
levels of haplotype diversity at all sites, suggesting that founding events
tended to involve many unrelated birds.

DID FISHERIES PLAY A ROLE IN THE SPREAD OF

NORTH ATLANTIC FULMARS?

While fisheries waste has proved an important source of food for fulmars
in certain areas, it is clearly not the sole cause underlying the population’s
expansion. Instead it seems much more likely that the population was
responding to multiple driving forces, leaving us with the challenge of deter-
mining the relative importance of different drivers. While this is beyond the
scope of this chapter, examination of the process by which explanations for
the spread were developed and interpreted provides some general lessons
for evaluating this and similar issues in the future.
One clear feature of criticisms of alternative hypotheses was that they
were often constrained by the limited information available on seabird life-
history patterns at that time. Perhaps the most obvious shortcoming is
that complex descriptions of the spatial pattern of spread were based on
the assumption that successful breeders produced young that recruited to
nearby colonies at just 1 or 2 years old (Fisher 1952). Similarly, lack of knowl-
edge about age at maturity and longevity meant that the sensitivity of popu-
lation trends to variations in reproduction and adult survival were not fully
appreciated (Fisher & Waterston 1941). There remains uncertainty over the
detailed ecology of many species involved in current fisheries interactions,
 Identifying drivers of change 153

and we should be careful not to dismiss hypotheses simply because they lie
outside our current understanding of the system.
Secondly, issues of scale heavily influenced perceptions of the impor-
tance of fisheries. At a fine scale, large feeding flocks of fulmars around
whale carcasses and trawlers appeared to provide convincing support for
the importance of these artificial prey supplies. Only since larger-scale stud-
ies have been conducted has it become clear that attraction to vessels is a
relatively local process (Skov & Durinck 2000). Indeed, our perception of
the rapid rise in abundance is itself biased because most studies have been
carried out at a local scale in newly colonized parts of the United King-
dom. Broader-scale assessments of population increases are less dramatic
(see Box 10.1), and may not be so very different from those observed in
other seabirds in the region (Mitchell et al. 2004). Another aspect of scale
that links to our understanding of life history, is that driving forces acting
upon reproduction or early survival will have lagged effects on measures of
population abundance. Specifically, given the low power of many marine
monitoring programmes, this means that we may need to look well back
into the past to identify drivers of recently detected changes.
The lack of opportunities to test alternative hypotheses directly appears
to have encouraged the champions of different hypotheses to become
increasingly entrenched in their opinions (Box 10.2). The dismissal of some
hypotheses now seems premature, as several arguments used against these
ideas are much less convincing in the light of current ecological understand-
ing. Fisher and Waterston’s (1941) assessment of the impact of hunting on
populations in both Iceland and St Kilda was based on the assumption that
only young birds were taken. However, other sources indicate that harvests
were of both adults and young, and there would also be additional losses
from egg harvesting. When Martin Martin visited St Kilda in 1697, he esti-
mated that the people of St Kilda had given their party 16 000 seabird eggs
during their stay; while his description of the delicate taste of the adult ful-
mars highlights the existence of a mixed harvest (cited in Gordon (1936)).
Given the sensitivity of populations of long-lived vertebrates to changes in
adult mortality, slight changes in the ratio of adults to young in reported har-
vest levels (for example in response to a known decline in the market for the
oil and down during the late nineteenth century (Harman 1997)) could have
had important impacts on population growth rates. Alternative hypotheses
clearly need to be kept under review; particularly where there may be oppor-
tunities to develop new techniques to explore some of these old questions.
Even if changes in top-down processes did not influence the early stages
of the expansion, reductions in hunting after the 1930s must have con-
tributed to the faster increases during the next few decades (see Box 10.1).
154 P. M. Thompson

With an estimated 20 000 birds per year currently caught by Norwegian

long-liners (Dunn & Steel 2001), it would be timely to assess the role that
this may play in the current levelling of European populations. Several dif-
ferent factors are also likely to have contributed to bottom-up influences
on food supply. Whaling and fisheries may have played a part, directly or
indirectly, but there have been other important responses of plankton com-
munities to ocean-climate variation (Beaugrand et al. 2002). In some cases,
there may be co-variation in responses of different population parameters
to each of these factors, but there is also evidence that different parameters
may be responding to different driving forces (Fig. 10.2). Modelling frame-
works exist to explore such issues (e.g. Jenouvrier et al. 2003), but there
remain important challenges when parameterizing models. In particular,
few detailed studies have estimated reproduction and survival rates, and
caution is required when scaling up from local studies to explore impacts
across the population’s range. Similarly, data on dispersal rates between
different colonies remain elusive, but are crucial to any attempts to model
changes beyond the colony scale. If these problems can be overcome, these
tools provide exciting, but challenging, opportunities for modelling how
these different factors could have influenced historical population trends
and how they may in turn influence future population levels.

REFERENCES
Allen, R. C. & Keay, I. (2001). The first great whale extinction: the end of the
bowhead whale in the eastern arctic. Explor. Econ. Hist., 38, 448–77.
Barbraud, C., Weimerskirch, H., Guinet, C. & Jouventin, P. (2000). Effect of
sea-ice extent on adult survival of an Antarctic top predator: the snow petrel
Pagodroma nivea. Oecologia, 125, 483–8.
Beaugrand, G., Reid, P. C., Ibanez, F., Lindley, J. A. & Edwards, M. (2002).
Reorganization of North Atlantic marine copepod biodiversity and climate.
Science, 296, 1692–4.
Brown, R. G. B. (1970). Fulmar distribution: a Canadian perspective. Ibis, 112,
44–51.
Burg, T. M., Lomax, J., Almond, R., Brooke, M. D. & Amos, W. (2003). Unravelling
dispersal patterns in an expanding population of a highly mobile seabird, the
northern fulmar (Fulmarus glacialis). Proc. R. Soc. Lond. B Biol. Sci., 270,
979–84.
Camphuysen, C. J. & Garthe, S. (1997). An evaluation of the distribution and
scavenging habits of northern fulmars (Fulmarus glacialis) in the North Sea.
ICES J. Mar. Sci., 54, 654–83.
Cramp, S., Bourne, W. & Saunders, D. (1974). The seabirds of Britain and Ireland.
London: Collins.
Dunn, E. & Steel, C. (2001). The Impact of Longline Fishing on Seabirds in the
North-east Atlantic: Recommendations for Reducing Mortality. Report of the
RSPB/NOF/JNCC/BirdLife International, Report No. 5. Sandy, UK: RSPB.
 Identifying drivers of change 155

Dunnet, G. M. (1991). Population studies of the Fulmar on Eynhallow, Orkney

Islands. Ibis, 133, 24–7.
Durant, J. M., Anker-Nilssen, T. & Stenseth, N. C. (2003). Trophic interactions
under climate fluctuations: the Atlantic puffin as an example. Proc. R. Soc.
Lond. B Biol. Sci., 270, 1461–6.
Fisher, J. (1952). The Fulmar. London: Collins.
(1966). The fulmar population of Britain and Ireland, 1959. Bird Study, 13,
334–54.
Fisher, J. & Waterston, G. (1941). The breeding distribution, history and population
of the Fulmar (Fulmarus glacialis) in the British Isles. J. Anim. Ecol., 10,
204–72.
Furness, R. W. & Todd, C. M. (1984). Diets and feeding of Fulmars, Fulmarus
glacialis, during the breeding-season: a comparison between St Kilda and
Shetland colonies. Ibis, 126, 379–87.
Gjerdrum, C., Vallee, A. M. J., St Clair, C. C. et al. (2003). Tufted puffin
reproduction reveals ocean climate variability. Proc. Nat. Acad. Sci. U. S. A.,
100, 9377–82.
Gordon, S. (1936). The fulmar petrel. Nature, 137, 173–6.
Hamer, K. C., Thompson, D. R. & Gray, C. M. (1997). Spatial variation in the
feeding ecology, foraging ranges, and breeding energetics of northern fulmars
in the north-east Atlantic Ocean. ICES J. Mar. Sci., 54, 645–53.
Harman, M. (1997). An Isle Called Hirte: History and Culture of the St Kildans to
1930. Isle of Skye, UK: Maclean Press.
Hatch, S. A. (1991). Evidence for color phase effects on the breeding and
life-history of Northern Fulmars. Condor, 93, 409–17.
Hobson, K. A. (1993). Trophic relationships among High Arctic seabirds: insights
from tissue-dependent stable-isotope models. Mar. Ecol. Prog. Ser., 95,
7–18.
(1999). Tracing origins and migration of wildlife using stable isotopes: a review.
Oecologia, 120, 314–26.
Hodgson, D. A., Johnston, N. M., Caulkett, A. P. & Jones, V. J. (1998).
Palaeolimnology of Antarctic fur seal Arctocephalus gazella populations and
implications for Antarctic management. Biol. Conserv., 83, 145–54.
Jenouvrier, S., Barbraud, C. & Weimerskirch, H. (2003). Effects of climate
variability on the temporal population dynamics of southern fulmars. J. Anim.
Ecol., 72, 576–87.
Jones, I. L., Hunter, F. M. & Robertson, G. J. (2002). Annual adult survival of Least
Auklets (Aves, Alcidae) varies with large-scale climatic conditions of the North
Pacific Ocean. Oecologia, 133, 38–44.
Lloyd, C., Tasker, M. L. & Partridge, K. (1991). The Status of Seabirds in Britain and
Ireland. London: T. & A. D. Poyser.
Mitchell, I., Newton, S., Ratcliffe, N. & Dunn, T. (2004). Seabird Populations of
Britain and Ireland. London: T. & A. D. Poyser.
Montevecchi, W. A. & Hufthammer, A. K. (1990). Zooarchaeological implications
for prehistoric distributions of seabirds along the Norwegian coast. Arctic, 43,
110–4.
O’Connell, J. M. & Tunnicliffe, V. (2001). The use of sedimentary fish remains for
interpretation of long-term fish population fluctuations. Mar. Geol., 174,
177–95.
156 P. M. Thompson

Ollason, J. & Dunnet, G. (1978). Age, experience and other factors affecting the
breeding success of the Fulmar, Fulmarus glacialis in Orkney. J. Anim. Ecol., 47,
961–76.
(1983). Modelling annual changes in numbers of breeding Fulmars, Fulmarus
glacialis, at a colony in Orkney. J. Anim. Ecol., 52, 185–98.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, F. (1998). Fishing
down marine food webs. Science, 279, 860–3.
Phillips, R. A., Petersen, M. K., Lilliendahl, K. et al. (1999). Diet of the northern
fulmar Fulmarus glacialis: reliance on commercial fisheries? Mar. Biol., 135,
159–70.
Rindorf, A., Wanless, S. & Harris, M. P. (2000). Effects of changes in sandeel
availability on the reproductive output of seabirds. Mar. Ecol. Prog. Ser., 202,
241–52.
Roman, J. & Palumbi, S. R. (2003). Whales before whaling in the North Atlantic.
Science, 301, 508–10.
Salomonsen, F. (1965). The geographical variation of the fulmar (Fulmarus glacialis)
and the zones of marine environment in the North Atlantic. Auk, 82, 327–55.
Serjeantson, D. (1988). Archaeological and ethnographic evidence for seabird
exploitation in Scotland. Archaeozoologia, 11, 209–24.
Skov, H. & Durinck, J. (2000). Seabird distribution in relation to hydrography in
the Skagerrak. Cont. Shelf Res., 20, 169–87.
Smith, S. J., Iverson, S. J. & Bowen, W. D. (1997). Fatty acid signatures and
classification trees: new tools for investigating the foraging ecology of seals.
Can. J. Fish. Aquat. Sci., 54, 1377–86.
Springer, A. M., Estes, J. A., Vliet, G. B. van et al. (2003). Sequential megafaunal
collapse in the North Pacific Ocean: an ongoing legacy of industrial whaling?
Proc. Natl. Acad. Sci. U. S. A., 100, 12 223–8.
Stenhouse, I. J. & Montevecchi, W. A. (1999). Increasing and expanding
populations of breeding Northern Fulmars in Atlantic Canada. Waterbirds, 22,
382–91.
Thompson, D. R., Furness, R. W. & Lewis, S. A. (1995). Diets and long-term
changes in delta-N-15 and delta-C-13 values in Northern Fulmars Fulmarus
glacialis from two Northeast Atlantic colonies. Mar. Ecol. Prog. Ser., 125, 3–11.
Thompson, P. M. & Ollason, J. C. (2001). Lagged effects of ocean-climate change
on fulmar population dynamics. Nature, 413, 417–20.
Trites, A. W. & Donnelly, C. P. (2003). The decline of Steller sea lions Eumetopias
jubatus in Alaska: a review of the nutritional stress hypothesis. Mamm. Rev., 33,
3–28.
Wooller, R. D., Bradley, J. S. & Croxall, J. P. (1992). Long-term population studies
of seabirds. Trends Ecol. Evol., 7, 111–14.
Wynne-Edwards, V. (1962). Animal Dispersion in Relation to Social Behaviour.
Edinburgh, UK: Oliver and Boyd.
 11
Monitoring predator–prey interactions using
multiple predator species: the
South Georgia experience
J . P . C R O XA L L

The inception (in 1976) and development of an annual programme mon-

itoring selected variables to characterize diet, foraging and breeding per-
formance of key krill-dependent top predators (Antarctic fur seal, gentoo
and macaroni penguin, and black-browed albatross) at Bird Island, South
Georgia is described. Criteria for choice of species and variables (the latter
covering the range of spatio-temporal scales of predator–prey interactions)
are provided, together with the current approaches to combining indices to
improve characterization of key relationships with prey availability. The suc-
cesses of the programme, particularly in relation to understanding predator
responses to changes in prey availability, are summarized, together with its
limitations – notably in respect of explaining or predicting changes in pop-
ulation size. The main challenges for the future include understanding the
predator–prey interactions within the full environmental context, linking
appropriately characterized functional relationships to population models
and incorporating predator data more effectively into the management of
the krill fishery.
Thirty years ago, in 1975, when the British Antarctic Survey (BAS) was
planning a new programme of research on marine vertebrates (seals and
seabirds) at Bird Island, South Georgia, long-term population studies (espe-
cially of marked individuals) were recognized as a vital tool for ecological
investigations. Long-term studies of population activities and responses to
ecological conditions, however, were often referred to as monitoring and

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
158 J. P. Croxall

were not regarded as serious science. Nevertheless, in parallel with starting

demographic studies on albatrosses, penguins, fur seals and elephant seals
at South Georgia, a programme to collect annually adjunct data on diet, for-
aging ecology, adult condition, offspring growth and breeding success was
started (Croxall & Prince 1979). This reflected the growing concern that the
serial over-exploitation of fur seals, baleen whales and certain fish stocks
in the Southern Ocean was about to be followed by unregulated harvest-
ing of krill – and that data would be needed to demonstrate how critical
this resource was as a keystone prey species for a wide range of vertebrate
predators.
These same concerns were shared internationally and ultimately led to
the development and implementation of the Convention on the Conserva-
tion of Antarctic Marine Living Resources (CCAMLR) which came into force
in 1982. Its fundamental principles (contained in Article II of the Conven-
tion) emphasized the need for:
(a) sustainable use of harvested resources;
(b) taking full account of the needs of dependent species;
(c) prevention of changes irreversible within 20 to 30 years.

These principles (in effect pioneering the potential practical use of the pre-
cautionary principle) and the explicit need to balance resource exploita-
tion and conservation within a single management regime, led to the first
attempts for marine systems explicitly to try to develop approaches to man-
agement on an ecosystem basis. This was developed through two main
approaches: creating an appropriate conceptual framework (e.g. as illus-
trated in CCAMLR (1995) and Fig. 11.3 later in this chapter) and establishing
a CCAMLR Ecosystem Monitoring Programme (CEMP). The CEMP was
designed mainly to:
(a) detect significant changes in critical components of the ecosystem;
(b) distinguish between changes due to harvesting and changes due to
environmental variability.
By 1985, when CEMP planning started, the principles and practices
established in the BAS monitoring and population programmes at Bird
Island had been expanded and endorsed by the Scientific Committee for
Antarctic Research (SCAR 1979) as an appropriate model for an interna-
tional network and scheme. They also formed the foundation for the devel-
opment of the CEMP (CCAMLR 1985, 1986, Croxall et al. 1988, Agnew
1997, Croxall & Nicol 2004) which started in 1987. The focus of this moni-
toring programme was on species dependent on krill; criteria for the selec-
tion of monitored species and variables are summarized in Box 11.1.
Box 11.1 Choice of monitoring species and variables
The criteria used to select species for the predator-based monitoring
at Bird Island (and CEMP) were:
r significance (important consumer of krill);
r specialist (krill consistently forming the main element of the
diet);
r widely distributed (a range of additional study sites potentially
available);
r feasibility (readily accessible breeding sites and tolerant of
human activity).
The resulting species chosen at Bird Island were gentoo and mac-
aroni penguins, black-browed albatross and Antarctic fur seal. (Addi-
tional species for CEMP were Adélie and chinstrap penguins, Cape
and Antarctic petrels and crabeater seal).
The variables selected were designed to cover a range of spatial
and temporal scales (Fig. 11.1). Inevitably the majority of these reflect
performance and conditions during the austral summer breeding
season, whereas only arrival mass (and breeding-population size, at
least in part) reflect conditions in winter (Fig. 11.1a). The variables
were also selected on the basis of ease and accuracy of measurement
(given that detecting change over relatively short periods requires
repeated acquisition of large samples of data). A handbook provid-
ing standard methods for measuring each variable was produced for
CEMP (CCAMLR 1987) and updated annually.
Most variables integrate processes and conditions operating on
scales of months (i.e. the duration of offspring-rearing events);
although the constituent elements of some of these variables (e.g.
foraging-trip duration, diet) are collected, and can reveal effects at,
smaller scales. Population variables represent variation at scales of
at least one year, with population size – which integrates annual and
multi-year effects – being particularly complex to interpret.
The spatial scales at which these variables integrate prey
and environmental conditions are particularly diverse (Fig. 11.1b),
ranging from 10 to 100 km in gentoo penguin to over 10 000 km in
black-browed albatross. Knowledge and awareness of the relation-
ship between temporal and spatial scales is often crucial to interpret-
ing monitoring data in terms of influence of prey and environment.
Monitoring and linking these at congruent scales to the particular
predator variables is obviously essential. The advent of satellite track-
ing has revolutionized our ability to understand the nature of forag-
ing range (and key feeding areas within these ranges) of predators.
160 J. P. Croxall

(a) Sep Oct Nov Dec Jan Feb Mar Apr

Arrive Hatch Crèche Fledge

Eggs

Breeding population
Chicks fledging
Adult arrival mass Foraging-trip duration
Growth rate
Incubation shift duration
Adult departure mass
Breeding success Chick fledging mass

(b)
Multi-year

n
tio
p ula
Po size
Year

ult l
Ad viva
r
su
Time
Months

g
din
r ee cess
B uc
s
g
rin g
fsp ionin
Weeks

f
O is
v
ing r pro
o rag viou
F ha
be

Black-browed
albatross 10 100 1000 10 000
Gentoo
penguin 1 10 100

Space (km)
Fig. 11.1 Examples of temporal and spatial scales of relevance to variables
selected for monitoring by the Bird Island and CCAMLR Ecosystem Monitoring
Programmes. (a) Timing and duration of various parameters for macaroni
penguin. (b) Temporal and spatial scales integrated by main categories of
monitoring variables for the two most extreme species at South Georgia:
black-browed albatross (upper spatial scale) and gentoo penguin (lower spatial
scale). After Reid et al. (2005).
 Monitoring predator–prey interactions 161

The aims of this chapter are to examine some of the achievements of

the monitoring programmes at Bird Island, to identify some of the main
problems and challenges that have emerged and to indicate some promising
or desirable initiatives for the future. I shall focus on predator–prey (krill)
interactions because Trathan et al. (Chapter 3 in this volume) have reviewed
the environmental context and influences, and Reid et al. (Chapter 17 in
this volume) will summarize the substantial contribution that the predator
monitoring programme came to make to the monitoring of krill population
structure and dynamics.

ACHIEVEMENTS, LIMITATIONS AND CHALLENGES

Long-term programmes using seabirds and land-based marine mammals

as indicators of conditions in marine systems are now relatively widespread
(e.g. Harris & Wanless 1990, Bost & Le Maho 1993, Ainley et al. 1995,
Monaghan 1996, Furness & Camphuysen 1997, Regehr & Montevecchi
1997, Weimerskirch et al. 2003, Montevecchi et al. (Chapter 8 in this
volume)). They share most, if not all, of the following aims:

(1) To detect changes in indices of the status (in either demographic or

physiological – e.g. condition – respects) and/or reproductive
performance of seabirds and seals.
(2) To relate these changes to indices of prey abundance and availability
(to the predators).
(3) To use predator indices, on the basis of relationships between
predators and prey developed above, as a measure of: (a) prey
availability (to the predators) and (b) prey stock abundance.
(4) To use the predator indices to detect changes in food availability that
result from commercial harvesting as distinct from changes due to
natural fluctuations in the biological and physical environment.

Detection of change

As time elapsed, the monitoring programmes at Bird Island provided

numerous examples of significant changes (but few systematic trends) in
index values (Croxall & Prince 1987, Croxall et al. 1988, Williams & Rothery
1990, Lunn et al. 1993, 1994, Boyd et al. 1994, Croxall & Rothery 1995). Only
a few of these changes related to body condition (i.e. arrival mass, adult
mass, fledging/weaning mass), whereas most reflected changes in one
or more aspects of reproductive performance (e.g. foraging-trip duration,
162 J. P. Croxall

4 Years of low krill availability

3
Reproductive output index
2

−1

−2 Antarctic fur seal

Macaroni penguin
−3 Gentoo penguin
Black-browed albatross
−4
1985 1990 1995 2000

Year
Fig. 11.2 Fluctuations in breeding performance – as measured by a reproductive
output index (ROI) – for Antarctic fur seals, gentoo penguins, macaroni
penguins and black-browed albatrosses breeding at Bird Island, South Georgia.
Details of variables comprising the ROI are given in Reid & Croxall (2001), from
which this figure is taken.

provisioning rate, breeding success). Most of the changes related to clear

differences in performance in years of differing availability of krill. Such
events were particularly clear for all the monitored species in a number of
years, particularly once the monitoring programme at Bird Island was fully
developed (in the early 1980s) (Fig. 11.2). When all performance variables
were combined, there was a statistically significant negative trend in preda-
tor performance through time, suggesting that there may have been less
krill available to predators in the 1990s compared with the 1980s (Reid &
Croxall 2001, Reid et al. 2005).

Relating predator data to changes in prey

Estimates of prey (krill) abundance in the vicinity of South Georgia have

only become available annually, from acoustic surveys (Brierley et al. 1999),
since 1990 (except 1995), although extensive surveys did take place in 1981,
1982 and particularly in 1986 (Reid et al. 2005). Therefore rigorous investi-
gation of functional predator–prey relationships could not commence until
the late 1990s. Consequently in the early years of the monitoring pro-
gramme the most convincing evidence of relationships between predator
 Monitoring predator–prey interactions 163

Box 11.2 Comparisons of predator performance in good and

bad years
Insights into the responses of top predators to changes in prey
availability are often achieved as a result of ‘natural experiments’
when relevant data are available for years of very different levels of
prey abundance. At South Georgia such a comparison was possible
between 1986 and 1994, years when krill abundance at sea (esti-
mated using standard acoustic survey techniques) and a wide variety
of predator response variables (from two penguin and two albatross
species, with supporting data from Antarctic fur seals) were mea-
sured independently and simultaneously (Croxall et al. 1999).
The four-fold difference in krill biomass in 1986 (c. 30 g m−2 )
compared with 1994 (c. 7 g m−2 ) was accompanied by: (a) an 88%
to 90% reduction in the mass of krill in predator diets (and some
increase in the fish component) (Table 11.1); (b) greater prey diver-
sity for most species; (c) reduced diet overlap between species; (d) a
switch from krill to amphipods in macaroni penguin but no equiv-
alent major dietary change in other species; (e) a major reduction
in the length–frequency (and length–biomass) composition of krill
eaten (see Fig. 11.3). Rates of provisioning of offspring decreased by
90% in gentoo penguin and 40% to 50% in the other three species;
this was due to reduced meal size in penguins (by 90% in gentoo
and 50% in macaroni penguins) and to doubling of foraging-trip
duration in albatrosses (Table 11.1). Breeding success was reduced by
50% in grey-headed albatross (the species least dependent on krill),
by 90% in black-browed albatross and gentoo penguin (only 3% to
4% of eggs producing fledged chicks), but by only 10% in maca-
roni penguin – presumably reflecting its ability to switch to small
prey unprofitable for the other species. However, all species (except
for black-browed albatross) – and particularly macaroni penguin –
produced fledglings significantly lighter than usual, which proba-
bly affected their subsequent survival. Some effects on adult survival
could also be inferred. Our results show a coherent, although com-
plex, pattern of within- and between-species similarities and differ-
ences. These mainly reflect the degree of dependence on krill, the
feasibility of taking alternative prey and constraints on trip duration
and/or meal size imposed by foraging adaptations (especially relat-
ing to travel speeds and diving abilities, whereby flightless divers and
pelagic foragers differ markedly).
164 J. P. Croxall

Table 11.1. Indices of diet content, offspring provisioning and overall

reproductive performance in a year of low krill availability (1994) as a
percentage of values for 1986, a year of average krill availability

Krill Meal Meal Provisioning Breeding

Species in diet size frequency rate success

Grey-headed albatross 9 80 53 41 45
Black-browed albatross 12 96 53 47 9
Gentoo penguin 13 9 78 9 10
Macaroni penguin 14 62 93 50 92

performance and krill availability came from data in single years of widely
different estimates of krill abundance (Box 11.2). The results presented
in Box 11.2 indicate that even with sufficiently major changes in prey
abundance, such that all dependent species show statistically significant
responses in terms of provisioning and productivity:

(a) few responses are proportional (i.e. show a linear relationship) to

changes in prey abundance;
(b) some species are more sensitive than others;
(c) differences between species mainly reflect the degree of dependence
on krill and provisioning constraints imposed by lifestyle and
physiological adaptation.

Prey abundance–availability relationships

In some cases (e.g. Fig. 11.4) there was also evidence of strong relation-
ships between krill abundance from acoustic surveys and one or more of
the diet indices of krill availability to predators. The ability to generate valu-
able data on the population structure of prey, particularly krill, from preda-
tor sampling became a particular achievement of the programme (e.g. Reid
et al. 1996, 1999a, 1999b) and is reviewed by Reid et al. (Chapter 17 in
this volume). Although these data are unlikely to provide measures of prey
stock abundance, they are likely to be increasingly valuable as proxies for
understanding the nature of krill availability to predators and potentially
contributing to predictions for future years (Reid et al. 1999a).
Now that over a decade of annual prey-abundance estimates are avail-
able for South Georgia, more progress is being made in defining and
understanding the functional relationships involved. This is greatly assisted
1986 0.3 0.3 0.3
n = 2673 n = 3215 n = 324
0.2 0.2 0.2

0.1 0.1 0.1

Proportion
Proportion
Proportion
0 0 0

1994 0.3 0.3 0.3

n = 35 n = 308
n = 175
0.2 0.2 0.2

0.1 0.1 0.1

Proportion
Proportion
Proportion

0 0 0
26 30 34 38 42 46 50 54 58 62 66 70 26 30 34 38 42 46 50 54 58 62 66 70 26 30 34 38 42 46 50 54 58 62 66 70
Krill length (mm) Krill length (mm) Krill length (mm)

Macaroni penguin Gentoo penguin Antarctic fur seal

Fig. 11.3 Differences in length–frequency distribution of Antarctic krill taken by predators in years of average (1986) and low (1994) availability
of krill. From Croxall et al. (1999).
166 J. P. Croxall

1996

Percentage krill in macaroni penguin diet

100
1997

80

1998
60

40
1999

20
1994 2000

0
10 15 20 25
Krill density (g m−2)

Fig. 11.4 Relationship between proportion (% wet mass) of krill in the diet of
macaroni penguins from Bird Island and estimated krill density (g m−2 ) at the
nearby north-western end of South Georgia between 1994 and 2000 (r2 = 0.96,
F1,5 = 92.9, p = 0.0006). From Barlow et al. (2002).

by recent developments that have facilitated the combining of monitoring

indices in a way that simplifies interpretation (Box 11.3).

Distinguishing changes due to harvesting and those due to

environmental variability

None of the changes in predator performance variables can be attributed to

effects of commercial fishing for krill. Most changes are believed to reflect
some combination of local and regional changes in the physical and bio-
logical environment (Trathan et al. (Chapter 3 in this volume)). A particular
difficulty in evaluating or differentiating the relative roles of commercial
fishing and of environmental influences at South Georgia is that the fish-
ery there occurs chiefly in winter, the time covered by fewest of the moni-
tored predator variables. This contrasts with the situation in the Antarctic
Peninsula sector where the fishery operates in summer and usually close
to sites where predator monitoring takes place, albeit monitoring that has
been initiated somewhat more recently (Hewitt et al. 2003).
It is unlikely that much progress will be made in detecting effects of krill
fishing on predators unless: (a) the fishery harvest increases substantially;
(b) the environmental effects on predators can be attributed more precisely;
or (c) the magnitude and timing of krill flux into the harvesting and moni-
toring areas can be estimated accurately.
 Monitoring predator–prey interactions 167

Box 11.3 New insights for combining indices from monitoring

data
Monitoring programmes tend to produce substantial amounts of
data on a diverse range of measures which potentially reflect aspects
of the status of species, processes or systems under study. Effective
interpretation and use can be greatly aided by simplification without
loss of information.
Using a data matrix of y years and a response indices, de la
Mare and Constable (2000) provided an effective way of combin-
ing indices by a process of transformation (to obtain standard nor-
mal distributions) and standardization (in respect of sign (positive
to indicate better conditions) and standard deviation (SD)). Each
response variable is standardized to a mean = 0 and SD = 1, pro-
ducing a sum I for each year y such that Iy = a  x y , where x is a vector
of values for all response variables in year y and a is an identity vector
of the same dimensions as x that takes a value of 1 for those vectors
where observations exist and 0 for missing data. The variance V of Iy
is given by Vy = a Sa  , where S is the covariance matrix of the stan-
dardized response vectors; hence the combined standardized index
(CSI) in year y was
√
CSI y = Iy / V y
De la Mare and Constable (2002) also address sensitivity to missing
values by utilizing a procedure for smoothing correlation matrices.
This approach was further developed, using data from the Bird
Island monitoring programme, by Boyd & Murray (2001) and
Reid et al. (2005), the latter also assessing the difference between
using CSIs and individual response variables (vectors). This showed
rather consistent patterns in the coefficients of variation of predator
response (Fig. 11.5), with vectors relating to body condition showing
least variation and those measuring breeding success the most. This
conforms to expectation, whereby the body condition of long-lived
vertebrates is probably better buffered against environmental condi-
tions than breeding success in any one year. Indeed animals that do
not attain a certain level of body condition may decide not to breed
(and would therefore be unavailable for shore-based monitoring) in
a particular year.
Evaluation of the nature of relationships between grouped vari-
ables and estimates of krill abundance (Table 11.2) offers several
important conclusions.
168 J. P. Croxall

Table 11.2. Goodness of fit (R2 ) of the relationship between krill

abundance and (a) predator response vectors and (b) combined
standardized indices (CSIs), grouped by event chronology and species.
(Data taken from Reid et al. (2005))

Vector CSI

Linear model Type II functional Type II functional

response responsea response

All 0.159 0.240 0.564

Summer 0.171 0.311 0.607
Winter 0.200 0.194 0.245
Multi-year 0.079 0.050 0.004
Antarctic fur seal 0.221 0.314 0.520
Gentoo penguin 0.185 0.319 0.724
Macaroni penguin 0.154 0.199 0.373
Black-browed albatross 0.046 0.103 0.112

a
See Asseburg et al. (Chapter 18 in this volume) for more details.

80
S
70 S
S
60 S
C
50 P
F
CV

40 F F F
P F S
P
P F
30 S
P
20 P
M C M
10 M S
M M M M M
M
0
0 5 10 15 20 25 30
Ranking

Fig. 11.5 Ranked coefficients of variation (CV) of predator response vectors in

relation to the characteristics or process they are measuring. Abbreviations are:
C, breeding chronology; F, foraging/diet; M, body mass; P, population size; S,
breeding success. From Reid et al. (2005).

r Most responses (60%) are fitted significantly better by

non-linear (here Holling Type II) functional responses; for only
15% a linear relationship was significantly better.
r CSIs give much better fits than any combination of vectors in
almost all cases.
 Monitoring predator–prey interactions 169

1.5

1.0

0.5
CSI summer

0.0

0.5

1.0

1.5

2.0

2.5
0 20 40 60 80 100 120

Krill density (g m )

Fig. 11.6 Relationship between krill density and the summer CSI for
predators breeding at Bird Island (F1,98 = 28.73, p < 0.001); from Reid et al.
(2005). For examples including typical standard deviations see Boyd and
Murray (2001).

r At the species level the ranking (in terms of functional

response) tends to reflect the scale of foraging and nature of
dispersal. Thus gentoo penguins are resident with inshore
ranges (<500 km2 ), whereas black-browed albatrosses travel to
South Africa in winter and have the biggest breeding range
(>25 000 km2 ); macaroni penguins and Antarctic fur seals are
intermediate (partial or shorter-distance migrants with ranges
of 5000 to 10 000 km2 ).
r Summer variables provide a much stronger functional
relationship than winter ones, explaining 61% of the variance.

This last result (Fig. 11.6) reinforces the conclusions of Boyd and
Murray (2001) and Boyd (2002) that such relationships provide a
potential basis for management of krill to minimize adverse effects
on dependent species. Thus in the example shown, krill densities of
below 20 to 30 g m−2 could be taken as levels at which fishing should
be particularly constrained (e.g. by closed season, area restrictions
or reduction in catch level) to avoid exacerbating problems already
being encountered by the dependent species (see Boyd (2002)). Sug-
gestions for future work on methodologies for using these preda-
tor response curves in assessing krill availability and/or identifying
anomalous (low) years of krill availability are provided by Constable
and Murphy (2003).
170 J. P. Croxall

Relating predator data to changes in population size

It has proved very difficult to link changes in population size to variation in

other performance indices or to changes in prey abundance. One factor is, of
course, that population sizes are products of several demographic processes
operating at different temporal (and often spatial) scales (e.g. adult sur-
vival, juvenile survival/recruitment, deferred (non-annual) breeding). With
long-lived top predators – where the interval between fledging/weaning and
recruitment to the breeding population ranges from 3 to 5 years (gentoo
penguin, Antarctic fur seal), to 6 to 10 years (macaroni penguin), to 7 to
15 years (black-browed albatross) – there are potentially lengthy and com-
plex lag effects which are difficult to incorporate into existing models. Also,
knowledge of the scale of operation and representativeness of the monitored
population may be a limitation.
Thus, there are a range of other potential effects on populations,
exemplified in some of the population trajectories for the monitored
species at Bird Island (Fig. 11.7). Thus although Antarctic fur seals (Fig.
11.7a) have increased from 1000 individuals in the 1930s to 100 000
in the 1960s and to more than 3 million today (Boyd 1993 and BAS
unpublished data, to 2001), the population at the Bird Island moni-
toring site essentially peaked in the late 1980s and has declined sub-
sequently, currently apparently stabilizing at a lower level. This may
reflect density-dependent processes, particularly those operating in sum-
mer when the Bird Island population is constrained to forage in rela-
tively limited areas around northwest South Georgia. Therefore, not only
may some monitoring indices reflect the status of this particular ele-
ment of the South Georgia population – other elements of which are still
increasing at sites distant from Bird Island (on the mainland of South
Georgia) – but they may, therefore, not be typical of the overall South
Georgia population.
The decline in macaroni penguins at Bird Island (Fig. 11.7d) – which
is consistent over much of South Georgia – may primarily be due to envi-
ronmental changes, including those linked to prey availability; however, the
rapidity and magnitude of the decline at Bird Island is probably exacerbated
by competition for food with the massive population of Antarctic fur seals
which have co-extensive foraging ranges in summer and take the same prey
(Barlow et al. 2002).
Decreases in black-browed albatross populations (Fig. 11.7b) – again
fairly consistent at colonies elsewhere on South Georgia – probably
 Monitoring predator–prey interactions 171
900 200
(a) 180 (b)
800 160
Population
140
700 120
100
600 80
60
500 40
20
400 0
1980 1985 1990 1995 2000 1980 1985 1990 1995 2000

6000 1600
(c) (d)
5000 1400
Population

1200
4000
1000
3000
800
2000
600

1000 400
1980 1985 1990 1995 2000 1980 1985 1990 1995 2000

Year Year

Fig. 11.7 Population changes in (a) Antarctic fur seals, (b) black-browed
albatrosses, (c) gentoo penguins, and (d) macaroni penguins breeding at Bird
Island, South Georgia, 1980–2000. Dotted lines show the measured values;
continuous lines show a smoothed spline through these points. From Reid and
Croxall (2001).

substantially (but not exclusively) reflect the effect of incidental mortality

in longline (and trawl) fisheries, both around South Georgia and elsewhere
in its breeding and wintering ranges in the South Atlantic (Arnold et al.
in press). These examples illustrate the potential difficulties of interpret-
ing population change, even when detailed demographic data are available,
without collateral information on other potential influences at the popula-
tion level and scale.

Challenges and opportunities for the future

(1) The monitoring programme at Bird Island suffers considerably from

lack of other sites in the region where at least some of the same
variables are measured. This is also – to some extent – a deficiency of
the CEMP programme, where monitoring sites are few (15) relative to
the vast extent of the Southern Ocean (CCAMLR 2004). In addition,
monitoring at similar sites, but with different proximities to the main
fishing grounds, would be particularly useful.
172 J. P. Croxall

(2) It would be useful to obtain a better understanding of the sources of

variation in predator indices (including spatial and temporal
variability) and the consequences of such variability on the power to
detect trends of varying magnitude, over varying lengths of time, at
different numbers of monitoring sites and under various levels of risk.
(3) More data on prey abundance and availability, involving multi-season
surveys, need to be derived from fishery-dependent data; failing this
the use of predator-derived data as proxies – including in developing
approaches to fisheries management – would be logical, appropriate
and feasible.
(4) The increasingly well understood relationships between predators and
prey need to be more precisely related to the environmental influences
and contexts in which they occur. This will require an understanding
of the interactions between large-scale effects, whether endogenous
(like cycles in Antarctic sea-ice) or exogenous (like the El Niño
Southern Oscillation), and environmental processes at the smaller
scales at which predator foraging occurs (Croxall 1992, Trathan et al.
(Chapter 3 in this volume)).
(5) Behavioural models based on interactions between the environment,
prey, predators and fisheries may be useful in a management context,
although correct parameterization and validation of such models is
critical to their use.
(6) Further refinement and application of approaches (especially in respect
of CSIs) to develop and utilize functional relationships with predators
and their prey availability are highly desirable.
(7) There is a particular need to link functional response relationships to
population demographics, especially by using matrix models (e.g.
Arnold et al. in press).

Many of the above recommendations reflect the increasingly urgent need

to develop effective ways for data on dependent species to be integrated
into the approaches and mechanisms for managing prey stocks and pop-
ulations. This will require some new approaches to fisheries management
as well as providing more sophisticated mechanisms beyond the existing
approaches that use discount factors to allow for the needs of dependent
species.
In addition to ensuring that the ‘burden of proof ’ is appropriately bal-
anced for incorporating outputs from predator-based and prey (fishery)-
based approaches, new types of decision rules – incorporating adaptive ref-
erence points – will need to be developed. Models that include uncertainly
 Monitoring predator–prey interactions 173

need to be used to underpin management, linked to action when the prob-

ability of adverse effects is <0.05.
Finally, the framework described here and elsewhere in this book will
rapidly become part of the search for appropriate ways in which to imple-
ment ecosystem-based approaches to management, in which data derived
from upper-trophic-level predators will need to play an increasingly impor-
tant part.

ACKNOWLEDGEMENTS

I thank particularly the late Peter Prince with whom I started these pro-
grammes that were made possible by the vision and foresight of Dick
Laws; successive Heads of Biosciences at BAS – Nigel Bonner, Andrew
Clarke and Paul Rodhouse – gave unstinting support. The commitment of
Steve Hunter, Bill Doidge, Nick Lunn, Tony Williams, Simon Berrow, Iain
Staniland, Kate Barlow and Ian Boyd ensured the programme prospered,
however, without the exceptional field work of the research-assistant team
at Bird Island over the last 30 years this would have been impossible. Keith
Reid, Dirk Briggs and Andy Wood played vital roles in data management
and quality control; Chris Ricketts and Pete Rothery contributed statisti-
cal rigour to our efforts; to these and many others we owe the success of
the long-term monitoring programme at Bird Island and of the science it
underpins and complements.
I also thank Elizabeth Dixon, Julie Leland and Janet Silk for assistance
in the preparation of this chapter, which benefited from comments by Ian
Boyd, Keith Reid, Phil Trathan and an anonymous reviewer.

REFERENCES
Agnew, D. J. (1997). The CCAMLR ecosystem monitoring programme. Antarct.
Sci., 9, 235–42.
Ainley, D. G., Sydeman, W. J. & Norton, J. (1995). Upper trophic level predators
indicate interannual negative and positive anomalies in the Californian
Current food web. Mar. Ecol. Prog. Ser., 118, 69–79.
Arnold, J. M., Brault, S. & Croxall, J. P. Albatross populations in peril? A
population trajectory for black-browed albatrosses at South Georgia. Ecological
Applic., in press.
Barlow, K. E., Boyd, I. L., Croxall, J. P. et al. (2002). Are penguins and seals in
competition for Antarctic krill at South Georgia? Mar. Biol., 140, 205–13.
Bost, C. A. & Le Maho, Y. (1993). Seabirds as bio-indicators of changing marine
ecosystems: new perspectives. Acta Ecol., 14, 463–70.
Boyd, I. L. (1993). Pup production and distribution of breeding Antarctic fur seals
(Arctocephalus gazella) at South Georgia. Antarct. Sci., 5, 17–24.
174 J. P. Croxall

(2002). Integrated environment–prey–predator interactions off South Georgia:

implications for management of fisheries. Aquat. Conserv.: Mar. Freshwater
Ecosyst., 12, 119–26.
Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Boyd, I. L., Arnould, J. P. Y., Barton, T. & Croxall, J. P. (1994). Foraging behaviour
of Antarctic fur seals during periods of contrasting prey abundance. J. Anim.
Ecol., 63, 703–13.
Brierley, A. S., Watkins, J. L., Goss, C., Wilkinson, M. T. & Everson, I. (1999).
Acoustic estimates of krill density at South Georgia, 1981 to 1998. CCAMLR
Sci., 6, 47–57.
CCAMLR (Convention on the Conservation of Antarctic Marine Living Resources)
(1985). Report of the 4th Meeting of the Scientific Committee. Annex 7. Report of the
Ad-hoc Working Group on Ecosystem Monitoring. Hobart, Australia: CCAMLR.
(1986). Report of the 5th Meeting of the Scientific Committee. Annex 6. Report of the
Working Group on the CCAMLR Ecosystem Monitoring Program. Hobart,
Australia: CCAMLR.
(1987). CCAMLR Ecosystem Monitoring Program. Standard Methods. Hobart,
Australia: CCAMLR.
(1995). Report of the 14th Meeting of the Scientific Committee. Annex 4. Report of the
Working Group on Ecosystem Monitoring and Management. Hobart, Australia:
CCAMLR.
(2004). Report of the 22nd Meeting of the Scientific Committee. Annex 4. Report of
the Working Group on Ecosystem Monitoring and Management, Appendix D,
CEMP Review Workshop. Hobart, Australia: CCAMLR.
Constable, A. & Murphy, E. (2003). Using predator response curves to decide on
the status of krill availability: updating the definition of anomalies in predator
condition – preliminary analyses. In Report of the 22nd Meeting of the Scientific
Committee. Annex 4. Report of the Working Group on Ecosystem Monitoring and
Management, Appendix D, CEMP Review Workshop, Attachment 3. Hobart,
Australia: CCAMLR.
Croxall, J. P. (1992). Southern Ocean environmental change: effects on seabird,
seal and whale populations. Phil. Trans. R. Soc. Lond. B, 338, 319–28.
Croxall, J. P. & Nicol, S. (2004). Management of Southern Ocean resources: global
forces and future sustainability? Antarct. Sci., 16, 569–84.
Croxall, J. P. & Prince, P. A (1979). Antarctic seabird and seal monitoring studies.
Polar Rec., 19, 573–95.
(1987). Seabirds as predators on marine resources, especially krill, at South
Georgia. In Seabirds: Feeding Ecology and Role in Marine Ecosystems, ed. J. P.
Croxall. Cambridge, UK: Cambridge University Press, pp. 347–68.
Croxall, J. P. & Rothery, P. (1995). Population change in gentoo penguins Pygoscelis
papua at Bird Island, South Georgia: potential roles of adult survival,
recruitment and deferred breeding. In The Penguins: Ecology and Management,
eds. P. Dann, I. Norman & P. Reilly. Chipping Norton, Australia: Surrey Beatty
& Sons, pp. 26–38.
Croxall, J. P., McCann, T. S., Prince, P. A. & Rothery, P. (1988). Reproductive
performance of seabirds and seals at South Georgia and Signy Island, South
Orkney Islands, 1976–1987: implications for Southern Ocean monitoring
 Monitoring predator–prey interactions 175

studies. In Antarctic Ocean and Resources Variability, ed. D. Sahrhage. Berlin:

Springer-Verlag, pp. 261–85.
Croxall, J. P., Reid, K. & Prince, P. (1999). Diet, provisioning and productivity
responses of marine predators to differences in availability of Antarctic krill.
Mar. Ecol. Prog. Ser., 177, 115–31.
de la Mare, W. K. & Constable, A. J. (2000). Utilising data from ecosystem
monitoring for managing fisheries: development of statistical summaries of
indices arising from the CCAMLR ecosystem monitoring programme.
CCAMLR Sci., 7, 101–17.
Furness, R. W. & Camphuysen, C. J. (1997). Seabirds as monitors of the marine
environment. ICES J. Mar. Sci., 54, 726–37.
Harris, M. P. & Wanless, S. (1990). Breeding success of kittiwakes Rissa tridactyla
in 1986–88: evidence for changing conditions in the northern North Sea. J.
Appl. Ecol., 27, 172–87.
Hewitt, R. P., Demer, D. A. & Emery, J. H. (2003). An 8-year cycle in krill biomass
density inferred from acoustic surveys conducted in the vicinity of the South
Shetland Islands during the austral summers of 1991–1992 through
2001–2002. Aquat. Living Resour., 16, 205–13.
Lunn, N. J., Boyd, I. L., Barton, T. & Croxall, J. P. (1993). Factors affecting the
growth rate and mass at weaning of Antarctic fur seal pups, at Bird Island,
South Georgia. J. Mammalogy, 74, 908–19.
Lunn, N. J., Boyd, I. L. & Croxall, J. P. (1994). Reproductive performance of female
Antarctic fur seals: the influence of age, breeding experience, environmental
variation and individual quality. J. Anim. Ecol., 63, 827–40.
Monaghan, P. (1996). Relevance of the behaviour of seabirds to the conservation of
marine environments. Oikos, 77, 227–37.
Regehr, H. M. & Montevecchi, W. A. (1997). Interactive effects of food shortage
and predation on breeding failure of black-legged kittiwakes: indirect effects of
fisheries activities and implications for indicator species. Mar. Ecol. Prog. Ser.,
155, 249–60.
Reid, K. & Croxall, J. P. (2001). Environmental response of upper trophic-level
predators reveals a system change in an Antarctic marine ecosystem. Proc. R.
Soc. Lond. B, 268, 377–84.
Reid, K., Trathan, P. N., Croxall, J. P. & Hill, H. J. (1996). Krill caught by predators
and nets: differences between species and techniques. Mar. Ecol. Prog. Ser.,
140, 13–20.
Reid, K., Barlow, K., Croxall, J. P. & Taylor, R. (1999a). Predicting changes in the
Antarctic krill Euphausia superba population at South Georgia. Mar. Biol., 135,
647–52.
Reid, K., Watkins, J., Croxall, J. P. & Murphy, E. (1999b). Krill population dynamics
at South Georgia 1991–1997, based on data from predators and nets. Mar. Ecol.
Prog. Ser., 117, 103–14.
Reid, K., Croxall, J. P., Briggs, D. & Murphy, E. (2005). Antarctic ecosystem
monitoring: quantifying the response of ecosystem indicators to variability in
Antarctic krill ICES J. Mar. Sci., 62, 366–73.
SCAR (Scientific Committee for Antarctic Research) (1979). Fifteenth meeting of
SCAR, Chamonix, 16–26 May 1978. Appendix A. Working Group on Biology.
Polar Record, 19, 304–12.
176 J. P. Croxall

Weimerskirch, H., Inchausti, P., Guinet, C. & Barbraud, C. (2003). Trends in bird
and seal populations as indicators of a system shift in the Southern Ocean.
Antarct. Sci., 15, 249–56.
Williams, T. D. & Rothery, P. (1990). Factors affecting variation in foraging and
activity patterns of gentoo penguins Pygoscelis papua during the breeding
season at Bird Island, South Georgia. J. Appl. Ecol., 27, 1042–54.
 12
Impacts of oceanography on the foraging
dynamics of seabirds in the North Sea
F . D AU N T , S . W A N L E S S , G . P E T E R S , S . B E N V E N U T I ,
J . S H A R P L E S , D . G R É M I L L E T A N D B . S C O T T

Prey densities of at least 100× the average are necessary for profitable
foraging by auks A. G. Gaston (2004)

To meet the above requirement, seabirds rely on prey being distributed in

patches (Gaston 2004). Oceanography has a profound impact on the dis-
tribution of marine life (Miller 2004), and top predators frequently con-
gregate in areas with a high prey biomass (Boyd & Arnbom 1991, Hunt
et al. 1999). However, the impact of ocean physics on top-predator forag-
ing behaviour is poorly understood, largely because of the complex trophic
linkages involved. In particular, a detailed understanding of the interaction
between seabirds and their prey is lacking. Two main methods are cur-
rently available to quantify seabird behaviour: animal-borne instrumenta-
tion and at-sea observations (see Box 12.1). In this chapter, we examine the
impacts of oceanography on the foraging dynamics of North Sea seabirds
during the breeding season. The seabirds of the North Sea are primarily
piscivorous, with the majority wholly or largely dependent on the lesser
sandeel Ammodytes marinus in summer (Furness & Tasker 2000). Using
three seabird species with contrasting foraging strategies and dependence
on sandeels, we test three specific predictions from the hypothesis that
oceanography determines seabird foraging location and behaviour, using
data from animal-borne instrumentation, oceanography and primary pro-
duction collected concurrently. We interpret our findings in the context of
the behaviour of seabirds’ prey.

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
178 F. Daunt et al .

Box 12.1 Seabirds and oceanography: general methods

Two main methods are available for the collection of data on seabird
behaviour and distribution in relation to the physical environment:
animal-borne instrumentation and at-sea surveys.

Animal-borne instrumentation
The data collected by a variety of instruments attached to birds
describe the three-dimensional distribution and foraging behaviour
of individuals in detail (Wilson et al. 2002). It is also possible to
demonstrate preferences for particular hydrographic conditions, if
data are collected concurrently with oceanography. However, ship-
based cruises may not coincide in timing or location with the bird
deployments, and satellite images are frequently lacking because of
cloud cover. In addition, the number of deployments is usually low,
and so population inferences are often challenging.

At-sea surveys
Unlike animal-borne instrumentation, at-sea surveys generally have
the advantage of large sample sizes. In addition, oceanographic data
can be collected concurrently, although the slow speed of ships
results in a lack of synoptic measurement across a study area. Direct
observation of certain behaviours – especially by surface-feeders –
and interactions between individuals and species, can be made.
However, population-based inferences are problematic because the
status and origin of birds is usually unknown, and thus the intrinsic
constraints within which individuals are operating cannot be incor-
porated. In addition, data collection is biased towards what is visible
at the sea surface, and therefore depth usage is unknown.

NORTH SEA OCEANOGRAPHY AND PRODUCTIVITY

The North Sea is a semi-enclosed shelf sea (Otto et al. 1990). In spring,
the interaction between tidal currents, solar irradiation, wind patterns and
bathymetry create a mosaic of mixed, stratified and frontal regions (see Scott
et al. (Chapter 4 in this volume)). The principal frontal zone (the tidal or
shallow sea front) occurs at a point between the shelf break and the coast
where the water is shallow enough for tidal mixing to reach the surface.
Inshore of the front the water is mixed, although under certain bathymetric
 Impacts of oceanography on seabird foraging dynamics 179

conditions small regions of stratified water and fronts may occur. Pri-
mary production is typically concentrated at frontal regions (Pingree
et al. 1975, Franks 1992) and thermoclines in stratified water (Harder 1968,
Barraclough et al. 1969). Superimposed on this broad seasonal pattern are
variations in water structure due to the tidal cycle and wind. The tides
cause changes in water depth, current speed and direction (Mann & Lazier
1996).

IMPACT OF OCEAN PHYSICS ON NORTH SEA

SEABIRD PREY

The lesser sandeel is the principal prey of most North Sea seabirds (Fur-
ness & Tasker 2000). For much of the summer, autumn and winter, adult
sandeels are buried in the substrate, only entering the water column briefly
in winter to spawn (Robards et al. 1999). In spring, adult sandeels are active
in the water column during the daytime, returning to the sand at night
(Winslade 1974). Within this diurnal pattern, their distribution in the water
column is expected to be dependent on the vertical distribution of their prin-
cipal prey, calanoid copepods. However, diurnal movements of calanoid
copepods are highly flexible, diverging from the typical migration from shal-
low depths at night to deep depths during the daytime depending on pre-
dation pressure (Frost & Bollens 1992). Because of this complex dynamic,
the distribution of adult lesser sandeels in relation to ocean physics and pri-
mary production is poorly understood. Larval sandeels metamorphose in
early spring into young-of-the-year fish which have an extended pre-settled
phase where they are present in the water column throughout the daily cycle
(Jensen et al. 2003). These young fish are also preyed upon by seabirds,
and are regularly aggregated in frontal zones (Camphuysen & Webb 1999,
Camphuysen et al. (Chapter 6 in this volume)), but other precise distri-
butions in the water column – including association with phytoplankton
biomass at the thermocline – are unknown.
Clupeids, and in particular sprats Sprattus sprattus, are important alter-
native prey for a number of seabird species in the North Sea. Like sandeels,
clupeids feed primarily on zooplankton but have a very different diurnal
distribution, foraging actively near the surface at night but being inactive
at deep depths during the daytime (Blaxter & Hunter 1982). As such, they
appear to match the typical vertical migration of their prey more closely,
but the extent to which this ties in with frontal features and thermoclines is
untested.
180 F. Daunt et al .

IMPACT OF OCEAN PHYSICS ON NORTH SEA SEABIRDS

The temporal and spatial variation in hydrographic structure has been

shown to have an impact on the distribution of top predators such as
seabirds. An association between seabird distribution and horizontal fronts
has been demonstrated repeatedly in shelf seas (reviewed in Hunt et al.
(1999)). In addition, stratified regions are known to be important to diving
species (Russell et al. 1999), although no preference for the thermocline
within stratified regions of shelf seas has been shown. Daily tidal advection
influences the horizontal distribution of seabirds (Coyle et al. 1992). The
interaction between bathymetry and daily tides may also drive seabird prey
closer to the surface at certain times, affecting the timing of seabird for-
aging (Irons 1998, Hunt et al. 1999, Camphuysen et al. (Chapter 6 in this
volume)).
The method by which a species obtains food is likely to have a profound
impact on the role that oceanography plays on its ecology. Here we consider
the relationship between foraging behaviour and marine physics for three
main groups which are principally diurnal foragers, but have contrasting
foraging strategies: surface-feeders, mid-water divers and benthic divers.
We test specific predictions on three representative species breeding on the
Isle of May, southeast Scotland.

Surface-feeding species

Surface-feeding seabirds require processes that bring prey to the sea sur-
face (Garthe 1997, Camphuysen & Webb 1999). Horizontal frontal systems
are predicted to provide such opportunities, by driving prey such as young-
of-the-year lesser sandeels to the surface, in particular under certain tidal
phases when strong currents interact with bathymetry (see Camphuysen
et al. (Chapter 6 in this volume)).
We tested the prediction that chick-rearing black-legged kittiwakes Rissa
tridactyla, which feed predominantly on young-of-the-year lesser sandeels
at this time (Lewis et al. 2001), would target frontal regions. We equipped
breeding birds with activity loggers (Box 12.2) that allowed us to estimate a
maximum foraging range of 69 ± 6 km (see Fig. 12.1a and b). This distance
accords well with the distance from the breeding colony to the shallow sea
front, which runs parallel to the coast in our study area. Thus, the front
appears to form an outer barrier for breeding black-legged kittiwakes, with
foraging occurring throughout the zone between the colony and the front.
Our findings agree with at-sea surveys and telemetry that demonstrate a
consistent pattern of distribution of kittiwakes and other surface feeders
 Impacts of oceanography on seabird foraging dynamics 181

Box 12.2 Methods

Study area
The study area is the North Sea off southeast Scotland (55◦ 30 N to
56◦ 30 N, 3◦ 00 W to 0◦ 30 W). A detailed description of the sum-
mer oceanography of this area is found in Scott et al. (Chapter 4 in
this volume).

Black-legged kittiwake frontal usage

The shallow sea front runs approximately north–south along the east
coast of Britain, and is relatively stable across years (Camphuysen &
Webb 1999). The precise location varies with season and tide, but
its typical location is shown in Fig. 12.1b.
Chick-rearing black-legged kittiwakes on the Isle of May were
equipped with activity loggers each June between 1999 and 2003.
Birds were recaptured after they had made a foraging trip (1999,
n = 20 trips; 2000, n = 12; 2001, n = 6; 2002, n = 12; 2003,
n = 13). The activity loggers distinguish nest attendance, presence
on the sea surface, foraging flight and travelling flight, and record
trip duration (see Daunt et al. (2002) for full details). Travelling flight
is the only activity during which significant displacement occurs.
During chick rearing, kittiwakes carry out trips with a direct or nar-
row eliptical flight path (Wanless et al. 1992, Humphreys 2002).
Based on travelling flight speeds of 13 m s−1 (Pennycuick 1997), we
estimated maximum foraging range from

(Travelling flight duration (s) × 13)/2

Maximum range (km) =
1000

We ran four models to describe the relationship between travelling

flight duration and trip duration: constant, linear, exponential and
broken stick (full details in Daunt et al. (2002)).

Guillemot thermocline use

Primary productivity and temperature was measured by a fixed
mooring placed at 56◦ 14.79 N, 01◦ 59.41 W equipped with a fluo-
rometer and temperature loggers. The data were used to parame-
terize a one-dimensional vertical-couple physical–biological model
(see Scott et al. (Chapter 4 in this volume)), which estimates chloro-
phyll concentrations at 1-m depth bands and the depth of the
182 F. Daunt et al .

thermocline. Mean concentration of chlorophyll was determined in

three categories: above, in and below the thermocline.
Chick-rearing adult common guillemots (n = 8) were equipped
with temperature–depth loggers (PreciTD, Earth and Ocean Tech-
nologies, Kiel, Germany) on the Isle of May in 2002. These loggers
have rapid-response temperature probes enabling accurate detection
of thermocline depth (Daunt et al. 2003). Dives (n = 1291) were split
into bouts (after Sibly et al. (1990)) and the top and bottom of the
thermocline were estimated for each foraging bout. The foraging
depth of each dive was determined using Multitrace (Jensen Soft-
ware Systems, Kiel, Germany) and classed in the same three cate-
gories as chlorophyll concentration.

European shag tidal preference

Tide tables were available for Ansthruther, 10 km from the Isle of
May, providing the timing and height of low and high tide through-
out the study period. Tidal height was estimated by running a sine
function through each cycle bounded by the known tidal heights at
high and low tide.
Chick-rearing adult European shags (n = 48) were caught on the
Isle of May in 2002 and equipped with temperature–depth loggers
(PreciTD). Mean foraging depth, time at foraging depth, dive dura-
tion and sea surface duration were determined for each dive (n =
3918) in Multitrace. Following Grémillet et al. (1998), dives were
defined as benthic (flat-bottom shape, 83%) or pelagic (V-shaped,
17%). Each benthic dive was associated with tidal height. To deter-
mine preference, the proportion of time foraging in each tidal band
was corrected for the availability of tidal height during daylight
(because shags do not feed at night, Wanless et al. (1993)).

at – and westwards of – the front, with very few east of the front (Cam-
phuysen & Webb 1999, Humphreys 2002). However, unlike at-sea surveys,
we found no strong evidence that birds were targeting the front over other
regions within the birds’ foraging range.

Mid-water divers

In the North Sea, the shallow sea front is important for mid-water divers
(Camphuysen & Webb 1999). In addition, these species can exploit the
water column, and would be expected to target depths where prey are con-
centrated. Primary production is aggregated at the thermocline in stratified
 Impacts of oceanography on seabird foraging dynamics 183

(a) Travelling flight duration (min) 300

250

200
179
150

100

50

0
414
0 100 200 300 400 500 600 700 800
Foraging trip duration (min)

1999 2000 2001 2002 2003

Fig. 12.1 (a) Relationship between travelling flight duration and trip duration
during foraging trips of black-legged kittiwakes in 1999–2003. A broken-stick
model with flat asymptote provided the best fit to the model (77.9% of the
variation explained). The slope of the line is initially estimated at 0.43 ± 0.018,
before flattening at a trip duration of 414 min and a flight duration of 179 min.
Thereafter, there is no increase in distance travelled with increasing trip
duration. (b) Map of the study area, showing the inferred mean ± SE maximum
foraging range (69 ± 6 km) of kittiwakes breeding on the Isle of May. Maximum
range coincided with the position of the shallow sea front, and kittiwakes foraged
throughout the zone between the colony and the front. The 30-, 40- and 50-m
bathymetric contours are shown. Also shown are 45◦ and 135◦ bearings relative to
the colony, between which most kittiwake foraging trips are located (Humphreys
2002).
184 F. Daunt et al .

6 1.0

5
Chl A concentration (mg m ) 0.8

4
0.6

Proportion of dives
3
0.4
2

0.2
1

0 0
Above thermocline In thermocline Below thermocline

Fig. 12.2 Chlorophyll A concentration (mean + SD, open bars) and proportion
of dives of common guillemots (filled bars) above, in and below the thermocline,
18–26 June 2002.

waters (Mann & Lazier 1996). However, the extent to which the thermo-
cline is an important zone for foraging seabirds is dependent on its use by
their prey.
We examined the importance of the thermocline to a mid-water diver,
the common guillemot Uria aalge, a species that feeds on both lesser
sandeels and sprats (Harris & Wanless 1985). We equipped common guille-
mots with rapid-response temperature–depth loggers that record external
temperature very accurately (Daunt et al. 2003). Data collected concurrently
by fixed moorings allowed primary production to be modelled throughout
the water column (Box 12.2), and demonstrated that primary production
was highest at the thermocline, lowest below the thermocline and inter-
mediate above the thermocline (Fig. 12.2). Although guillemots foraged
almost exclusively in stratified water (Daunt et al. 2003), foraging effort
was strongly targeted at the zone below the thermocline (generalized lin-
ear mixed model (GLMM) with individual as random effect, W = 28.0,
p < 0.001; Fig. 12.2; Box 12.2). This distribution matched the daytime dis-
tribution of sprats rather than lesser sandeels and accorded well with sprat
being the principal prey delivered to chicks during the study period (78% of
prey deliveries).

Benthic divers

In contrast to other species groups, the feeding distribution of benthic-

feeding species is not associated with fronts or thermoclines (Daunt et al.
 Impacts of oceanography on seabird foraging dynamics 185

(a) 25 (b) 0.38

Proportion of time at foraging depth

24
0.36
Foraging depth (m)

23

22 0.34

21
0.32

20
0.30
19

18 0.28
0 1 2 3 4 5 0 1 2 3 4 5
Tidal height band (m) Tidal height band (m)

(c) 0.20
Proportion of foraging time

0.15

0.10

0.05

0
0 1 2 3 4 5
Tidal height band (m)

Fig. 12.3 (a) Foraging depth; (b) proportion of time in water spent at foraging
depth; (c) proportion of time spent foraging by European shags in each tidal
height band (1 = 0.5 to 1.5 m, 2 = 1.5 to 2.5 m, etc.) corrected for availability.

2003). Rather, the tidal cycle is expected to be an important physical driver of

foraging behaviour, because it has a direct impact on the distance between
the sea surface and the foraging habitat. Thus, benthic divers are predicted
to forage preferentially at low tide.
We examined diving behaviour during different phases of the tide in the
European shag Phalacrocorax aristotelis, a benthic-diving species specializ-
ing on adult lesser sandeels (Harris & Wanless 1991). We found that dive
depth closely matched tidal height, with a 1-m change in tidal height corre-
sponding on average to a 1.2-m increase in dive depth (Fig. 12.3a; regression,
depth = 18.6 + 1.2 × height; restricted maximum likelihood (REML) with
individual as random effect: tidal height, F1,46 = 234.7, p < 0.001). The
increased time spent travelling through the water column, together with a
disproportionately longer recovery period between dives (see Wanless et al.
186 F. Daunt et al .

1993), resulted in the proportion of time at foraging depth being 20% lower
at high tide than low tide (ratio of time at foraging depth to dive + surface
duration, REML, tidal height: F1,46 = 76.1, p < 0.001; Fig. 12.3b). Despite
this, there was no evidence that birds preferred to forage during low tide
(Fig. 12.3c; GLMM: W = 0.2, p = 0.94).

CONCLUSIONS

There is support both for and against a direct association between oceanog-
raphy and the foraging dynamics of seabirds in the North Sea. The shallow
sea front is favoured by both surface-feeding and mid-water-diving species,
supporting the view that these physical features form an important focus
for marine life at all trophic levels. There is also some indication that the
tidal cycle has an impact on temporal foraging opportunities for surface-
feeders (see Camphuysen et al. (Chapter 6 in this volume)), although the
potential impact of tidal current and direction on horizontal distribution
remains unclear. In contrast, the shallow sea front was not favoured by
breeding black-legged kittiwakes over other habitats within their foraging
range. In addition, we found no evidence for a preference for the thermo-
cline by common guillemots diving in stratified regions, with birds consis-
tently diving through the thermocline to areas with the lowest levels of pri-
mary production. Finally, there was no evidence that benthic-feeding shags
adjusted their foraging activity in response to the tidal cycle.
There are a number of reasons why the link between oceanography and
avian top predators may be weak. The most significant of these is likely to be
the number of trophic links between ocean physics and seabirds (typically
four levels in the North Sea). The association between primary production
and oceanography is strong. However, as one moves up the food web the
interactions become more complex. North Sea seabirds are feeding primar-
ily on small fish that are active swimmers and may only aggregate at regions
of high productivity when zooplankton, their principal prey, are feeding in
these zones. Depth utilization and observations of prey delivered to chicks
both suggest that common guillemots are targeting sprats which are inac-
tive at the sea floor during the day. As such, a close association between
guillemots and primary production is not expected. The lack of a preference
for different phases of the tide by European shags suggests that there are
patterns in the behaviour of adult lesser sandeels that are more important
in determining the timing of foraging than reduced distance between sea
surface and foraging habitat apparent during low tide. Clearly, an important
focus for future research is to gain a better understanding of lesser sandeel
 Impacts of oceanography on seabird foraging dynamics 187

and sprat behaviour under different oceanographic conditions. In contrast,

the more direct associations between seabirds and oceanography demon-
strated in the Bering Sea (reviewed in Hunt et al. (1999)) may be in part
due to their largely planktonic diet, and thus a closer coupling between the
predators, productivity and physics.
Seabird distribution is also dictated by the behaviour of other species.
Multispecies feeding associations are common in the North Sea (Camphuy-
sen & Webb 1999, Camphuysen et al. (Chapter 6 in this volume)). In these
associations, diving species such as cetaceans and auks drive fish to the
surface where they become available to surface-feeding species such as the
black-legged kittiwake. This interaction is maintained until the shoal dis-
perses or is entirely consumed, or the flock is disrupted by large gulls or
northern gannets Morus bassanus. In such situations, seabirds may occur in
locations that could not be predicted by ocean physics. Equally, if an asso-
ciation with a front is detected, it may not be directly due to this feature –
but rather to the presence of other predators facilitating the availability of
prey.
Intrinsic mechanisms are also likely to be important. During the spring
and summer the breeding component of the population is under two impor-
tant constraints. Firstly, individuals are restricted in their foraging range
by the need to return repeatedly to the colony to feed the young. Thus,
there may be a trade-off between habitat profitability and distance from
the nest site, such that birds may not always prefer the highest-quality
areas (Davoren et al. 2003). The stronger preference for the shallow sea
front demonstrated from at-sea surveys (Camphuysen & Webb 1999) com-
pared with colony-based work (this study) may in part be due to the former
comprising a large proportion of non-breeding individuals. Secondly, the
elevated energetic costs associated with rearing offspring may force birds
to forage at times that are sub-optimal. This constraint is compounded
by the need to allocate a large proportion of available time to offspring
attendance.
The best approach to understanding the impact of marine physics on
seabird foraging dynamics is to combine animal-borne instrumentation
with at-sea survey data. Further, the relationship between oceanography
and the foraging dynamics of seabirds must be interpreted in tandem with
other key processes, notably fish distribution and behaviour (Montevecchi
et al. (Chapter 8 in this volume)), if we are to quantify the role of physics
in clumping prey at densities that can be foraged profitably by seabirds. A
better understanding of the physical mechanisms driving seabird foraging
dynamics will enable us to understand and predict population responses
188 F. Daunt et al .

of top predators to physical changes in the marine environment (Parrish

& Zador 2003). This approach is very relevant to the North Sea, which is
currently undergoing considerable physical changes that are having a sig-
nificant impact on the ecology of the ecosystem (Beaugrand 2004).

ACKNOWLEDGEMENTS

The work was funded by European Commission project ‘Interactions

Between the Marine Environment, Predators and Prey: Implications for
Sustainable Sandeel Fisheries (IMPRESS; QRRS 2000-30864)’. We thank
Scottish Natural Heritage for permission to work on the Isle of May. Many
thanks are due to Mike Harris, Linda Wilson, Sue Lewis and Debbie Russell
for help in the field. Oliver Ross, Simon Greenstreet, Helen Fraser, Gayle
Holland, Sarah Hughes, John Dunn and George Slessor provided support
with collection and analysis of the mooring data. We thank Alberto Ribolini
and Luigi Dall’Antonia for activity-logger and associated software develop-
ment, Paolo Lambardi for analysis of activity data and Morten Frederiksen
for comments on the manuscript. The coastline and bathymetry are repro-
duced by permission of the Controller of Her Majesty’s Stationery Office
and the UK Hydrographic Office.

REFERENCES
Barraclough, W. E., LeBrasseur, R. J. & Kennedy, O. D. (1969). Shallow scattering
layer in the subarctic Pacific Ocean: detection by high-frequency echo sounder.
Science, 166, 611–13.
Beaugrand, G. (2004). The North Sea regime shift: evidence, causes, mechanisms
and consequences. Prog. Oceanogr., 60, 245–62.
Blaxter, J. H. S. & Hunter, J. R. (1982). The biology of the clupeoid fishes. Adv. Mar.
Biol., 20, 3–203.
Boyd, I. L. & Arnbom, T. (1991). Diving behavior in relation to water temperature
in the Southern elephant seal: foraging implications. Polar Biol., 11, 259–66.
Camphuysen, C. J. & Webb, A. (1999). Multi-species feeding associations in
North Sea seabirds: jointly exploiting a patchy environment. Ardea, 87, 177–98.
Coyle, K. O., Hunt, G. L., Decker, M. B. & Weingartner, T. J. (1992). Murre
foraging, epibenthic sound scattering and tidal advection over a shoal near St
George Island, Bering Sea. Mar. Ecol. Prog. Ser., 83, 1–14.
Daunt, F., Benvenuti, S., Harris, M. P. et al. (2002). Foraging strategies of the
black-legged kittiwake Rissa tridactyla at a North Sea colony: evidence for a
maximum foraging range. Mar. Ecol. Prog. Ser., 245, 239–47.
Daunt, F., Peters, G., Scott, B., Grémillet, D. & Wanless, S. (2003). Rapid response
recorders reveal interplay between marine physics and seabird behaviour. Mar.
Ecol. Prog. Ser., 255, 283–8.
 Impacts of oceanography on seabird foraging dynamics 189

Davoren, G. K., Montevecchi, W. A. & Anderson, J. T. (2003). Distribution

patterns of a marine bird and its prey: habitat selection based on prey and
conspecific behaviour. Mar. Ecol. Prog. Ser., 256, 229–42.
Franks, P. J. S. (1992). Sink or swim: accumulation of biomass at fronts. Mar. Ecol.
Prog. Ser., 82, 1–12.
Frost, B. W. & Bollens, S. M. (1992). Variability of diel vertical migrations in the
marine planktonic copepod Pseudocalanus newmani in relation to its predators.
Can. J. Fish. Aquat. Sci., 49, 1137–41.
Furness, R. W. & Tasker, M. L. (2000). Seabird–fishery interactions: quantifying
the sensitivity of seabirds to reductions in sandeel abundance, and
identification of key areas for sensitive seabirds in the North Sea. Mar. Ecol.
Prog. Ser., 202, 354–64.
Garthe, S. (1997). Influence of hydrography, fishing activity, and colony location
on summer seabird distribution in the south-eastern North Sea. ICES J. Mar.
Sci., 54, 566–77.
Gaston, A. G. (2004). Seabirds: A Natural History. London: T. & A. D. Poyser.
Grémillet, D., Argentin, G., Schulte, B. & Culik, B. M. (1998). Flexible foraging
techniques in breeding cormorants Phalacrocorax carbo and shags
Phalacrocorax aristotelis: benthic or pelagic feeding? Ibis, 140, 113–19.
Harder, W. (1968). Reactions of plankton organisms to water stratification. Limnol.
Oceanogr., 13, 156–68.
Harris, M. P. & Wanless S. (1985). Fish fed to young guillemots, Uria aalge,
and used in display on the Isle of May, Scotland. J. Zool. Lond., 207,
441–58.
(1991). The importance of the lesser sandeel Ammodytes marinus in the diet of
the shag Phalacrocorax aristotelis. Ornis Scand., 22, 375–82.
Humphreys, E. M. (2002). Energetics of spatial exploitation of the North Sea by
kittiwakes breeding on the Isle of May, Scotland. Unpublished Ph.D. thesis,
University of Stirling, Scotland, UK.
Hunt, G. L. Jr, Mehlum, F., Russell, R. W. et al. (1999). Physical processes, prey
abundance, and the foraging ecology of seabirds. In Proceedings of the 22nd
International Ornithological Congress, Durban Johannesburg, 16–28 August, 1998,
eds. N. J. Adams & R. Slotow. Johannesburg, South Africa: BirdLife South
Africa, pp. 2040–56.
Irons, D. B. (1998). Foraging area fidelity of individual seabirds in relation to tidal
cycles and flock feeding. Ecology, 79, 647–55.
Jensen, H., Wright, P. J. & Munk, P. (2003). Vertical distribution of the pre-setttled
sandeel (Ammodytes marinus) in the North Sea in relation to size and
environmental variables. ICES J. Mar. Sci., 60, 1342–51.
Lewis, S., Wanless, S., Wright, P. J. et al. (2001). Diet and breeding performance of
black-legged kittiwakes Rissa tridactyla at a North Sea colony. Mar. Ecol. Prog.
Ser., 221, 277–84.
Mann, K. H. & Lazier, J. R. N. (1996). Dynamics of Marine Ecosystems. Oxford, UK:
Blackwell Science.
Miller, G. B. (2004). Biological Oceanography. Oxford, UK: Blackwell Science.
Otto, L., Zimmerman, J. T. F., Furnes, G. K. et al. (1990). Review of the physical
oceanography of the North Sea. Neth. J. Sea Res., 26, 161–238.
190 F. Daunt et al .

Parrish, J. K. & Zador, S. G. (2003). Seabirds as indicators: an exploratory analysis

of physical forcing in the Pacific Northwest coastal environment. Estuaries, 26,
1044–57.
Pennycuick, C. J. (1997). Actual and ‘optimum’ flight speeds: field data reassessed.
J. Exp. Biol., 200, 2355–61.
Pingree, R., Pugh, P. R., Holligan, P. M. & Forster, G. R. (1975). Summer
phytoplankton blooms and red tides in the approaches to the English Channel.
Nature, 258, 672–7.
Robards, M. D., Willson, M. F., Armstrong, R. H. & Piatt, J. F. (eds.) (1999). Sand
Lance: A Review of Biology and Predator Relations and Annotated Bibliography.
Portland, OR: US Department of Agriculture.
Russell, R. W., Harrison, N. M. & Hunt, G. L., Jr (1999). Foraging at a front:
hydrography, zooplankton, and avian planktivory in the northern Bering Sea.
Mar. Ecol. Prog. Ser., 182, 77–93.
Sibly, R. M., Nott, H. M. R. & Fletcher, D. J. (1990). Splitting behaviour into bouts.
Anim. Behav., 39, 63–9.
Wanless, S., Monaghan, P., Uttley, J. D., Watton, P. & Morris, J. A. (1992). A
radio-tracking study of kittiwakes (Rissa tridactyla) foraging under suboptimal
conditions. In Wildlife Telemetry: Remote Monitoring and Tracking of Animals,
eds. I. G. Priede & S. M. Swift. New York: Ellis Horwood, pp. 580–90.
Wanless, S., Corfield, T., Harris, M. P., Buckland, S. T. & Morris, J. A. (1993).
Diving behaviour of the shag Phalacrocorax aristotelis (Aves: Pelicaniformes) in
relation to water depth and prey size. J. Zool. Lond., 231, 11–25.
Wilson, R. P., Grémillet, D., Syder, J. et al. (2002). Remote-sensing systems and
seabirds: their use, abuse and potential for measuring marine environmental
variables. Mar. Ecol. Prog. Ser., 228, 241–61.
Winslade, P. (1974). Behavioural studies on the lesser sandeel Ammodytes marinus
(Raitt) ii. The effect of light intensity on activity. J. Fish Biol., 6, 577–86.
 13
Foraging energetics of North Sea birds
confronted with fluctuating prey availability
M . R . E N S T I P P , F . D AU N T , S . W A N L E S S ,
E. M. HUMPHREYS, K. C. HAMER, S. BENVENUTI
A N D D . G R É M I L L E T

In the western North Sea, a large seabird assemblage exploits a limited

number of fish species. Sandeels are particularly important prey items in
this system, with populations that show strong spatial and temporal vari-
ability. This variability might be triggered by oceanic climatic features but
could also be influenced by human activities, especially fisheries. In order to
assess how different sandeel consumers are buffered against fluctuations in
prey availability, we studied the foraging energetics of common guillemots,
black-legged kittiwakes, European shags and northern gannets at two major
colonies in southeast Scotland. Our analysis was based on: (a) time budgets
recorded with data loggers attached to breeding adults foraging at sea; (b)
metabolic measurements of captive and free-ranging individuals; and (c)
information on diet and parental effort. We calculated daily food intake and
feeding rates of chick-rearing adults and examined a number of hypothet-
ical scenarios, to investigate how birds might be buffered against reduced
sandeel availability. Our results suggest that under the conditions currently
operating in this region, shags and guillemots may have sufficient time and
energy available to increase their foraging effort considerably, whereas kitti-
wakes and gannets are more constrained by time and energy respectively. Of
the species considered here, gannets are working at the highest metabolic
level during chick rearing, and hence, have the least physiological capacity
to increase foraging effort. However, to compensate for their energetically
costly life, gannets might make use of a highly profitable foraging niche.

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
192 M. R. Enstipp et al.

Human activities, such as commercial fisheries, have produced major

changes in the structure of marine food webs (Pauly et al. 1998) but we
know very little about the mechanisms involved. Species at intermediate
trophic levels in such webs undergo strong spatial and temporal fluctu-
ations, making it difficult to assess and monitor their populations. Con-
versely, predators at upper trophic levels, such as seabirds, are very conspic-
uous and are potentially reliable indicators of the state of marine systems
(Cairns 1987, Montevecchi 1993, Furness & Camphuysen 1997). Thus,
studying higher marine predators can provide insights into the mecha-
nisms structuring marine food webs.
In the western North Sea, a large seabird assemblage exploits a small
number of fish species. Sandeels, especially the lesser sandeel (Ammodytes
marinus), are important prey items in this system and comprise a major
component of the diet of seabirds, marine mammals and predatory fish
(see Furness & Tasker (2000)). Sandeel populations show strong spatial
and temporal variability, which is poorly understood. A marked decline in
sandeels around Shetland in the mid 1980s had adverse effects on many
seabird species. Surface-feeders like Arctic terns (Sterna paradisaea) and
black-legged kittiwakes (Rissa tridactyla) had greatly reduced breeding suc-
cess, while diving species like common guillemots (Uria aalge) were able
to compensate for the reduction in sandeel availability to some extent by
increasing their foraging effort (Monaghan 1992, Monaghan et al. 1996).
In 1990 a sandeel fishery opened around the Firth of Forth area (southeast
Scotland) and expanded rapidly, coinciding with a decline in breeding per-
formance of kittiwakes from nearby colonies (Tasker et al. 2000). Concern
for the future of these predators culminated in the closure of the fishery in
2000.
Furness and Tasker (2000) found that small seabirds with high ener-
getic costs during foraging and a limited ability to switch diet (e.g. many
surface-feeders) were most sensitive to a reduction in sandeel abundance.
Larger species with less costly foraging modes and a greater ability to switch
diet (e.g. many pursuit-diving species) were less sensitive. Furness and
Tasker were, however, uncertain about the relative importance of some fac-
tors – such as foraging energetics. Hence, it is important to test the hypoth-
esis that the impacts of reduced sandeel availability on seabirds depend on
energetic and behavioural constraints during foraging.
The current study addresses these issues in four North Sea seabird
species during the period of chick rearing in southeast Scotland. Our
study included: two pursuit-diving species, the common guillemot and the
 Foraging energetics of North Sea birds 193

European shag (Phalacrocorax aristotelis); one surface-feeding species, the

black-legged kittiwake; and one plunge-diving species, the northern gannet
(Morus bassanus). We calculated the daily food intake (DFI) from knowledge
of time–activity budgets, energy expenditures and diet (Box 13.1 and Table
13.1), allowing the estimation of required feeding rates (catch per unit effort
(CPUE)) under a number of scenarios that investigated the capacity of the
four species to compensate for a reduction in sandeel availability by altering
their foraging behaviour.

Time–activity/energy budgets

The daily time–activity budget indicated that all species except shags spent
about 50% of their time at the colony and 50% at sea. Shags on the other
hand allocated only about 15% of their time towards food acquisition, and
stayed at the colony for the remainder of the time. Kittiwakes, gannets and
guillemots spent a considerable amount of their time at sea resting (15% to
30%), but resting at sea was negligible in shags. Shags and guillemots spent
a much smaller proportion of their time flying than kittiwakes and gannets,
reflecting the use of prey patches closer to the colony. Daily energy expendi-
ture (DEE) calculated for the four species considered (Table 13.2) compared
well with reported energy expenditures measured in the field using doubly
labelled water (DLW), where available. The time–energy budget emphasized
the relative importance of energetically expensive activities, especially flight,
on the overall daily energy expenditure. While birds spent only between 13%
and 34% of their day active at sea, this period accounted for 39% to 60%
of their daily energy expenditure. Gannets worked the hardest with a field
metabolic rate (FMR) of 3.9 × BMR, while all other species worked at a level
of around 3 × BMR (Table 13.2).
CPUE values (based on active time spent at sea; see Table 13.2) for
shags and gannets were high compared with the other species, with shags
foraging most efficiently (Table 13.3; foraging efficiency is defined as the
ratio of metabolizable energy gained during foraging to energy used during
foraging).

Sensitivity analysis

An assessment of the sensitivity of the calculation of prey requirements to

each variable used in the calculation (Table 13.4) indicated that the time
spent in each activity and the caloric density of the prey ingested had the
194 M. R. Enstipp et al.

Box 13.1 Methods used to establish time--energy budgets

Shags and guillemots were equipped with compass loggers and/or
precision temperature–depth recorders (PreciTD; both from Earth
and Ocean Technologies, Kiel, Germany). These provided very fine-
scale activity data that distinguished between phases of rest on land
or at sea from flight and diving. A flight activity sensor combined
with a saltwater switch was deployed on kittiwakes (Instituto di Elab-
orazione dell’Informazione, C. N. R., Pisa, Italy) and this allowed us
to distinguish between periods of flight associated with travelling or
foraging, and periods of rest on land and at sea. Satellite tags (PTT;
Microwave Telemetry, Inc., Columbia, Maryland, USA) on gannets
enabled us to distinguish between periods spent at the colony and
periods at sea; PreciTD loggers allowed us to distinguish time spent
in flight, time spent submerged and time spent resting at sea. All
field data for kittiwakes, shags and guillemots were collected dur-
ing the early chick-rearing period (June to July) from 1999 to 2003
on the Isle of May, Firth of Forth, southeast Scotland. Field data
for the gannets were collected from the nearby Bass Rock breeding
colony during early to mid chick rearing in 2003. Input values for
our algorithm were generated from yearly mean values for the time
that birds spent in various activities per day, weighted according to
sample size.
Activity-specific metabolic rates for shags were measured directly
via respirometry. This included measurement of basal metabolic
rate (BMR) and metabolic rate during resting on water and dur-
ing diving, incorporating the effect of water temperature (Enstipp
et al. 2005). All other values were compiled from the literature. For
kittiwakes all metabolic rates, except for those in flight, were taken
from Humphreys (2002). BMR for gannets was taken from Bryant
and Furness (1995) and metabolic rate during resting at sea was
taken from Birt-Friesen et al. (1989). For guillemots we used the
BMR value given by Hilton et al. (2000a) who established a regres-
sion equation from all published BMR values. Metabolic rate during
resting at sea and during diving (incorporating the effect of water
temperature) was taken from Croll and McLaren (1993). To account
for activities at the nest such as chick feeding and preening, which
will increase metabolic rate above BMR, we assumed a metabolic
 Foraging energetics of North Sea birds 195

rate at the nest that was twice the BMR; this value was used for all
species except for the kittiwake where we used the measured value
from Humphreys (2002). To incorporate the effect of water tem-
perature on metabolic rate during resting at sea for kittiwakes and
gannets we used the slope given by Croll and McLaren (1993) for
guillemots. In the absence of data we assumed that metabolic costs
of travel flight and forage flight for the kittiwake are identical and
the same assumption was made for flying and plunge-diving for the
gannet. All estimates of energetic costs during flight were calculated
using the aerodynamic model of Pennycuick (1989), using the latest
version ‘Flight 1.13’. Wing morphology values were taken from Pen-
nycuick (1987). We accounted for the presumably higher flight costs
during the return trip, after birds have ingested food and carry food
for their chicks. Estimates of the daily energy expenditure of chicks
were based on those provided by Visser (2002) for all species except
the guillemot – which was taken from Harris and Wanless (1985),
corrected for assimilation efficiency.
Diet samples were collected as regurgitations, observations of
prey delivered to chicks or from food dropped at the ledge. A mean
calorific value for prey taken was established for each species based
on the biomass proportions of prey and its size. Calorific values of
the various prey items were taken from the literature (Hislop et al.
1991, Bennet & Hart 1993, Pedersen & Hislop 2001) accounting
for seasonal effects. We took assimilation efficiencies for the gan-
net from Cooper (1978) and for all other species from Hilton et al.
(2000b). Assimilation efficiency for chicks was assumed to be the
same as in adults except in kittiwakes, for which we took the value
from Gabrielsen et al. (1992).
Body masses were obtained from birds during routine handling
associated with ringing. Breeding success was determined as the
number of chicks fledged from surveyed nests where eggs had been
laid. We took water temperatures from Daunt et al. (2003) who, in
the same area, measured water temperatures directly from foraging
shags and guillemots during chick rearing.
The algorithm used to compile the time–energy budgets (‘base-
line situation’, see Table 13.1 for key input values) and to investi-
gate the different scenarios was based on Grémillet et al. (2003)
but incorporated the energetic requirements of chicks. CPUE val-
ues (Table 13.2) are based on the time spent underwater for shags
196 M. R. Enstipp et al.

and guillemots, the time spent in forage flight for kittiwakes and the
total time spent at sea for gannets (a CPUE value based on the active
time spent at sea is included in brackets to allow comparison across
species). We conducted a sensitivity analysis (Table 13.4) to test the
robustness of our algorithm (Grémillet et al. 2003).

strongest influence on the total energy expenditure. The calculations for

shags and guillemots were particularly sensitive to variation of the amount
of time spent flying per day. In contrast, kittiwakes were most sensitive to
time spent resting at the colony, whereas gannets were equally sensitive
to time spent flying, resting at sea and resting at the colony. These results
emphasize the importance of measuring these variables as precisely and
accurately as possible.

Potential responses to decreased sandeel availability

Seabirds foraging in the North Sea are constrained by a delicate balance of

the following three components: (a) the time they can allocate towards food
acquisition; (b) the energy demands associated with their activities; and (c)
the food they are able to acquire. Confronted with a decline in availability of
a particular prey species (e.g. sandeel), seabirds have a number of potential
options to maintain their DFI at a sustainable level. For some it might be
possible to switch to other prey species (e.g. clupeids or gadids) or to make
greater use of fish discarded as bycatch in certain fisheries. Alternatively,
they might be able to increase their foraging effort in a number of ways.
In the following scenarios we explored the capacity of the four species to
increase their foraging effort within the constraints imposed upon them
by time, energy and food. In all scenarios the increased amount of time
allocated towards prey acquisition was balanced by reducing the time spent
resting at sea and at the colony. While decreasing resting time at sea to zero
we decreased resting time at the colony only to a minimum of 50% of the
daily total, assuming that chicks were not left unattended. We also took into
consideration that all species were inactive for some part of the night, dur-
ing which no foraging activity occurred (shags: 8 h, Wanless et al. (1999);
guillemots: 1 h, F. Daunt, unpublished observations, 2004; kittiwakes: 3 h,
Daunt et al. (2002); gannets: 5 h, E. M. Humphreys, unpublished observa-
tions, 2004). Assuming that partners shared the available time equally, the
total time that could potentially be allocated towards foraging activity by an
adult per day ranged from 8 h for shags to 11.5 h for guillemots.
Table 13.1. Some of the input values (means ± SD) used to compile a time–energy budget (‘baseline situation’) for four
North Sea seabirds during chick rearing

Black-legged kittiwake European shag Northern gannet Common guillemot

Body mass (g) 361.64 ± 36.14 1780.43 ± 97.63 2998 ± 234 920.34 ± 57.44
Assimilation efficiency for chick (%) 80.00 ± 1.25
Calorific value of fish (kJ g−1 wet mass) 5.0 ± 0.5 5.4 ± 0.5 5.8 ± 0.6 5.1 ± 0.5
Water temperature at surface (◦ C) 11.1 ± 0.5 11.1 ± 0.5 11.1 ± 0.5 12.0 ± 0.5
Water temperature at bottom (◦ C) 10.3 ± 0.4 8.8 ± 0.5
BMR (kJ day−1 ) 267.28 726.07 ± 46.15 1256.28 ± 227.94 584.48
Energy costs, resting at colony (W kg−1 ) 13.69 ± 1.20 9.44 ± 0.6 9.70 ± 1.76 14.70 ± 1.47
Energy costs, resting at sea (W kg−1 ) 12.82 ± 2.56 17.18 ± 2.02 12.46 ± 2.16 10.19 ± 1.02
Energy costs, flying (W kg−1 ) 44.83 ± 4.48 98.07 ± 9.81 43.69 ± 4.37 92.58 ± 9.26
Energy costs, foraging (W kg−1 ) 44.83 ± 4.48 20.58 ± 2.8 43.69 ± 4.37 23.83 ± 2.38
DEE of chick (kJ day−1 ) 525.71 ± 52.57 1203.98 ± 120.40 1593.30 ± 159.33 221.71 ± 22.17

BMR, basal metabolic rate; DEE, daily energy expenditure.

Table 13.2. Daily energy expenditure (DEE), field metabolic rate (FMR expressed as a multiple of BMR), daily food intake
(DFI) and feeding rate (catch per unit effort, CPUE) for four North Sea seabirds (‘baseline situation’)

Black-legged kittiwake European shag Northern gannet Common guillemot

Adult
DEE (kJ day−1 ) 786.74 2249.25 4856.01 1641.01
FMR (× BMR) 2.9 3.1 3.9 2.8
DFI (g fish day−1 ) 211 514 1114 415
Chick
DEE (kJ day−1 ) 525.71 1203.98 1593.30 221.71
DFI (g fish day−1 ) 131 275 366 56
No. of chicks fledged/pair 0.71 1.51 0.67 0.69
DFI (g fish day−1 , portion/adult) 47 208 122 19
Total
DFI (g fish day−1 ) 258 722 1237 434
CPUE (g fish min−1 ) 1.35 (0.50) 10.10 (3.84) 1.63 (3.89) 2.45 (1.18)

CPUE values are based on the time spent underwater for shags and guillemots, the time spent in forage flight for kittiwakes and the total time
spent at sea for gannets. To allow comparison across species a CPUE value based on the active time spent at sea (excluding periods of rest at sea)
is included in brackets.
Table 13.3. Foraging efficiency (ratio of metabolizable energy gained during foraging to energy used during foraging) and
foraging range of four North Sea seabirds

Energy acquired at Energy expenditure

sea per day to meet at sea per day Foraging range (km)
adult and chick Foraging efficiency
requirements Total Activea ‘Baseline Potential
Species (kJ day−1 ) (kJ day−1 ) (kJ day−1 ) At sea Activea situation’ increase

Black-legged kittiwake 972.07 534.44 473.59 1.82 2.05 49.6 +34.0

European shag 3158.25 932.76 879.27 3.39 3.59 10.4 +11.1
Northern gannet 5389.76 3484.77 2500.13 1.55 2.16 282.4 +26.8
Common guillemot 1715.00 960.93 784.20 1.78 2.19 21.8 +41.5

a
Excludes periods of rest at sea.
Table 13.4. Sensitivity analysis for the time–energy budget of four North Sea seabirds

Black-legged kittiwake European shag Northern gannet Common guillemot

Variation of Range Variation of Range Variation of Range Variation of Range

mean DFI (%) used mean DFI (%) used mean DFI (%) used mean DFI (%) used

Body mass (g) ± 8.1 SD ± 1.7 SD ± 5.2 SD ± 8.3 SD

Time resting at colony (min day−1 ) ± 10.8 SD ± 2.7 SD ± 9.9 SD ± 1.0 SD
Time resting at sea (min day−1 ) ± 2.3 SD ± 0.3 SD ± 9.4 SD ± 6.6 SD
Time spent flying (min day−1 ) ± 2.7 SD ± 5.3 SD ± 10.1 SD ± 18.5 SD
Time spent foraging (min day−1 ) ± 1.5 SD ± 2.0 SD ± 0.3 SD ± 5.9 SD
Assimilation efficiency (%) ± 1.5 SD ± 1.4 SD ± 5.9 SD ± 2.2 SD
Assimilation efficiency (%) for chick ± 0.4 SD
Calorific value of fish (kJ g−1 wet mass) ± 10.5 10% ± 9.6 10% ± 10.8 10% ± 10.4 10%
Water temperature at surface (◦ C) ± 0.4 SD ± 0.1 SD ± 0.5 SD ± 0.6 SD
Water temperature at bottom (◦ C) ± 0.1 SD ± 0.1 SD
Energy costs, resting at colony (W kg−1 ) ± 1.9 SD ± 2.6 SD ± 4.2 SD ± 3.8 10%
Energy costs, resting at sea (W kg−1 ) ± 1.2 SD ± 0.2 SD ± 3.3 SD ± 1.0 10%
Energy costs, flying (W kg−1 ) ± 3.1 10% ± 1.9 10% ± 4.8 10% ± 2.4 10%
Energy costs, foraging (W kg−1 ) ± 1.9 10% ± 1.2 SD ± 0.1 10% ± 2.3 10%
DEE of chick (kJ day−1 ) ± 1.9 10% ± 2.9 10% ± 1.0 10% ± 0.5 10%
All parametersa
Variation of mean DFI (%) ± 36.0 ± 32.2 ± 49.7 ± 54.8
179–365 527–992 750–1979 245–721
Absolute range of DFI (g fish day−1 )

Minimum and maximum input values for each parameter were used (see Table 13.1) to compute the individual variation in mean DFI (%). Minimum
and maximum values for all parametersa combined were computed for the most and least demanding situation, which indicates the maximum range of
potential DFI values for the birds.
 Foraging energetics of North Sea birds 201

Energy expenditure of endotherms sustained over a longer time period

is limited by physiological constraints (e.g. digestive capacities, Weiner
(1992)). Hence, if energy expenditure of animals in the wild approaches
such a ceiling, fitness costs may be incurred (e.g. reduced survival). Here
we assumed the metabolic ceiling of 4 × BMR as suggested by Drent and
Daan (1980) for birds raising chicks. In scenario 1 birds increased their for-
aging time spent within a prey patch. In scenario 2 birds made use of a prey
patch at a further distance from the colony, increasing the amount of time
spent flying. Birds flew to a further prey patch and foraged for a longer time
within the prey patch in scenario 3 (both variables raised equally). Finally,
we investigated the effect that feeding on a diet of lower caloric density
(4.0 kJ g−1 wet mass) had in combination with the above scenarios.

Responses relating to time and energy

If seabirds are to increase their foraging effort in response to a reduction
in sandeel availability, the first constraint encountered is likely to be the
availability of spare time. Time–activity budget analysis illustrated that, with
the exception of the shag, no species could reduce their resting time at the
colony much further, unless they left their chicks unattended. Doing so
could potentially reduce their breeding success drastically, especially when
the chicks are small. While non-attendance of chicks has been recorded in
all four species (Harris & Wanless 1997, Daunt 2000, Lewis et al. 2004), we
assumed that birds normally avoided this. All species except the shag, how-
ever, spent a considerable amount of time resting at sea. In a first step then,
birds are predicted to reduce their resting time at sea to a minimum before
starting to reduce their resting time at the colony. Based on the time–activity
budgets, birds could potentially reallocate between 10% (kittiwake) and 22%
(guillemot) of their daily time towards an increase in foraging effort.
Increasing foraging effort in these scenarios led to an increase in the
amount of required daily food (lower panels in Figs 13.1 to 13.3; see also
Box 13.2). This was especially drastic in scenario 2, where birds made use
of a prey patch at a greater distance from the colony, requiring longer
flight times. The exact relationship depended on the strategy being pur-
sued (commuting to a further prey patch, foraging for longer in a par-
ticular prey patch or a combination of both) and on the specific costs of
the associated activities (e.g. flight versus diving). It also depended on the
benefits accruing as a result of the increased effort. Figures 13.1 to 13.3
(upper panels; see also Box 13.2) clearly underline the limited possibili-
ties for gannets to increase their foraging effort because of energetic con-
straints. Gannets spent about 30% of their daily time resting at sea, of which
202 M. R. Enstipp et al.

Box 13.2 Increasing foraging effort to buffer reduced sandeel

availability: three scenarios
The upper panels in Figs 13.1 to 13.3 plot the daily energy expendi-
ture for the four species (as multiples of BMR) against the increase
in foraging effort considered in scenarios 1 to 3. The solid upper line
indicates the presumed metabolic ceiling of 4 × BMR. Increases in
foraging effort that lead to an increase in energy expenditure beyond
this ceiling are assumed to be unsustainable, indicating a physio-
logical constraint. The x-axis indicates the scope that the bird may
have to reallocate time towards an increase in foraging effort, with
zero being the ‘baseline situation’ before increasing foraging effort.
If the plot for an individual bird stops before reaching the physio-
logical ceiling this indicates a time constraint because birds have no
time left to increase their foraging effort. The percentages given indi-
cate the relative increase in foraging effort that is possible before a
constraint is reached. The lower panels in Figs 13.1 to 13.3 indicate
changes in the required DFI (g day−1 ) that accompany the increase
in foraging effort considered in scenarios 1 to 3.

17% could potentially be allocated towards increasing their foraging effort

(taking into account the inactive period at night). Since gannets already
worked close to the presumed maximal energetic capacity, however, they
could only increase their foraging efforts in very small increments before
reaching the assumed metabolic ceiling in any of the three scenarios. Birt-
Friesen et al. (1989) also reported high metabolic rates for northern gannets
during chick rearing and attributed these to the high costs of thermoregu-
lation and flapping flight. The gannets in our study were already working at
a much higher level than the other birds considered here, hence the phys-
iological limitations to an increase in foraging effort were not surprising.
One possible strategy for gannets, which is not explored in this analysis
since it assumes a balanced energy budget, could be that they incur an
energy debt over a short period which is paid off at a later time. In fact,
Nelson (1978) suggested that body condition declines in gannets over the
course of the breeding season. A possible explanation for the high FMR
values we calculated for the gannets could be the large size of the colony
from which our data were collected. Foraging-trip duration and foraging
range of gannets nesting at the Bass Rock colony are high when compared
with a colony of smaller size (see Hamer et al. (Chapter 16 in this volume)),
 Foraging energetics of North Sea birds 203

4.2
Energy expenditure (× BMR) 4 x BMR
4.0

3.8
+222%
3.6

+75%
3.4
+111%
3.2
Kittiwake
3.0 Shag
Gannet
2.8 Guillemot
1400
1300
Daily food intake (g day−1)

800

700

600

500

400

300

200
0 50 100 150 200 250 300
−1
Increase in foraging time (min day )
Fig. 13.1 Scenario 1 (increasing foraging time within a prey patch). Zero
indicates the ‘baseline situation’ (i.e. before increasing foraging time). See
Box 13.2 for details.

resulting in higher energy expenditures per foraging trip. Hence, gan-

nets breeding at a smaller colony might not experience the same ener-
getic constraint as the gannets in our study. The other three species had
much more scope (in terms of time and energy) to increase their foraging
effort. While in all three scenarios kittiwakes were ultimately limited by the
amount of time they could reallocate towards an increase in foraging effort
(Figs 13.1 to 13.3), shags and guillemots were mostly constrained by the ener-
getic demand that accompanied such an increase. This difference can be
attributed to the relatively high costs of flapping flight in the latter two
species. However, the overall capacity to increase foraging effort in shags
and guillemots was quite considerable. Shags could potentially increase
204 M. R. Enstipp et al.

Energy expenditure (× BMR)

4.2
4 x BMR
4.0
+9% +106% +190%
3.8

3.6 +69%

3.4

3.2
Kittiwake
3.0 Shag
Gannet
2.8 Guillemot
1400
1300
Daily food intake (g day−1)

900

800

700

600

500

400

300

200
0 25 50 75 100 125 150 175 200

Increase in flight time (min day−1)

Fig. 13.2 Scenario 2 (foraging at a more distant prey patch). Zero indicates the
‘baseline situation’ (i.e. before increasing flight time). See Box 13.2 for details.

their foraging time by about 222%, their flight time by about 106% and
their total active time at sea by about 112%. Comparable values for guille-
mots were 111%, 190% and 65%, respectively. Increased flight times in sce-
nario 2 potentially doubled the foraging range of shags while it tripled that of
guillemots (Table 13.3). Kittiwakes and gannets had the potential to increase
their foraging range by 69% and 9% respectively.

Food responses
Birds are constrained by the amount of food available and by the rates at
which they can acquire food. In many cases we know little about sandeel
abundance in the North Sea but we know even less about the prey-capture
capacities of seabirds and the fish densities they require to forage effectively.
 Foraging energetics of North Sea birds 205

4.2
Energy expenditure (× BMR) 4 x BMR
4.0
+9% +112% +65%
3.8

3.6

3.4 +29%

3.2
Kittiwake
3.0 Shag
Gannet
2.8 Guillemot
1400
1300
Daily food intake (g day−1)

900

800

700

600

500

400

300

200
0 50 100 150 200 250
−1
Increase in active time at sea (min day )

Fig. 13.3 Scenario 3 (foraging at a more distant prey patch and for a longer time
within that prey patch). Zero indicates the ‘baseline situation’ (i.e. before
increasing flight and foraging time). See Box 13.2 for details.

Under conditions of reduced prey availability birds presumably have diffi-

culties finding sufficient food to meet their energy requirements and those
of their chicks. Any additional increase in foraging effort, as suggested by
our scenarios, will lead to an even higher requirement for food. Foraging
effort has to be even greater when birds are forced to feed on prey of lower
energy density. Combining the above scenarios with a diet switch to prey
of lower calorific value (4.0 kJ g−1 wet mass; Fig. 13.4) did not change
the basic outcome of our calculations in terms of time and energy con-
straints. However, it drastically increased the food requirement and the
associated feeding rates for all species. Figure 13.4a illustrates this for sce-
nario 2 in the guillemot (foraging at a distant prey patch) and Fig. 13.4b
206 M. R. Enstipp et al.

(a)

Daily energy expenditure (kJ day −1)

2600 5.0
Guillemot
4.5
2400 4 x BMR

CPUE (g min )
−1
4.0
2200

3.5

2000
3.0

1800
2.5

1600 2.0
0 50 100 150
Increase in flight time (min day−1)

(b)
Daily energy expenditure (kJ day −1)

3200 15
Shag
14
3000
4 x BMR 13

CPUE (g min )
−1
12
2800

11

2600
10

9
2400
8

2200 7
0 50 100 150

Increase in active time at sea (min day−1)

Fig. 13.4 (a) Scenario 2 (foraging at a more distant prey patch), and combined
with feeding on prey of reduced caloric density for the guillemot. (b) Scenario 3
(foraging at a more distant prey patch and for a longer time within that prey
patch), and combined with feeding on prey of reduced caloric density for the
shag. Circles indicate DEE, while squares indicate CPUE. Filled symbols indicate
scenarios 2 and 3, while open symbols indicate combination of the respective
scenario with feeding on less profitable prey. CPUE values are based on the time
spent underwater.
 Foraging energetics of North Sea birds 207

shows scenario 3 in the shag (foraging at a distant prey patch for a longer
period).
How steep the increase in foraging effort will need to be depends on how
energetically expensive the associated activities are. Making use of a prey
patch at a greater distance from the colony, requiring longer flight times,
greatly increased daily food requirements for most species considered here
(Fig. 13.2). There will be a limit, imposed by the food availability, at which a
further increase in foraging effort becomes unsustainable.
Feeding rates reported in the literature (6 to 12 g fish min−1 underwater
for shags, Wanless et al. (1998); 0.5 to 1.3 g fish min−1 at sea for gannets,
Garthe et al. (1999)) are typically within the range or slightly lower than our
estimates for the ‘baseline situation’ (before increasing foraging effort). The
same holds true for DFI values reported for kittiwakes and guillemots. This
could indicate that birds might not be able to achieve the feeding rates that
would be required in the above scenarios when foraging effort is drastically
increased.

Further potential responses

An alternative strategy to increasing foraging effort might be to switch to
exploiting other prey types. Unlike the situation in Shetland where sandeels
are the only small, shoaling forage fish, other prey species are present in
the Firth of Forth area that are potentially available to seabirds (Daan et al.
1990). Dietary information suggests that kittiwakes and shags may be less
able to switch to alternative prey compared with guillemots and gannets. In
the case of kittiwakes this might be exacerbated by its surface-feeding habit
which limits its foraging abilities to prey items at, or close to, the surface
(Lewis et al. 2001).
In addition to switching to other live prey species, seabirds can also
potentially exploit fishery discards. Most fisheries in the North Sea produce
bycatch which is discarded and can be consumed by seabirds. While most
pursuit-diving species tend to ignore these discards, many surface-feeders
readily feed on them. Of the four species considered here, only the kittiwake
and gannet are observed in substantial numbers at fishing boats (Garthe
et al. 1996). However, with the volume of fishery discards in the North Sea
potentially declining (see Votier et al. (2004)), this might not be a sustain-
able option.
This chapter has highlighted the interactions between physiological and
behavioural constraints that condition the different responses of seabirds
in the Firth of Forth area to reduced sandeel availability. While shags and
guillemots may have sufficient time and energy to allow them to increase
their foraging effort considerably, kittiwakes and gannets appear more
208 M. R. Enstipp et al.

constrained by time and energy respectively. Our analysis was relatively

restricted in time and space. Clearly, including activity data from a larger
geographical area and over a longer time period to establish the time–energy
budgets (‘baseline situation’) would be desirable and would minimize any
bias that years of high or low prey availability might introduce. As previ-
ously recognized by Furness and Tasker (2000), consideration of energetic
constraints is essential in order to fully evaluate the capacity of species to
cope with food, particularly sandeel, shortages. Gannets scored low for the
criteria used by Furness and Tasker to establish vulnerability and sensitiv-
ity indices for seabirds in the North Sea and the authors concluded that
this species was generally well buffered against change. In contrast, our
study suggests that during chick rearing, gannets are working at the highest
metabolic level of all species considered and, hence, have the least physio-
logical capacity to increase foraging effort. This indicates that gannets could
potentially be very sensitive to a reduction in sandeel availability. To com-
pensate for their energetically costly life, however, gannets might make use
of a highly profitable foraging niche.

ACKNOWLEDGEMENTS

Most of this work was funded by the European Commission project ‘Inter-
actions Between the Marine Environment, Predators and Prey: Implica-
tions for Sustainable Sandeel Fisheries (IMPRESS; QRRS 2000-30864)’.
We thank Svein-Håkon Lorentsen, Claus Bech, Geir Håvard Nymoen,
Hans Jakob Runde and Magali Grantham for help with the captive work
on shags and guillemots. The Norwegian Animal Research Authority
granted research permits for this study (reference numbers 7/01 and
1997/09618/432.41/ME). Thanks also to the many people who helped with
field work, particularly Stefan Garthe, Mike Harris, Janos Hennicke, Sue
Lewis, Gerrit Peters and Linda Wilson. Scottish Natural Heritage and Sir
Hew Hamilton-Dalrymple granted permission to work on the Isle of May
and the Bass Rock, respectively. Luigi Dall’Antonia and Alberto Ribolini
developed the flight activity sensors used on kittiwakes and provided soft-
ware for data analysis. Thanks to Ian Boyd and two anonymous referees for
improving an earlier version of this chapter.

REFERENCES
Bennet, D. C. & Hart, L. E. (1993). Metabolizable energy of fish when fed to captive
great blue herons (Ardea herodias). Can. J. Zool., 71, 1767–71.
Birt-Friesen, V. L., Montevecchi, W. A., Cairns, D. K. & Macko, S. A. (1989).
Activity-specific metabolic rates of free-living northern gannets and other
seabirds. Ecology, 70, 357–67.
 Foraging energetics of North Sea birds 209

Bryant, D. M. & Furness, R. W. (1995). Basal metabolic rates of North Atlantic

seabirds. Ibis, 137, 219–26.
Cairns, D. K. (1987). Seabirds as indicators of marine food supplies. Biol.
Oceanogr., 5, 261–71.
Cooper, J. (1978). Energetic requirements for growth and maintenance of the cape
gannet (Aves: Sulidae). Zool. Afr., 13, 305–17.
Croll, D. A. & McLaren, E. (1993). Diving metabolism and thermoregulation in
common and thick-billed murres. J. Comp. Physiol. B, 163, 160–6.
Daan, N., Bromley, P. J., Hislop, J. R. G. & Nielsen, N. A. (1990). Ecology of North
Sea fish. Neth. J. Sea Res., 26, 343–86.
Daunt, F. (2000). Age-specific breeding performance in the shag Phalacrocorax
aristotelis. Unpublished Ph.D. thesis, University of Glasgow, Scotland, UK.
Daunt, F., Benvenuti, S., Harris, M. P. et al. (2002). Foraging strategies of the
black-legged kittiwake Rissa tridactyla at a North Sea colony: evidence for a
maximum foraging range. Mar. Ecol. Prog. Ser., 245, 239–47.
Daunt, F., Peters, G., Scott, B., Grémillet, D. & Wanless, S. (2003). Rapid-response
recorders reveal interplay between marine physics and seabird behaviour. Mar.
Ecol. Prog. Ser., 255, 283–8.
Drent, R. H. & Daan, S. (1980). The prudent parent: energetic adjustments in
avian breeding. Ardea, 68, 225–52.
Enstipp, M. R., Grémillet, D. & Lorentsen, S.-H. (2005). Energetic costs of diving
and thermal status in European shags (Phalacrocorax aristotelis). J. Exp. Biol.,
208, 3451–61.
Furness, R. W. & Camphuysen, C. J. (1997). Seabirds as monitors of the marine
environment. ICES J. Mar. Sci., 54, 726–37.
Furness, R. W. & Tasker, M. L. (2000). Seabird–fishery interactions: quantifying
the sensitivity of seabirds to reductions in sandeel abundance, and
identification of key areas for sensitive seabirds in the North Sea. Mar. Ecol.
Prog. Ser., 202, 253–64.
Gabrielsen, G. W., Klaassen, M. & Mehlum, F. (1992). Energetics of black-legged
kittiwake (Rissa tridactyla) chicks. Ardea, 80, 29–40.
Garthe, S., Camphuysen, C. J. & Furness, R. W. (1996). Amounts of discards by
commercial fisheries and their significance as food for seabirds in the North
Sea. Mar. Ecol. Prog. Ser., 136, 1–11.
Garthe, S., Grémillet, D. & Furness, R. W. (1999). At-sea activity and foraging
efficiency in chick-rearing northern gannets Sula bassana: a case study in
Shetland. Mar. Ecol. Prog. Ser., 185, 93–9.
Grémillet, D., Wright, G., Lauder, A., Carss, D. N. & Wanless, S. (2003). Modelling
the daily food requirements of wintering great cormorants: a bioenergetics tool
for wildlife management. J. Appl. Ecol., 40, 266–77.
Harris, M. P. & Wanless, S. (1985). Fish fed to young guillemots, Uria aalge, and
used in display on the Isle of May, Scotland. J. Zool. Lond., 207, 441–58.
(1997). Breeding success, diet, and brood neglect in the kittiwake (Rissa
tridactyla) over an 11 year period. ICES J. Mar. Sci., 54, 615–23.
Hilton, G. M., Ruxton, G. D., Furness, R. W. & Houston, D. C. (2000a). Optimal
digestion strategies in seabirds: a modelling approach. Evol. Ecol. Res., 2,
207–30.
Hilton, G. M., Furness, R. W. & Houston, D. C. (2000b). A comparative study of
digestion in North Atlantic seabirds. J. Avian Biol., 31, 36–46.
210 M. R. Enstipp et al.

Hislop, J. R. G., Harris, M. P. & Smith, J. G. M. (1991). Variation in the calorific

value and total energy content of the lesser sandeel (Ammodytes marinus) and
other fish preyed on by seabirds. J. Zool. Lond., 224, 501–17.
Humphreys, E. M. (2002). Energetics of spatial exploitation of the North Sea by
kittiwakes breeding on the Isle of May, Scotland. Unpublished Ph.D. thesis,
University of Stirling, Scotland, UK.
Lewis, S., Wanless, S., Wright, P. J. et al. (2001). Diet and breeding performance of
black-legged kittiwakes Rissa tridactyla at a North Sea colony. Mar. Ecol. Prog.
Ser., 221, 277–84.
Lewis, S., Hamer, K. C., Money, L. et al. (2004). Brood neglect and contingent
foraging behaviour in a pelagic seabird. Behav. Ecol. Sociobiol., 56, 81–8.
Monaghan, P. (1992). Seabirds and sandeels: the conflict between exploitation and
conservation in the northern North Sea. Biodiversity Conserv., 1, 98–111.
Monaghan, P, Wright, P. J., Bailey, M. C. et al. (1996). The influence of changes in
food abundance on diving and surface-feeding seabirds. In Studies of
High-Latitude Seabirds. 4. Trophic Relationships and Energetics of Endotherms in
Cold Ocean Systems. Canadian Wildlife Service Occasional Paper 91. Ottowa,
Ontario, Canada: Canadian Wildlife Service, pp. 10–19.
Montevecchi, W. A. (1993). Birds as indicators of change in marine prey stocks. In
Birds as Monitors of Environmental Change, eds. R. W. Furness & J. J. D.
Greenwood. London: Chapman and Hall, pp. 217–65.
Nelson, J. B. (1978). The Gannet. London: T. & A. D. Poyser.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, T., Jr (1998). Fishing
down marine food webs. Science, 279, 860–3.
Pedersen, J. & Hislop, J. R. G. (2001). Seasonal variations in the energy density of
fishes in the North Sea. J. Fish Biol., 59, 380–9.
Pennycuick, C. J. (1987). Flight of auks (Alcidae) and other northern seabirds
compared with southern Procellariformes: ornithodolite observations. J. Exp.
Biol., 128, 335–47.
(1989). Bird Flight Performance: A Practical Calculation Manual. Oxford, UK:
Oxford University Press.
Tasker, M. L., Camphuysen, C. J., Cooper, J. et al. (2000). The impact of fishing on
marine birds. ICES J. Mar. Sci., 57, 531–47.
Visser, G. H. (2002). Chick growth and development in seabirds. In Biology of
Marine Birds, eds. E. A. Schreiber & J. Burger. Boca Raton, FL: CRC Press, pp.
439–65.
Votier, S. C., Furness, R. W., Bearhop, S. et al. (2004). Changes in fisheries discard
rates and seabird communities. Nature, 427, 727–30.
Wanless, S., Grémillet, D. & Harris, M. P. (1998). Foraging activity and
performance of shags Phalacrocorax aristotelis in relation to environmental
characteristics. J. Avian Biol., 29, 49–54.
Wanless, S., Finney, S. K., Harris, M. P. & McCafferty, D. J. (1999). Effect of the
diel light cycle on the diving behaviour of two bottom feeding marine birds: the
blue-eyed shag Phalacrocorax atriceps and the European shag P. aristotelis. Mar.
Ecol. Prog. Ser., 188, 219–24.
Weiner, J. (1992). Physiological limits to sustainable energy budgets in birds and
mammals: ecological implications. Trends Ecol. Evol. 7, 384–8.
 14
How many fish should we leave in the sea for
seabirds and marine mammals?
R. W. FURNESS

Harvesting reduces fish populations. Using empirical data and stock-

recruitment models, a biomass limit reference point may be set to maintain
recruitment. This minimum is often <20% of the biomass that would be
present without fishing. However, such a low biomass might be inadequate
to sustain top predators, especially where an exploited fish stock is a key food
for wildlife. Another limit reference point might be devised to account for
food needs of predators. Empirical evidence shows that bioenergetics esti-
mates of the quantities of food required by top predators are inadequate to
set reference points. Top predators such as marine mammals and seabirds
show population declines or breeding failures when food stocks remain far
above the minimum amount these predators need to eat. The density of the
prey field for foraging predators is probably a crucial factor affecting their
foraging performance. Not all species of seabirds or marine mammals are
equally vulnerable to impacts on their populations through food shortage. It
is possible to identify the aspects of predator ecology most likely to require
high densities of food to permit economically profitable foraging. It may be
useful to select for study ‘sensitive species’ of seabirds and marine mam-
mals; this might permit the development – from empirical studies – of ref-
erence points based on the ecology of these species. Such reference points
for ‘sensitive species’ could be used in a precautionary way as proxies to
protect the broad community of dependent wildlife.
Harvesting generally reduces fish stock size. A reduction in spawning-
stock biomass may increase the risk of recruitment failure. A limit refer-
ence point can be set based on spawning-stock biomass to reduce this risk,

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
212 R. W. Furness

by aiming to keep spawning-stock biomass above a threshold level. For

example, the North Pacific Fishery Management Council’s Fishery Man-
agement Plan involves explicit definitions for an over-fishing level (OFL),
and for a minimum stock size threshold (MSST). The OFL is generally
set at a level corresponding to FMSY , the fishing mortality rate associated
with (single-species) maximum sustainable yield (MSY). The MSST is gen-
erally set at one-half of BMSY , the biomass associated with MSY (Goodman
et al. 2002). The Eastern Bering Sea walleye pollock Theragra chalcogramma
stock is managed with the arithmetic mean of the probability density func-
tion for FMSY set as the over-fishing limit. The use of FMSY as a limit
to exploitation rate is consistent with current international best practice
for single-species harvest strategies (e.g. FAO 1995, Jennings et al. 2001).
The generally accepted proxy for BMSY for walleye pollock is B35% and there-
fore the limit reference point is B17.5% . But is a biomass of only 17.5% of the
amount that would be present in the absence of a fishery also an adequate
food abundance level for predators that depend on this fish stock as food?
Or might there be a need to set a limit at a higher biomass level in order
to accommodate the needs of dependent predators and healthy ecosystem
function?
Such a higher limit could be set on a precautionary basis, or by assess-
ment of empirical data from a long-established fishery or from ecological
studies of the relationships between specific predator populations and their
food stocks. In the case of walleye pollock, this species is also a major food
for many top predators, including the Steller sea lion Eumetopias jubatus –
a species that has been listed as ‘Endangered’ under the US Endangered
Species Act. Therefore in this case any impact of the fishery on this top
predator is a matter of particular concern for managers. For the walleye pol-
lock fishery, a precautionary limit has been set at B20% ; if the stock fell below
this threshold the fishery would close to minimize risk to the Steller sea lion.
However, the biomass of pollock required to sustain the Steller sea lion is
unknown, and B20% represents an arbitrary limit reference point, albeit one
that is above that set by single-species assessment considerations alone. Fur-
thermore, it may be availability of age-0 and age-1 pollock, not the biomass
of adult pollock, that is important. Since adult pollock can be cannibalistic,
high adult biomass does not necessarily improve foraging conditions. This
chapter focuses on the North Sea sandeel Ammodytes marinus stock, where a
large industrial fishery and top predators potentially compete for the same
fish. However, at present there is no precautionary biomass limit set for
sandeel management equivalent to the precautionary approach being used
in the pollock fishery. An example of a highly conservative but arbitrary
 How many fish should we leave in the sea? 213

ecological limit reference point is seen in the management of krill within

the Convention on the Conservation of Antarctic Marine Living Resources
(CCAMLR); this sets, on a precautionary basis, a target stock level of B75% ,
since krill is a keystone prey species in the Southern Ocean ecosystem (Con-
stable et al. 2000). A number of recent papers consider the general ques-
tions of defining over-fishing from an ecosystem perspective and taking
into account the needs of top predators (for example, Hollowed et al. 2000,
Murawski 2000, Sainsbury et al. 2000, Boyd and Murray 2001). In this
chapter, I ask the specific question: Can we find evidence from studies of
predator ecology of what biomass limits might be appropriate to set for
specific fish stocks? The chapter also discusses the possibilities of extend-
ing two theoretical approaches that have emerged to help evaluate the ref-
erence point approach with regard to the food needs of top predators in
ecosystems subject to major fishery harvests. The first theoretical approach
is the use of bioenergetics estimates of the amount of food required by top-
predator populations; from these estimates it may be possible to compute
how much fish should be ‘set aside’ for predators to consume rather than be
taken by fisheries. The second approach is the estimation of the prey abun-
dance (or prey density) required by top predators to permit profitable forag-
ing that can sustain successful breeding or can sustain healthy top-predator
populations.

BIOENERGETICS CONSIDERATIONS

Many studies in different marine ecosystems find a consistent tendency for

fish consumption by seabirds and marine mammals to be less than the har-
vest by major fisheries, and considerably less than the consumption of food
fish by predatory fish (Box 14.1). Studies of seabird community bioenerget-
ics tend to show consumption by seabirds of around 1 t km−2 in a range
of seas supporting major fisheries, and somewhat higher consumption in
sub-Antarctic and Antarctic waters; in the North Sea, the same pattern can
be seen (Box 14.1).
It might seem that a limit reference point could be estimated from these
bioenergetics calculations of food requirements. For example, a sandeel
biomass of 350 000 t in the North Sea would appear to provide the amount
required by the populations of seabirds and marine mammals in 1999
(Box 14.1). However, such a calculation assumes that the predators can find
every fish and that predators and prey have the same geographical distri-
butions. In practice, predators require a certain density of food in order to
be able to meet their energy demands. The amount of food they require to
214 R. W. Furness

Box 14.1 Estimates of food consumption by seabirds

‘Food-fish’ consumption
Estimated consumption of ‘food fish’ by major consumer groups in
various large marine ecosystems; data from Bax (1991).

Consumer Minimum (t km−2 ) Maximum (t km−2 )

Predatory fish 5.0 56.0

Fisheries 1.4 6.1
Marine mammals 0.1 5.4
Seabirds 0.1 2.0

Consumption by seabirds
Estimated food consumption by seabirds in various regions.

Consumption
Region (t km−2 ) Reference

Northeast Atlantic 0.5 Furness (1994)

Southeast Bering Sea 1.0 Schneider et al. (1986)
Barents Sea 1.0 Mehlum and Gabrielsen (1995)
Eastern Canada 1.1 Diamond et al. (1993)
Bear Island 1.3 Mehlum and Gabrielsen (1995)
Georges Bank 1.6 Schneider et al. (1987)
Shetland 1.9 Furness (1978)
Prince Edward Island 4.7 Brown (1989)
South Georgia 7.8 Croxall and Prince (1987)

Sandeel consumption
Consumption of sandeels Ammodytes by major consumer groups in
the North Sea (ICES Areas IV a,b,c) in 1969 and 1999; from Furness
(2002).

Consumer 1969 1999

Predatory fish 5 000 000 t 2 000 000 t

Industrial fishery 100 000 t 800 000 t
Marine mammals 80 000 t 150 000 t
Seabirds 150 000 t 200 000 t
 How many fish should we leave in the sea? 215

consume may be much less than the amount that they need to have in the
environment in order to forage effectively.

PREY ABUNDANCE REQUIREMENTS OF TOP PREDATORS

A number of datasets provide indications of the prey density needed by

top predators. This was clearly demonstrated by Boyd and Murray (2001)
in their analysis of 22 years of monitoring data of top predators and krill
around South Georgia. They showed that an index of population ecology –
including aspects of numbers and breeding performance – increased with
krill density following a non-linear asymptotic model. Both marine mam-
mals and seabirds showed improvements in the index score with krill den-
sities up to about 60 g m−2 .
A dramatic decrease in sandeel stock biomass in Shetland in the 1980s
presented an opportunity to examine predator breeding ecology and success
over a large range of sandeel abundances (Klomp & Furness 1992, Phillips
et al. 1996, Caldow & Furness 2000, 2001, Oro & Furness 2002, Ratcliffe
et al. 2002). One striking feature was that even though most seabirds in
Shetland fed predominantly on sandeels while breeding, an 80% reduc-
tion in the Shetland sandeel stock biomass caused breeding failure of some
species, reduced breeding success in other species and had a negligible
impact on a few species (Furness & Tasker 2000). Clearly there were large
species-specific differences in vulnerability to reduced food supply related to
individual species’ ecology. In the Arctic skua Stercorarius parasiticus, breed-
ing success varies from year to year but is correlated across the different
colonies within Shetland, suggesting that all Shetland Arctic skua colonies
are affected by the local sandeel stock abundance in each year (Mavor
et al. 2003). Over years when sandeel stock abundance data for Shetland
were available from survey and VPA, Arctic skua breeding success increased
with sandeel abundance and can be well described by either a sigmoidal or
a logarithmic relationship, with high chick production when sandeel stock
exceeded about 80 000 t (Fig. 14.1). The implication here is that Arctic skuas
require rather a high density of sandeels in the Shetland ecosystem, and
are unlikely to breed successfully when sandeel biomass is below 50 000 t.
The Arctic skua population consumes 66.5 t of sandeels (Box 14.2) but
the empirical data show that the species requires the sandeel stock to be
in excess of 50 000 t in order to achieve successful breeding. Similarly,
kittiwake breeding success at Foula, and generally across Shetland, cor-
related with sandeel abundance; this suggests that a threshold of around
50 000 t is required to permit a breeding success of over 0.5 chicks per nest
216 R. W. Furness

Box 14.2 Relationships between seabird breeding success and

food-fish stock biomass, and the quantity of this food required to
supply the breeding season needs of the bird population
Calculating how much food the birds eat
The Shetland Arctic skua population was around 1950 pairs in
1982–6, the period when the relationship with sandeel abundance
was measured. This equates to about 4000 breeding adults, to
which we can add another 25% to allow for non-breeders and pas-
sage migrants. Arctic skuas arrive in Shetland from their southern-
hemisphere wintering grounds around 1 May, and depart 31 July;
therefore, they are present for about 92 days. Adult Arctic skuas
have a field metabolic rate of 752 kJ day−1 (Hunt & Furness 1996).
In Shetland during this period Arctic skuas fed almost exclusively
on sandeels, which have a calorific value of 6.5 kJ g−1 (fresh mass)
(Hunt & Furness, 1996). Assuming a food utilization efficiency of
80%, we can estimate the consumption of sandeels by Arctic skuas
at Shetland as:
5000 birds × 92 days × 752 kJ day−1 × 1/0.8 = 4.3 × 108 kJ
= 66.5 t

of sandeels.

(Figs. 14.2 and 14.3), yet (following an equivalent calculation to that above)
Shetland kittiwakes consumed less than 1% of this amount. Similar log-
arithmic or sigmoidal relationships between predator local abundance or
breeding success and food fish abundance have been reported for guille-
mot species (Mehlum et al. 1999, Fauchald et al. 2000, Fauchald & Erik-
stad 2002), for Atlantic puffins Fratercula arctica (Durant et al. 2003), for
least auklets Aethia pusilla (Hunt 1997), for kelp bass Paralabrax clathra-
tus (Anderson 2001) and for basking sharks Cetorhinus maximus (Sims
2000).
Another example where predator performance can be related to fish
stock biomass is the case of the common guillemot Uria aalge in the Barents
Sea, where this species feeds almost exclusively on capelin Mallotus villosus
(Barrett & Krasnov 1996). Food requirements of common guillemots in the
Barents Sea have been estimated at 70 000 t per year (Mehlum & Gabrielsen
1995). The capelin stock biomass was around 6 million tonnes in 1980, but
fell rapidly in the mid 1980s to about 500 000 t in 1985. In 1985–7 over 90%
 How many fish should we leave in the sea? 217

Arctic skua breeding success at Foula 1.4

1.2

1.0

0.8

0.6

0.4
y = 0.5428 ln(x) − 5.2379
0.2 R 2 = 0.8032
0
0 50000 100000 150000 200 000
−0.2
Shetland sandeel total stock biomass (tonnes)

Fig. 14.1 Breeding success of Arctic skuas in Foula, Shetland from 1975 to
1994, in relation to the estimated abundance of sandeels at Shetland (VPA
estimate of total stock biomass in tonnes; data from ICES). The fitted line is a
logarithmic regression.
Kittiwake breeding success at Foula

1.6
1.4
1.2
1.0
0.8
0.6
0.4 y = 0.4735 ln(x ) − 4.3601
0.2 R 2 = 0.602
0
0 50 000 100 000 150 000 200 000
Shetland sandeel total stock biomass (tonnes)

Fig. 14.2 Breeding success of kittiwakes in Foula, Shetland from 1975 to 1994,
in relation to the estimated abundance of sandeels at Shetland (VPA estimate of
total stock biomass in tonnes; data from ICES). The fitted line is a logarithmic
regression.

of common guillemots in the Barents Sea died, apparently of starvation in

winter (Barrett & Krasnov 1996). So a critical abundance of capelin for com-
mon guillemot survival was in excess of 500 000 t despite the fact that the
population only needed to consume about 12% of this amount.

CONCLUSION

It seems clear that calculating the amount of fish eaten by top predators
does not provide a useful approach to defining the minimum stock biomass
necessary to support these top-predator populations. The seabird examples
218 R. W. Furness

1.2

Kittiwake breeding success

1.0

0.8
Shetland
0.6

0.4
y = 0.3673 ln(x) − 3.4806
0.2
R 2 = 0.7351
0
0 20000 40000 60000 80000 100000 120000 140000
Shetland sandeel total stock biomass (tonnes)

Fig. 14.3 Breeding success of kittiwakes in Shetland colonies monitored by

JNCC from 1986–1994, in relation to the estimated abundance of sandeels at
Shetland (VPA estimate of total stock biomass in tonnes; data from ICES). The
fitted line is a logarithmic regression.

quoted above show that populations require very much larger biomasses of
food fish in order to allow them to forage to obtain the relatively smaller
quantity of food that they need. The few empirical examples available from
the literature suggest that a decrease in food-fish stock biomass to about
20% of ‘normal’ may cause catastrophic reproductive failure in sensitive
seabird species, at least in the short term. However, the impact of such a
large reduction may be very slight for species of low sensitivity. Whether the
concept of species-specific sensitivity based on the ecology of each species
can be extended to other top predators – such as marine mammals and pis-
civorous fish – remains to be seen, but such an approach identifying ‘sen-
tinel’ species of high sensitivity as a proxy to represent the community of
top predators may provide a way to define prey-fish stock biomass (or den-
sity) limits that should provide adequate foraging opportunities for most top
predators in particular ecosystems. This approach has already been put into
effect with the management of the North Sea sandeel fishery; low breed-
ing success of black-legged kittiwakes in southeast Scotland and northeast
England was taken as a signal that sandeel fishing in that area should be
halted due to its effect on populations of dependent species.
It is noteworthy that at least the more sensitive seabird species suffer
reproductive failure at a food-fish stock biomass level that is somewhat
higher than might be set as a limit reference point on the basis of single-
stock assessment to protect recruitment. Reference-point management for
ecosystem sustainability probably cannot simply adopt reference points set
on the basis of fish stock recruitment needs alone. Although this paper
 How many fish should we leave in the sea? 219

has focused entirely on the reference-point approach (but extending this

to considerations of the wider ecosystem), it should be borne in mind that
a reference-point approach cannot take account of many ecological interac-
tions between fisheries and top predators. These interactions can be com-
plex. For example, some species of seabirds have been able to take advan-
tage of fishery discards and offal as a novel food source that they could not
access in the absence of fisheries (Garthe et al. 1996). Recent reductions in
stocks – and consequently catches – of cod, haddock Melanogrammus aeglefi-
nus and whiting Merlangius merlangus in the northwest North Sea have led
to smaller quantities of fish being discarded in that region in recent years
(Reeves & Furness 2002). Scavenging seabirds have switched diet to com-
pensate for this and have increased predation rates on some other seabirds
such as black-legged kittiwakes (Phillips et al. 1997, Bearhop et al. 2001,
Oro & Furness 2002, Votier et al. 2004). A reference-point limit on sandeel
fishing in areas of the North Sea to ensure adequate sandeel biomass for
good breeding success by a sensitive species such as the black-legged kit-
tiwake would not necessarily ensure that black-legged kittiwake popula-
tion size remained stable; increased predation due to reduced discarding
could lead to large declines in kittiwake numbers even if breeding success
remained high (Furness 2003). A reference-point approach aimed at tak-
ing account of the wider ecosystem and food needs of top predators may
be appropriate, but would not necessarily ensure that the populations of
those top predators would remain stable, and reference points might not
remain appropriate over extended periods of changing ecological relation-
ships. This leaves managers in a difficult position, and the way forward may
be to collect empirical data to describe these relationships to see how con-
sistent an approach this represents.

ACKNOWLEDGEMENTS

I thank the European Commission for funding the project ‘DISCBIRD’ and
the Natural Environment Research Council and the Shetland Oil Termi-
nal Environmental Advisory Group (SOTEAG) for funding research into
seabird ecology in Shetland over much of the period 1971–2003. I am grate-
ful to Mark Tasker for helpful comments on an earlier draft of this chapter.

REFERENCES
Anderson, T. W. (2001). Predator responses, prey refuges, and density-dependent
mortality of a marine fish. Ecology, 82, 245–57.
220 R. W. Furness

Barrett, R. T. & Krasnov, J. V. (1996). Recent responses to changes in fish stocks of

prey species by seabirds breeding in the southern Barents Sea. ICES J. Mar.
Sci., 53, 713–22.
Bax, N. J. (1991). A comparison of fish biomass flow to fish, fisheries, and
mammals in six marine ecosystems. ICES Mar. Sci. Symp., 193, 217–24.
Bearhop, S., Thompson, D. R., Phillips, R. A. et al. (2001). Annual variation in
great skua diets: the importance of commercial fisheries and predation on
seabirds revealed by combining dietary analyses. Condor, 103, 802–9.
Boyd, I. L. & Murray, W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Brown, C. R. (1989). Energy requirements and food consumption of Eudyptes
penguins at the Prince Edward Islands. Antarct. Sci., 1, 15–21.
Caldow, R. W. G. & Furness, R. W. (2000). The effect of food availability on the
foraging behaviour of breeding great skuas Catharacta skua and Arctic skuas
Stercorarius parasiticus. J. Avian Biol., 31, 367–75.
(2001). Does Holling’s disc equation explain the functional response of a
kleptoparasite? J. Anim. Ecol., 70, 650–62.
Constable, A. J., de la Mare, W. K., Agnew, D. J., Everson, I. & Miller, D. (2000).
Managing fisheries to conserve the Antarctic marine ecosystem: practical
implementation of the Convention on the Conservation of Antarctic Marine
Living Resources (CCAMLR). ICES J. Mar. Sci., 57, 778–91.
Croxall, J. P. & Prince, P. A. (1987). Seabirds as predators on marine resources,
especially krill, of South Georgia waters. In Seabirds: Feeding Ecology and Role in
Marine Ecosystems, ed. J. P. Croxall. Cambridge, UK: Cambridge University
Press, pp. 347–68.
Diamond, A. W., Gaston, A. W. & Brown, R. G. B. (1993). Studies of High-Latitude
Seabirds. 3. A Model of the Energy Demands of the Seabirds of Eastern and Arctic
Canada, ed. W. A. Montevecchi. Canadian Wildlife Service Occasional Paper
77. Ottowa, Ontario, Canada: Canadian Wildlife Service.
Durant, J. M., Anker-Nilssen, T. & Stenseth, N. C. (2003). Trophic interactions
under climate fluctuations: the Atlantic puffin as an example. Proc. R. Soc.
Lond. B, 270, 1461–6.
FAO (Food and Agriculture Organization) (1995). Code of Conduct for Responsible
Fisheries. Rome, Italy: FAO.
Fauchald, P. & Erikstad, K. E. (2002). Scale-dependent predator–prey interactions:
the aggregative response of seabirds to prey under variable prey abundance and
patchiness. Mar. Ecol. Prog. Ser., 231, 279–91.
Fauchald, P., Erikstad, K. E. & Skarsfjord, H. (2000). Scale-dependent
predator–prey interactions: the hierarchical spatial distribution of seabirds and
prey. Ecology, 81, 773–83.
Furness, R. W. (1978). Energy requirements of seabird communities: a
bioenergetics model. J. Anim. Ecol., 47, 39–53.
(1994). An estimate of the quantity of squid consumed by seabirds in the eastern
North Atlantic and adjoining seas. Fish. Res., 21, 165–77.
(2002). Management implications of interactions between fisheries and
sandeel-dependent seabirds and seals in the North Sea. ICES J. Mar. Sci., 59,
261–9.
(2003). Impacts of fisheries on seabird communities. Sci. Mar., 67 (Suppl. 2),
33–45.
 How many fish should we leave in the sea? 221

Furness, R. W. & Tasker, M. L. (2000). Seabird–fishery interactions: quantifying

the sensitivity of seabirds to reductions in sandeel abundance and
identification of key areas for sensitive seabirds in the North Sea. Mar. Ecol.
Prog. Ser., 202, 253–64.
Garthe, S., Camphuysen, C. J. & Furness, R. W. (1996). Amounts of discards by
commercial fisheries and their significance as food for seabirds in the North
Sea. Mar. Ecol. Prog. Ser., 136, 1–11.
Goodman, D., Mangel, M., Parkes, G. et al. (2002). Scientific Review of the Harvest
Strategy Currently Used in the BSAI and GOA Groundfish Fishery Management
Plans. Report prepared for North Pacific Fishery Management Council,
November 2002. Anchorage, Alaska: North Pacific Fishery Management
Council.
Hollowed, A. B., Ianelli, J. N. & Livingston, P. (2000). Including predation
mortality in stock assessments: a case study for Gulf of Alaska walleye pollock.
ICES J. Mar. Sci., 57, 279–293.
Hunt, G. L. (1997). Physics, zooplankton, and the distribution of least auklets in
the Bering Sea: a review. ICES J. Mar. Sci., 54, 600–7.
Hunt, G. L. & Furness, R. W. (1996). Seabird–Fish Interactions, with Particular
Reference to Seabirds in the North Sea. ICES Cooperative Research Report 216.
Copenhagen, Denmark: ICES.
Jennings, S., Kaiser, M. J. & Reynolds, J. D. (2001). Marine Fisheries Ecology.
Oxford, UK: Blackwell Science.
Klomp, N. I. & Furness, R. W. (1992). Non-breeders as a buffer against
environmental stress: declines in numbers of great skuas on Foula, Shetland,
and prediction of future recruitment. J. Appl. Ecol., 29, 341–8.
Mavor, R. A., Parsons, M., Heubeck, M., Pickerell, G. & Schmitt, S. (2003). Seabird
Numbers and Breeding Success in Britain and Ireland, 2002. UK Nature
Conservation, No. 27. Peterborough, UK: Joint Nature Conservation
Committee.
Mehlum, F. & Gabrielsen, G. W. (1995). Energy expenditure and food
consumption by seabird populations in the Barents Sea region. In Ecology of
Fjords and Coastal Waters, eds. H. R. Skjoldal, C. Hopkins, K. E. Erikstad & H.
P. Leinaas. Amsterdam, the Netherlands: Elsevier Science, pp. 457–70.
Mehlum, F., Hunt, G. L., Jr, Klusek, Z. & Decker, M. B. (1999). Scale-dependent
correlations between the abundance of Brunnich’s guillemots and their prey.
J. Anim. Ecol., 68, 60–72.
Murawski, S. A. (2000). Definitions of overfishing from an ecosystem perspective.
ICES J. Mar. Sci., 57, 649–58.
Oro, D. & Furness, R. W. (2002). Influences of food availability and predation on
survival of kittiwakes. Ecology, 83, 2516–28.
Phillips, R. A., Caldow, R. W. G. & Furness, R. W. (1996). The influence of food
availability on the breeding effort and reproductive success of Arctic skuas
Stercorarius parasiticus. Ibis, 138, 410–9.
Phillips, R. A., Catry, P., Thompson, D. R., Hamer, K. C. & Furness, R. W. (1997).
Inter-colony variation in diet and reproductive performance of great skuas
Catharacta skua. Mar. Ecol. Prog. Ser., 152, 285–93.
Ratcliffe, N., Catry, P., Hamer, K. C., Klomp, N. I. & Furness, R. W. (2002). The
effect of age and year on the survival of breeding adult great skuas Catharacta
skua in Shetland. Ibis, 144, 384–92.
222 R. W. Furness

Reeves, S. A. & Furness, R. W. (2002). Net Loss – Seabirds Gain? Implications of

Fisheries Management for Seabirds Scavenging Discards in the Northern North Sea.
Sandy, UK: RSPB.
Sainsbury, K. J., Punt, A. E. & Smith, A. D. M. (2000). Design of operational
management strategies for achieving fishery ecosystem objectives. ICES J.
Mar. Sci., 57, 731–41.
Schneider, D. C., Hunt, G. L., Jr & Harrison, N. M. (1986). Mass and energy
transfer to seabirds in the southeastern Bering Sea. Cont. Shelf Res., 5, 241–57.
Schneider, D. C., Hunt, G. L. & Powers, K. D. (1987). Energy flux to pelagic birds: a
comparison of Bristol Bay (Bering Sea) and Georges Bank (Northwest Atlantic).
In Seabirds: Feeding Ecology and Role in Marine Ecosystems ed. J. P. Croxal.
Cambridge, UK: Cambridge University Press, pp. 259–77.
Sims, D. W. (2000). Can threshold foraging responses of basking sharks be used
to estimate their metabolic rate? Mar. Ecol. Prog. Ser., 200, 289–96.
Votier, S. C., Furness, R. W., Bearhop, S. et al. (2004). Changes in fisheries discard
rates and seabird communities. Nature, 427, 727–30.
 15
Does the prohibition of industrial fishing for
sandeels have any impact on
local gadoid populations?
S. P. R. GREENSTREET

Industrial fisheries remove large quantities of small fish from the North Sea
ecosystem each year. Since these small fish constitute the prey of marine top
predators, such activities are considered to pose a potential threat to marine
food-web dynamics. The risk to seabird and marine mammal communities
has in the past received most attention, but more recently concern has been
expressed regarding the possible consequences of industrial fishing for pis-
civorous fish populations, often the target of fisheries for human consump-
tion. These concerns are addressed in this chapter. A major industrial fish-
ery for sandeels opened on the Wee Bankie in the northwestern North Sea in
the early 1990s. Subsequently, in 2000, this fishery was closed in response
to concern over its possible impact on local seabird populations. The effect
of this closure on the abundance of sandeels in the area – and on local gadoid
population abundance, diet, food consumption rates and body condition –
are described to examine the effects of the sandeel fishery on these pis-
civorous, predatory-fish populations. Although closing the sandeel fishery
resulted in an immediate increase in the local abundance of sandeels, no
beneficial effect on local gadoid populations was detected. Gadoid predators
in the area prey almost entirely on 0-group sandeels (fish‘born’ in the cur-
rent year), while the fishery took predominantly older-aged sandeels. Thus
these two consumers appear not to have directly competed for the same
resource.
Industrial fishing, the catching of fish for industrial purposes, such as
for the production of fishmeal, rather than for human consumption, has

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
224 S. P. R. Greenstreet

taken place in the North Sea for several decades. However, in the mid 1970s
the industrial fishery expanded considerably. At this time the percentage of
the total fish removals from the North Sea that was landed for industrial use
increased from less than 20% to over 50%. This expansion coincided with
increased targeting of sandeels by the fishery (ICES 2002). The industrial
fishery has long been a source of controversy and several different concerns
have been raised. Firstly, the large biomass of fish removed by the fishery
each year has worried many people, over 1.7 million tonnes in 1975 and close
to or exceeding 1 million tonnes every year since (Arnott et al. 2002, ICES
2002). Such catch levels represent the removal by the industrial fishery of
between 10% and 15% of the North Sea fish standing-crop biomass each
year (Yang 1982, Daan et al. 1990, Sparholt 1990). Secondly, in times when
fisheries for many species are in crisis, it is possible that large bycatches of
the juvenile stages of more valuable human-consumption species – such
as herring, cod and haddock – might be included in the catches taken by
the industrial fishery. Fishermen fear that these bycatches reduce recruit-
ment potential, contributing to stock declines and inhibiting recovery in
depleted stocks. Finally, the fish landed by the industrial fisheries consti-
tute the prey of predators higher up the food web. Consequently, there has
been concern over the potential for competition between industrial fisheries
and marine top predators (Avery & Green 1989, Furness 1996, Naylor et al.
2000).
When these concerns come to the fore, the traditional management
response has been to adopt a precautionary approach and to implement
‘local’ industrial-fishery moratoria. Thus in 1983 the winter sprat fishery in
the Moray Firth, northeast Scotland, was closed due to increasing concern
over the growing bycatch of juvenile herring among the fish caught (Hop-
kins 1986). In 1991 the fishery for sandeels around the Shetland Islands, to
the north of Scotland, was closed in an attempt to reverse a worrying down-
ward trend in the breeding success of seabirds in the area (Wright 1996).
Similar concerns were raised when, in the mid 1990s, seabird breeding suc-
cess started to decline at colonies in the Firth of Forth in southeast Scotland,
shortly after a new sandeel fishery commenced on the nearby Wee Bankie
and Marr Bank. In 2000, with no significant indication of any improvement
in the situation, a precautionary stance was taken and fishing for sandeels
along much of the east coast of Scotland was prohibited (Wright et al.
2002).
It is interesting to note that, to date, the top-predator concerns that have
elicited a management response have always involved seabirds. However,
during the 1990s fishermen were also making similar comments to those
 Prohibition of industrial fishing for sandeels 225

being aired by seabird conservationists, but their concerns related to the

threat to predatory-fish populations. They considered that the industrial
fishery for sandeels on the Wee Bankie was depleting prey stocks, in par-
ticular for gadoid predators, species that are heavily dependent on fish prey
(Daan 1989, Hislop 1997, Greenstreet et al. 1998). They feared that compe-
tition with the fishery for sandeel resources would either result in increased
mortality through starvation or cause fish to move away from traditional
grounds, making it more difficult for the smaller fishing vessels to oper-
ate profitably. In addition, dwindling food supplies could lead to reduced
gamete production and affect recruitment. Growth rates and fish condition
(weight at length) may also fall, affecting fish quality and price at the mar-
ket. Declining growth rates – leading to reduced weight at age – may result
in higher than anticipated fishing mortality for a given total allowable catch
(TAC) (stipulated in tonnes), perhaps resulting in the need for even more
stringent TACs in subsequent years.
The fishermen’s concerns were not groundless. Major cod fisheries exist
in Arctic waters across the North Atlantic where capelin Mallotus villosus are
their primary prey. When capelin stocks decline, as a consequence of either
over-fishing or lower recruitment, cod switch to alternative prey. In the
Barents Sea, cod switched to the shrimp Pandalus borealis. Within a year
or so, as a result of the higher predation mortality, the population biomass
of shrimps began to decline. Unable to find sufficient alternative prey,
rates of cannibalism among cod increased and cod food consumption rates
fell. This, combined with a reduction in the quality of the prey consumed
(alternative prey were of a lower energy density than capelin), resulted in
reduced cod growth rates and lowered fecundity. With the reduction in
weight at age, more individual fish were taken in the fishery for a given
TAC than was anticipated by the fishery managers. The cod stock declined
and TACs had to be reduced (Mehl & Sunnanå 1991, Bogstad & Mehl
1997, Bogstad & Gjøsæter 2001). Modelling of the predator–prey interac-
tions of cod and capelin in Icelandic waters suggested that when the capelin
biomass dropped below 2 million tonnes, the consumption of capelin by cod
fell to the point where cod were unable to compensate by consuming alter-
native prey. Total food consumption rates declined causing a reduction in
cod growth rates, stock biomass and fisheries yield (Magnússon & Pálsson
1991). Off southern Labrador and northeastern Newfoundland, cod were
unable to consume sufficient alternative prey to compensate for the decline
in their consumption of capelin in years when capelin biomass was low,
suggesting that a sustained reduction in capelin stock would affect cod pro-
ductivity (Lilly 1991).
226 S. P. R. Greenstreet

Box 15.1 Hypotheses and predictions

Hypotheses
The following hypotheses are implicit in the fishermen’s beliefs:
r Industrial fishing for sandeel depressed sandeel populations in
the Wee Bankie–Marr Bank area.
r Responding to the reduction in prey abundance, populations of
piscivorous-fish predators declined.
r Responding to the reduction in sandeel abundance,
piscivorous-fish predators were forced to switch to alternative
prey.
r The change in diet was not fully compensatory, consequently
overall food intake rates fell.
r As a result of reduced food intake rates, body condition declined
within the populations of piscivorous-fish predators.
Predictions
If these hypotheses are correct then specific predictions regarding
the effect of closing the sandeel fishery can be derived. Following
the closure, one should expect to see:
r an increase in sandeel population size in the area;
r increases in the population sizes of piscivorous-fish predators;
r an increase in the sandeel fraction in the diets of these
predators;
r increased predator food intake rates;
r increased body condition within the populations of
piscivorous-fish predators.

The closure of the sandeel fishery along the Scottish east coast in 2000 pro-
vided an opportunity to explore the validity of the fishermen’s concerns. In
1997, in response to the outcry over declining breeding success at seabird
colonies in the Firth of Forth and elsewhere along the Scottish east coast, a
research programme was initiated to examine interactions between marine
top predators and their sandeel prey. Field studies focused on the nearby
Wee Bankie–Marr Bank region, which in the early 1990s had become a
major sandeel fishing ground. By 2000 therefore, 3 years of data regard-
ing changes in the abundance, distribution and age–length composition
of the sandeel population – and changes in the diet, food consumption
rates and body condition of the main piscivorous fish species populations
in the area – had already been gathered. During this time the sandeel
 Prohibition of industrial fishing for sandeels 227

fishery had been in operation. Subsequently a further 4 years of identi-

cal data were collected while the sandeel fishery moratorium was in force,
allowing the direct effects of the cessation of fishing for sandeels on sandeel
prey populations and their marine top-predator populations to be examined.
This chapter focuses specifically on the trophic interactions between
sandeels and the fish predators in the area and sets out to test some of the
hypotheses implicitly posed by fishermen. These hypotheses, and the main
predictions derived from them, are listed in Box 15.1.

TESTS OF THE HYPOTHESES

Fishing for sandeels off the east coast of Scotland – principally on the Wee
Bankie, Marr Bank and Berwick Bank – commenced in the early 1990s. In
1993 catches from the area exceeded 100 000 t. In 2000 the fishery was
closed to protect breeding seabirds, and since this time fishing removals
(continued for scientific purposes) have not exceeded 5000 t in any year
(Fig. 15.1). The largest proportion of the sandeels removed by the fish-
ery over this entire period has consisted of sandeels of at least 1 year old.
Two separate fisheries-independent assessment methods indicated a con-
siderable increase in the biomass of sandeels present in the Wee Bankie–
Marr Bank–Berwick Bank complex following closure of the sandeel fishery
(Fig. 15.1).
The predominant piscivorous-fish species in the area were cod, haddock
and whiting. Sandeels constituted around 50% of the diet of cod up to 30
cm in length, falling to less than 10% of the diet of cod greater than 45 cm in
length. Approximately 50% of the diet of haddock smaller than 20 cm con-
sisted of sandeels, and this percentage increased to around 75% in haddock
of 45 cm in length. Whiting were fish specialists; 85% of the diet of 15-cm
whiting consisted of sandeels and this percentage declined to around 75% in
whiting of over 30 cm in length (Fig. 15.2). The biomass of all three gadoid
predators declined over the period 1997–2003 (Fig. 15.2). This change in
predator biomass was opposite to the change expected given the increase
in sandeel abundance in the area. Variation in cod and haddock biomass
in the area was strongly correlated with changes in stock size in the entire
North Sea (Fig. 15.2). These two predator populations in the Wee Bankie–
Marr Bank region appeared more strongly influenced by factors affecting
stocks in the whole of the North Sea, such as variability in recruitment and
fishing mortality, than by local changes in management policy regarding
the sandeel fishery.
Annual variation in the proportion of sandeels in the diets of the three
gadoid predators and variation in their daily food consumption rates were
228 S. P. R. Greenstreet

56.50

50-m depth
contour
Marr
Degrees latitude Bank

56.25

Isle of May Berwick

Bank
Firth of Forth Wee Bankie

56.00
−3 −2.75 −2.5 −2.25 −2 −1.75 −1.5 −1.25 −1
Degrees longitude
120000 300 25

Mean trawl density (kg.km−2)

Acoustic estimate (1000 tonnes)
Sandeel landings (tonnes)

Trends in sandeel landings from Trends in the abundance

the Wee Bankie/Marr Bank area of sandeels
20
Acoustic
80000 200 Trawl
15

10
40000 Fishery 100
closed
Fishery 5
closed
0 0 0
1990 1995 2000 2005 1996 1998 2000 2002 2004
Year Year

Fig. 15.1 Chart of the study area showing the main sandbanks where the
sandeel fishery was carried out. The size of the symbol indicates the numbers of
vessels recorded at each location by the Scottish Fisheries Protection Agency in
1996 and 1997. Landings of sandeels taken from the study area are presented
and the period of the moratorium is indicated. Trends in two
fisheries-independent, sandeel-abundance indicators – a demersal trawl survey
(see Fig. 15.2 legend) and an acoustic survey (Mosteiro et al. 2004) – show
changes in the size of the sandeel population within the area over the duration of
the study; again the fishery closure period is shown.

either unrelated to, or negatively correlated with, changes in sandeel abun-

dance in the area. Closure of the sandeel fishery therefore had no beneficial
effect on either of these two feeding parameters (Fig. 15.3). Instead, had-
dock and whiting daily consumption rates were positively correlated with
variation in water temperature close to the seabed (Fig. 15.4). In the light of
these results, it is no surprise that the body condition of cod, haddock and
whiting was also independent of variation in the abundance of sandeels in
the study area and therefore was unaffected by the closure of the sandeel
fishery (Fig. 15.5).
Thus, the fishermen’s contention that the sandeel fishery depressed
sandeel abundance in the Wee Bankie – Marr Bank area does appear to
be supported by the data, yet this seems to have had no perceived knock-
on effects on the gadoid predators in the area. Their population abundance
does not appear to have been limited by sandeel abundance, and there was
 Prohibition of industrial fishing for sandeels 229

Correlation between biomass

Change in diet with increase in predator body length in the study area and biomass
in the whole North Sea
1.0
Cod Haddock Whiting 0.4 r = 0.881, p = 0.020

Log study area abundance

Proportion (by weight)

0.8 0
Diet
0.6 Invertebrates 0.4
Other fish
0.4 Flatfish
0.8 Cod
Gadoids
Clupeids
0.2 1.2
Sandeels
0 Y = 2.85114 X 1.7938
1.6
0 20 40 60 80 10 20 30 40 50 10 15 20 25 30 35 40 0.2 0.4 0.6 0.8
Length (cm) Length (cm) Length (cm) 2.0
r = 0.917, p = 0.010

Log study area abundance

Whiting and haddock biomass (tonnes)

1.8

10 000 300 1.6

Trends in abundance

Cod biomass (tonnes)

1.4 Haddock
Haddock 250
8000
Cod 1.2

Whiting 200 Y = 0.53512 X + 0.3015

1.0
6000
1.6 2.0 2.4 2.8 3.2
150

Log study area abundance

1.95
4000 r = 0.125, p = 0.814
100
1.90

2000 50 Whiting
1.85

0 0 1.80

1996 1998 2000 2002 2004 1.75

Year 1.2 1.28 1.36 1.44

Log North Sea abundance

Fig. 15.2 Nineteen evenly spaced demersal trawl samples were collected in each
year and the catches quantified (number per size class per square meter swept).
These trawl densities were raised to the size of study area to estimate the total
biomass of each species present in each year. Size-stratified (12–14.9, 15–19.9,
20–24.9, 25–29.9, 30–34.9, 35–39.9, 40–49.9 and then every 10 cm thereafter)
sub-samples of the catch of cod, haddock and whiting were dissected for stomach
analysis. Sufficient fish, up to a limit of 25, were examined so as to obtain
between 10 and 15 stomachs with food contents per size class of each predator in
each trawl. Changes in the diet with length of cod, haddock and whiting; trends
in the abundance of these species in the area over the period 1997 – 2003 (note
different Y-axis for cod abundance); and correlations between these study-area
trends and similar trends for the entire North Sea stock are illustrated.

certainly no ‘release’ of their populations when the fishery was closed. The
proportion of sandeels in the diets of the gadoid predators, and their daily
food consumption rates, were not positively related to changes in sandeel
abundance; and there was no increase in either factor when the sandeel fish-
ery was closed. Consequently gadoid-predator body condition also remained
unaffected by the fishery moratorium. Why was this?
The hypotheses posed by the fishermen assume that the industrial fish-
ery and the gadoid predators in the Wee Bankie–Marr Bank region com-
peted directly for the same sandeel resource, but was this actually the case?
230 S. P. R. Greenstreet

100
Cod Haddock
Mann–Whitney p = 0.827 Mann–Whitney p = 0.827
Diet (% sandeel) 80

60
Whiting
40 Mann–Whitney p = 0.827

20

0
Daily consumption (% BM)

3
Cod Haddock
Mann–Whitney p = 0.827 Mann–Whitney p = 0.127
2.5

2
Whiting
1.5 Mann–Whitney p < 0.05

1
Yes No Yes No Yes No
Sandeel fishery in operation

Fig. 15.3 The mean weight of stomach contents was determined for each size
class of each predator in each cruise and Jones’ (1974, 1978) digestion rate model
applied to estimate predator daily rations (g. day−1 ). Daily food consumption
rates as a percentage of predator biomass were determined by dividing daily
rations by predator body weights. The effects of closure of the sandeel fishery on
the percentage (by weight) of sandeels in the diet of each gadoid predator, and on
daily food consumption rates (expressed as a percentage of predator biomass
(BM)), are illustrated.

The sandeel fishery on the Wee Bankie–Marr Bank–Berwick Bank complex

has always taken almost entirely 1-year-old and older sandeels. In contrast,
over 95% of all the sandeels consumed by gadoid predators over the entire
6-year period were less than 8.5 cm in length, putting them firmly in the
0-group age category. Furthermore, closure of the sandeel fishery had no
effect on the size of sandeels taken by the three gadoid species (Fig. 15.6).
Thus the likely explanation as to why closing the sandeel fishery had no
obvious effect on the gadoid predators, is that the fishery and the gadoids
were not after all in direct competition. They both utilized different compo-
nents of the sandeel resource.

CONCLUDING REMARKS

The precautionary closure of fisheries appears to be the usual response of

managers in situations where the fishery in question appears to be having
 Prohibition of industrial fishing for sandeels 231

3
Cod Haddock Whiting
Daily ration (% of predator biomass)

2.5

1.5

Y = 0.2487 × X − 0.3069 Y = 0.1991 × X + 0.6635

R = 0.890 R = 0.689
p = 0.018 p = 0.130
1
7 8 9 10 11 7 8 9 10 11 7 8 9 10 11
Temperature (°C) Temperature (°C) Temperature (°C)

Fig. 15.4 Relationships between gadoid-predator daily food consumption rates

(expressed as a percentage of predator biomass) and average water temperature
in the proximity of the seabed within the study area. Temperature profiles
through the entire water column were determined at each trawl location by
deploying a conductivity, temperature and depth (CTD) profiler.

Cod Haddock Whiting

1.06 1.12 1.07

1.05 1.06
1.10
1.04 1.05
Condition factor

Condition factor

Condition factor

1.08 1.04
1.03
1.03
1.02 1.06
1.02
1.01
1.04 1.01
1.00 1.00
1.02
0.99 0.99
0.98 1.00 0.98
No Yes No Yes No Yes
Sandeel fishing in operation Sandeel fishing in operation Sandeel fishing in operation

Fig. 15.5 Predator body condition (a ratio of weight per unit length) factors were
determined following the methods proposed by Kruuk et al. (1987) and
Thompson et al. (1996). Single weight-at-length relationships (of the form W =
aLb ) were derived by linear regression on all the log-transformed data collected
over the whole 7-year period for each species. Individual condition factors were
determined by dividing observed individual weights by the weight predicted by
the regression. The individual weight data used were fish tissue weight only, i.e.
with all the weight of food in the stomachs and intestines subtracted. The effects
of closure of the sandeel fishery on fish condition factors within the cod, haddock
and whiting populations in the Wee Bankie–Marr Bank study area are illustrated.
232 S. P. R. Greenstreet

Sandeels taken by:

60 1600 800
Cod Haddock Whiting

1200 600
40
Number

800 400

20
400 200

0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Length (cm)

Cod Haddock Whiting

Proportion of sandeels > 8.5 cm

1.2 0.20 0.20

1.0
0.15 0.15
Proportion

Proportion
0.8

0.6 0.10 0.10

0.4
0.05 0.05
0.2

0.0 0.00 0.00

No Yes No Yes No Yes
Sandeel fishery operating Sandeel fishery operating Sandeel fishery operating

Fig. 15.6 Length–frequency distributions of the size of sandeels taken by cod,

haddock and whiting collected over the whole 6-year period of the study, and the
effect of closure of the sandeel fishery on the proportion of sandeel prey taken
that exceeded 8.5 cm in length, for each of the gadoid predators.

a detrimental effect on marine top predators. In order to be confident that

such stringent management action has a reasonable chance of being effec-
tive, a good ‘working knowledge’ of the feeding ecology of the predators
concerned is required. If a fisheries closure is to be effective, both fish-
ery and predators have to utilize the same resource; and for competition
to occur, this resource has to be unable to support both sets of demands.
The effectiveness of the sandeel-fishery moratorium on the Wee Bankie–
Marr Bank as a means of safeguarding seabird predators has still to be fully
assessed. As a means of protecting the food supplies of gadoid predators in
the area, however, the data presented here suggest that prohibiting sandeel
fishing was unlikely to have been of any benefit. Since the fishery concen-
trated on 1-year-old and older sandeels, while the gadoids mainly preyed on
0-group sandeels, the two consumers did not compete directly for the same
resource. Only if the fishery were to limit the supply of 0-group sandeels
would it have any detrimental impact on the feeding potential of the area
for gadoid predators.
 Prohibition of industrial fishing for sandeels 233

Rather than a single well-mixed, ‘global’ population, sandeels in the

North Sea consist of several more or less self-contained meta-populations
(Proctor et al. 1998, Pedersen et al. 1999). One such meta-population, off
the east coast of Scotland, influenced the demarcation of the sandeel fishery-
ban area. If an active sandeel fishery in this area were to reduce the size of
the sandeel spawning stock to the point where future recruitment potential
was compromised, then the fishery could ultimately affect 0-group sandeel
production. Under such circumstances, because all age components in the
stock would be impacted, it is not only the gadoid predators but all marine
top predators utilizing the sandeel resource that would be adversely affected.
However, the evidence presented here suggests that this point had not been
reached in the Wee Bankie–Marr Bank area. Since the population in the area
started to increase in the first year that the fishery was closed, the fishery
had clearly not affected recruitment potential in previous years.
Sandeel recruitment in the North Sea appears to be negatively related
to the biomass of older sandeels (Arnott & Ruxton 2002, Furness 2002),
implying a degree of density-dependent negative feedback – such as compe-
tition for a limiting resource for example. Thus, the increase in the biomass
of 1-year-old and older sandeels in the Wee Bankie–Marr Bank area could
have had a negative influence on the local abundance of 0-group sandeels.
Under such circumstances, continuation of a controlled sandeel fishery in
the area could possibly benefit local gadoid predators. Such a fishery would
need to limit the abundance of sandeels of 1 year of age or more. This mea-
sure could encourage 0-group settlement but would need to be controlled
to prevent over-exploitation of these older sandeels because of the effect that
over-exploitation could have on the food available to top predators, such as
several seabirds and seals (Wanless et al. 1998). Factors affecting variation
in 0-group recruitment to the sandeel population in the area, before and
after the closure of the sandeel fishery, clearly require further investigation.

ACKNOWLEDGEMENTS

Many colleagues at the Marine Laboratory assisted both with the collection
of the data at sea and in its subsequent analysis; particularly Eric Arm-
strong, Iain Gibb, Helen Fraser, Gayle Holland, Cathy Doyle, Rose Li and
Mike Robertson. Conversations with Mike Heath, Peter Wright and others
influenced the direction of the study and I am grateful to Peter Wright for
his comments on the manuscript. I also thank Ian Boyd and two anony-
mous referees for their help in improving the manuscript. These data were
collected as part of two EC projects: The Effect of Large-Scale Industrial
234 S. P. R. Greenstreet

Fisheries on No. Species (ELIFONTS) (95/78 DGXIV) and Interactions

between the Marine Environment, Predator Implications for Sustainable
Sandeel Fisheries (IMPRESS) (Q5RS-2000–30864), and as part of the
Scottish Executive Environment and Rural Affairs Department (SEERAD)-
funded Fisheries Research Services (FRS) project MF0463.

REFERENCES
Arnott, S. A. & Ruxton, G. D. (2002). Sandeel recruitment in the North Sea:
demographic, climatic and trophic effects. Mar. Ecol. Prog. Ser., 238, 199–210.
Arnott, S. A., Ruxton, G. D. & Poloczanska, E. S. (2002). Stochastic dynamic
population model of North Sea sandeels, and its application to precautionary
management procedures. Mar. Ecol. Prog. Ser., 235, 223–34.
Avery, M. & Green, R. (1989). Not enough fish in the sea. New Sci., 1674, 28–9.
Bogstad, B. & Gjøsæter, H. (2001). Predation by cod (Gadus morhua) on capelin
(Mallotus villosus) in the Barents Sea: implications for capelin stock assessment.
Fish. Res., 53, 197–209.
Bogstad B. & Mehl, S. (1997). Interactions between Atlantic cod (Gadus morhua)
and its prey in the Barents Sea. In Proceedings of the International Symposium on
the Role of Forage Fishes in Marine Ecosystems. Alaska Sea Grant College
Program Report AK-SG-97-01. Fairbanks, Alaska: University of Alaska
Fairbanks, pp. 591–616.
Daan, N. (1989). Database Report of the Stomach Sampling Project 1981. ICES
Cooperative Research Report 164. Copenhagen, Denmark: ICES, pp. 1–144.
Daan, N., Bromley, P. J., Hislop, J. R. G. & Nielsen, N. A. (1990). Ecology of North
Sea fish. Neth. J. Sea Res., 26, 343–86.
Furness, R. W. (1996). A review of seabird responses to natural or
fisheries-induced changes in food supply. In Aquatic Predators and Their Prey,
eds. S. P. R. Greenstreet & M. L. Tasker. Oxford, UK:, Blackwell Science, pp.
166–73.
(2002). Management implications of interactions between fisheries and sandeel
dependent seabirds and seals in the North Sea. ICES J. Mar. Sci., 59, 261–9.
Greenstreet, S. P. R., McMillan, J. A. & Armstrong, F. (1998). Seasonal variation in
the importance of pelagic fish in the diet of piscivorous fish in the Moray Firth,
NE Scotland: a response to variation in prey abundance? ICES J. Mar. Sci., 55,
121–33.
Hislop, J. R. G. (1997). Database Report of the Stomach Sampling Project 1991. ICES
Cooperative Research Report 219. Copenhagen, Denmark: ICES, pp. 1–442.
Hopkins, P. J. (1986). Exploited fish and shellfish populations in the Moray Firth.
Proc. R. Soc. Edinburgh, Ser. B, 91, 57–72.
ICES (2002). Report of the Working Group on the Assessment of Demersal Stocks in the
North Sea and Skagerrak. ICES CM 2002, ACFM01. Copenhagen, Denmark:
ICES, pp. 1–555.
Jones, R. (1974). The rate of elimination of food from the stomachs of haddock
Melanogrammus aeglefinus, cod Gadus morhua, and whiting Merlangius
merlangus. J. Cons. Int. Explor. Mer, 35, 225–43.
(1978). Estimates of the food consumption of haddock (Melanogrammus
aeglefinus) and cod (Gadus morhua). J. Cons. Int. Explor. Mer., 38, 18–27.
 Prohibition of industrial fishing for sandeels 235

Kruuk, H., Conroy, J. H. W. & Moorhouse, A. (1987). Seasonal reproduction,

mortality and food of otters (Lutra lutra L.) in Shetland. Symp. Zool. Soc. Lond.,
58, 263–78.
Lilly, G. R. (1991). Interannual variability in predation by cod (Gadus morhua) on
capelin (Mallotus villosus) and other prey off southern Labrador and
northeastern Newfoundland. ICES Mar. Sci. Symp., 193, 133–46.
Magnússon, K. G. & Pálsson, Ó. (1991). Predator–prey interactions of cod and
capelin in Icelandic waters. ICES Mar. Sci. Symp., 193, 153–70.
Mehl, S. & Sunnanå, K. (1991). Changes in growth of Northeast Arctic cod in
relation to food consumption in 1984–1988. ICES Mar. Sci. Symp., 193, 109–12.
Mosteiro, A., Fernandes, P. G., Armstrong, F. & Greenstreet S. P. R. (2004). A
Dual Frequency Algorithm for the Identification of Sandeel School Echotraces. ICES
CM 2004, R12. Copenhagen, Denmark: ICES.
Naylor, R. L., Goldburg, R. J., Primavera, J. H. et al. (2000). Effect of aquaculture
on world fish supplies. Nature, 405, 1017–24.
Pedersen, S. A., Lewy, P. & Wright, P. (1999). Assessments of the lesser sandeel
(Ammodytes marinus) in the North Sea based on revised stock divisions. Fish.
Res., 41, 221–41.
Proctor, R., Wright, P. J. & Everitt, A. (1998). Modelling the transport of larval
sandeels on the north west European shelf. Fish. Oceanogr., 7, 347–54.
Sparholt, H. (1990). An estimate of the total biomass of fish in the North Sea.
J. Cons. Int. Explor. Mer, 46, 200–10.
Thompson, P. M., Tollit, D. J., Greenstreet, S. P. R., Mackay, A. & Corpe, H. M.
(1996). Between year variations in the diet and behaviour of harbour seals
(Phoca vitulina) in the Moray Firth; causes and consequences. In Aquatic
Predators and their Prey, eds. S. P. R. Greenstreet & M. L. Tasker. Oxford, UK:
Blackwell Science, pp. 44 – 52.
Wanless, S., Harris, M. P. & Greenstreet, S. P. R. (1998). Summer sandeel
consumption by seabirds breeding in the Firth of Forth, southeast Scotland.
ICES J. Mar. Sci., 55, 1141–51.
Wright, P. J. (1996). Is there a conflict between sandeel fisheries and seabirds? A
case history at Shetland. In Aquatic Predators and their Prey, eds. S. P. R.
Greenstreet & M. L. Tasker. Oxford, UK: Blackwell Science, pp. 154–65.
Wright, P. J., Jensen, H., Mosegaard, H., Dalskov, J. & Wanless, S. (2002).
European Commission’s Annual Report on the Impact of the Northeast Sandeel
Fishery Closure and Status Report on the Monitoring Fishery in 2000 and 2001.
Yang, J. (1982). An estimate of the fish biomass in the North Sea. J. Cons. Int.
Explor. Mer, 40, 161–72.
 16
Use of gannets to monitor prey availability in the
northeast Atlantic Ocean: colony size, diet and
foraging behaviour
K. C. HAMER, S. LEWIS, S. WANLESS, R. A. PHILLIPS,
T . N . S H E R R AT T , E . M . H U M P H R E Y S , J . H E N N I C K E
AND S. GARTHE

Large seabirds such as northern gannets Morus bassanus have very flex-
ible time–activity budgets; this means that changes in variables such as
foraging-trip duration could provide a rapid indicator of changes in food
supply, an indicator that could not be obtained from smaller species. More-
over, larger birds often have longer foraging ranges, giving them the poten-
tial to integrate information about changes in food availability over large
areas of ocean. There is insufficient information on temporal variation in
fish stocks exploited by far-ranging species to determine how the birds’ for-
aging ecology varies with prey abundance, but comparing colonies of dif-
ferent size potentially presents an opportunity to examine empirically how
foraging ecology varies in relation to prey availability. For gannets in Britain
and Ireland, there were major differences between colonies in foraging and
food-provisioning behaviour, but the relationship between trip duration and
foraging range was remarkably constant. Moreover, there was a strong rela-
tionship between trip duration and the square root of colony size, which was
very similar within colonies between years and between colonies within a
single year. This relationship could provide a powerful tool for gauging the
importance of changes in trip duration in terms of changes in per-capita
prey availability. Over 4 years, annual variation in diet at one colony to some
extent reflected variation in trip durations and foraging locations, although

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Use of gannets to monitor prey availability 237

one year was anomalous and a combination of trip duration and diet pro-
vided a much more complete picture than either did on their own.
Many studies have shown that seabirds are sensitive to changes in food
supply and so have the potential to act as monitors of fish stocks (see reviews
by Montevecchi (1993), and Furness and Camphuysen (1997)). Responses
vary among species but small seabirds such as terns, which spend a high
proportion of available time foraging, are generally considered the most sen-
sitive indicators (Furness & Ainley 1984, Monaghan 1992). Larger birds
have more flexibility in their time–activity budgets and so are often consid-
ered to be less sensitive because they can buffer the impacts of reductions
in food supply on variables such as breeding success and chick growth by
increasing their time spent foraging. However, the very fact that they can do
this means that changes in variables such as foraging-trip duration could
provide a rapid indicator of changes in food supply, which could not be
obtained from smaller species. Moreover, larger birds often have longer for-
aging ranges (e.g. >500 km in northern gannets Morus bassanus, hereafter
termed gannets; Hamer et al. (2000)), giving them the potential to integrate
information about changes in food availability over large areas of ocean.
The main problem to date in using far-ranging seabirds such as gan-
nets as monitors of marine fish stocks is that there is insufficient infor-
mation over sufficiently large spatial scales on temporal variation in fish
stocks exploited as prey, or on how variables other than stock size affect
prey availability. Thus the relationships between foraging-trip duration and
prey availability or abundance are unknown. This is a particular problem
for gannets, which exploit a wide range of species and sizes of prey (Hamer
et al. 2000, Lewis et al. 2003) – each of which would need to be assessed
independently to give a full picture of prey population sizes. Moreover,
while a reduction in prey availability would intuitively be expected to result
in longer foraging trips, this need not be the case. For instance, individuals
of many species, including gannets, spend a proportion of the time at sea
apparently inactive (Hamer et al. 2001, Lewis et al. 2002), and may respond
to a reduction in food supply by reducing this time without any change in
overall trip duration (Enstipp et al. (Chapter 13 in this volume)).
There are currently 21 separate gannet colonies in the British Isles, rang-
ing in size from <150 to >60 000 breeding pairs (Wanless & Harris 2004).
If birds at a colony are competing for food resources, then those at larger
colonies should experience lower per-capita prey availability as a result of
greater competition. If gannets compete for available prey that is randomly
distributed in discrete patches, then they should forage over approximately
the same total area per bird to obtain the same amount of food, irrespec-
tive of colony size. The area covered increases with the square of the mean
238 K. C. Hamer et al.

foraging radius, and so foraging range would be predicted to increase with

the square root of colony size (Lewis et al. 2001). Comparing colonies of dif-
ferent size could therefore present an opportunity to examine empirically
how the foraging ecology of gannets varies in relation to prey availability. In
this chapter we shall: (a) explore how the trip durations and foraging ranges
and behaviour of individuals differ between two colonies of markedly differ-
ent size; (b) investigate how trip duration varies over a wide range of colony
sizes; (c) examine annual variation in trip duration, foraging location and
diet at a single colony in the context of potential changes in per-capita prey
availability.

TRIP DURATIONS, AND FORAGING RANGES AND

BEHAVIOUR AT SEA

One of the largest gannet colonies in Britain is at the Bass Rock, southeast
Scotland (56◦ 6 N, 2◦ 36 W; >40 000 breeding pairs) and one of the small-
est is at Great Saltee, southeast Ireland (52◦ 7 N, 6◦ 37 W; 2000 pairs). We
examined foraging-trip durations and ranges, and at-sea behaviour of birds
at these two colonies in 1998 and 1999 respectively, using satellite teleme-
try (Box 16.1). At the Bass Rock, the mean duration of foraging trips was
31.3 h (SD, ± 11.0h; range, 13.1 to 84.0 h) and the mean distance to desti-
nation was 223 km (SD, ± 95 km; range, 39 to 540 km). Destinations of
foraging trips covered a wide area of ocean encompassing >200 000 km2
within the northwest, west and central North Sea (Fig. 16.1). Mean trip dura-
tion at Great Saltee was significantly shorter than at the Bass Rock (11.9 ±
6.7 h; range, 2.8 to 42.8 h; t-test using mean values for each bird, t12 = 10.5,
p < 0.001) as was the mean distance to destination (89 ± 49 km; range, 14
to 238 km; t12 = 9.5, p < 0.001). Destinations of foraging trips from Great
Saltee encompassed an area of about 45 000 km2 between the coasts of
northwest Wales, southwest England and southern Ireland (Fig. 16.1). This
was about one-quarter of the area covered by birds from the Bass Rock. In
association with this difference in foraging area, individual birds at the Bass
Rock showed a high degree of foraging-area fidelity, with successive trips
having very similar bearings and significant repeatability in distances trav-
elled; in contrast, birds at Great Saltee did not show this behaviour (Hamer
et al. 2001).
There was a highly significant relationship between maximum dis-
tance from the colony and trip duration at both the Bass Rock (F1,67 =
988.7, p < 0.0001, R2 = 0.94) and Great Saltee (F1,58 = 305.4, p < 0.0001,
R2 = 0.84). Average speed ( ± SE) over complete foraging trips was
14.1 ± 0.4 km h−1 at the Bass Rock and 13.8 ± 0.8 km h−1 at Great Saltee
 Use of gannets to monitor prey availability 239

Box 16.1 Satellite telemetry of gannets

Chick-rearing adults at nests with hatching dates ± 2 weeks from the
mode were captured at the nest using a roach pole with a brass noose.
A platform terminal transmitter (PTT; Microwave Telemetry Inc.,
Columbia, MD, USA) weighing 30 g (c. 1% of adult mass) and with
a duty cycle of continuous transmission was then attached with self-
amalgamating tape (RS Components, Northants, UK) to the under-
side of the four central tail feathers, close to the base of the tail with
the aerial pointing upwards through the feathers. This arrangement
minimized drag during flight and prevented tags being displaced
during plunge-diving. Attachment of tags took about 5 minutes and,
after release, every bird returned to the nest almost immediately.
Birds were then tracked for 14 to 23 days each (mean, 16 days), after
which time the tag was removed. Data provided by PTTs were pro-
cessed using the ARGOS facility (CNES, France) and locations of
birds at sea were overlaid on a universal transverse Mercator projec-
tion using Arcview.
Durations of foraging trips were calculated from the time of the
first location after the bird had left the colony until the time of the
first location after it had returned. This was done only for trips with
at least eight locations per day, giving an average error of ± 3 h
for departure and arrival times. Average travel speed during each of
these trips was calculated as twice the slope of the linear regression
of maximum distance from the colony upon trip duration (see Fig.
16.2).
In order to examine movements over shorter intervals within
the total foraging ranges of birds, we estimated travel speeds dur-
ing short sections of each trip as the distance between consecutive
pairs of locations divided by the time elapsed between them. Very
short intervals between locations could produce erroneous estimates
of speed and to avoid this problem, we used only pairs of locations
at sea that were separated by more than 1 h.

(Hamer et al. 2001). Analysis of covariance indicated no difference in speed

between colonies (F1,116 = 0.6, p = 0.4) and this was confirmed by com-
parison of the mean travel speeds for individual birds (Bass Rock, mean =
15.7 km h−1 , n = 9 birds, SD ± 2.5; Great Saltee, mean = 15.0 km h−1 , n =
5, SD ± 2.3; t-test using pooled variance estimate, t12 = 0.52, p = 0.6).
 N

60°N

Bass Rock
58°N

56°N

54°N

52°N

100 km
Great Saltee
50°N

10° W 5°W 0° 5° E 10° E

Fig. 16.1 Foraging ranges and destinations of foraging trips by gannets from the
Bass Rock, southeast Scotland and Great Saltee, southeast Ireland. Open circles,
locations of adults at sea; filled circles, destinations of foraging trips. Reproduced
from Marine Ecology Progress Series, with permission (Hamer et al. 2000).

600

500
Maximum distance (km)

400

300

200

100

0
0 20 40 60 80 100
Trip duration (h)

Fig. 16.2 Maximum distance from the colony and foraging-trip duration at the
Bass Rock (filled circles and continuous line) and at Great Saltee (open circles
and dashed line).
Bass Rock: Maximum distance (km) = 7.05 (SE, ± 0.22) × trip duration (h)
Great Saltee: Maximum distance (km) = 6.88 (SE, ± 0.39) × trip duration (h)
 Use of gannets to monitor prey availability 241

Speeds of travel over intervals within trips were calculated using con-
secutive pairs of locations with >1 h between them (Box 16.1). The mean
( ± SD) of these values at the Bass Rock (18.1 ± 16.6 km h−1 , n = 797)
was very similar to the mean at Great Saltee (17.3 ± 17.2 km h−1 , n = 237)
and there was no difference in the frequency distribution of travel speeds
at the two colonies (Kolmogorov–Smirnov two-sample test, Z = 0.47, n =
1034, p = 0.98). At both the Bass Rock and Great Saltee, the mean ( ± SD)
speed during hours of darkness was very low (1.6 ± 1.9 km h−1 , n = 30
and 4.0 ± 5.4 km h−1 , n = 11 respectively), with much higher speeds dur-
ing daylight (22.3 ± 31.6 km h−1 , n = 767 and 21.8 ± 27.3 km h−1 , n =
226 respectively).
These data indicate that despite the large differences between colonies in
trip durations, distances travelled and foraging-area fidelity, the behaviour
of birds during foraging trips was very similar at the two colonies.
There was, however, a major difference in foraging and food-provisioning
behaviour that was not indicated by these data. At Great Saltee, chicks were
always attended by at least one parent whereas at the Bass Rock, chicks >4
weeks old were sometimes left unattended at the nest while both parents
foraged simultaneously (Lewis et al. 2004). Such trips were only about half
as long on average as attended trips and increased in frequency as chicks
grew, comprising almost 40% of all trips by the time chicks were >9 weeks
old. These unattended trips typically occurred after a longer than average
period of attendance at the nest, and so they are likely to be more frequent at
colonies where long foraging trips are relatively common. Since these unat-
tended trips were much shorter than attended trips, their occurrence could
have influenced the relationship between colony size and mean trip dura-
tion, although this is unlikely: the duration of unattended trips increased
as chicks grew (Lewis et al. 2004) and by the time they were sufficiently
frequent to have a significant impact on overall trip duration, they were of
similar duration to attended trips.

TRIP DURATION, COLONY SIZE AND PER-CAPITA

PREY AVAILABILITY

We recorded trip durations at nine gannet colonies in summer 2000

(Fig. 16.3; Lewis et al. 2001). At each colony, we observed approximately
20 breeding pairs with chicks 5 to 10 weeks old for 10 to 55 h. By recording
the arrival and departure times of adults at these nests we estimated the
overall changeover rate of gannets per nest per day. The mean trip duration
at each colony was then calculated by dividing the mean time available per
242 K. C. Hamer et al.

N N

BR

100 km

Fig. 16.3 Location and size of gannet colonies studied. The area of each circle is
proportional to the colony’s size in 2000 (A, Ailsa Craig; B, Bempton; BR, Bass
Rock; F, Fair Isle; G, Great Saltee; H, Hermaness; I, Ireland’s Eye; N, Noss; T,
Troup Head).

day for foraging (local daylight time minus the time birds were together at
the nest) by the estimated changeover rate. We also obtained historical data
on trip durations at four colonies from personal (S. Wanless, unpublished
data for 1980) and published data (Nelson 1978, Garthe et al. 1999).
Across the nine colonies studied, there was a significant positive corre-
lation between trip duration (hence foraging range) and the square root of
 Use of gannets to monitor prey availability 243

25 175

150
20 BR
125

Foraging range (km)

Trip duration (h)

A
15
100
H
G
H 75
10 N
F A
BR
I 50
B
5
T
25
B

0 0
0 50 100 150 200 250
Square root of colony size

Fig. 16.4 The relationship between foraging-trip duration of gannets in 2000

and the square root of colony size, for nine colonies in Britain and Ireland (see
Fig. 16.3 for colony names and locations). Open circles are earlier data for four
colonies (BR in 1966, B in 1972, A in 1975 and H in 1997). The line of best fit for
2000 data is shown. Reproduced, with permission, from Lewis et al. (2001).

colony size (r = 0.90, d.f. = 7, p < 0.005) with trip duration increasing
three-fold between the smallest and largest colony (Fig. 16.4). Pairs at the
smaller colonies spent significantly more time together at the nest (square-
root-transformed data: r = −0.73, d.f. = 7, p < 0.05), but the rate of
provisioning of chicks by the parents still tended to decrease with increas-
ing population size (square-root-transformed data: r = −0.66, d.f. = 7,
p = 0.052; log-transformed data: r = −0.67, d.f. = 7, p < 0.05). His-
torical data on foraging-trip duration at four of these colonies (Fig. 16.4)
fitted the same regression model well (2000-only regression slope =
0.0492, SE = 0.009 07; pooled within colony regression slope for the four
colonies = 0.0622, SE = 0.0102; combined slope from a linear mixed
model = 0.0548, SE = 0.0128), indicating that the relationship within
colonies between years was similar to the relationship between colonies
within a single year. Data for St Kilda (the largest colony in the North
Atlantic) confirmed the generality of this relationship: in 1980 the colony
held 40 000 breeding sites and the average trip duration was around 21 h
244 K. C. Hamer et al.

Box 16.2 Dietary analysis

To minimize disturbance, most samples were collected from birds
at the periphery of the colony, the majority of which were likely to be
non-breeders. Comparisons of the frequency of occurrence of the
main prey items in these samples with those in samples known
to have come from adults with chicks showed no evidence of any
significant differences in diet between the two groups (all χ 2 tests,
p > 0.05). Each regurgitate was stored separately in a sealed poly-
thene bag and transported to the laboratory where it was weighed (to
the nearest 1g) and the prey were identified either visually or from
sagittal otoliths and vertebrae extracted from the sample (Härkönen
1986, Watt et al. 1997). In addition, sandeels were aged by the
absence (0-group) or presence (older age groups) of annual growth
rings in the otoliths (Anon. 1995).

(S. Wanless, unpublished data). Finally, of two pairs of colonies that were
very similar in size (Great Saltee (G) and Bempton (B); Fair Isle (F) and
Troup Head (T)) one of each pair was above the regression line of trip dura-
tion against colony size, and one was below. The colonies that were above
the regression line (G and F) are both approximately 50 km from other
large colonies, whereas the colonies below the regression line (B and T)
are both over 100 km away from other large colonies. This suggests that
foraging birds from neighbouring colonies may to some extent compete for
food.

ANNUAL VARIATION IN TRIP DURATION, FORAGING

LOCATION AND DIET AT A SINGLE COLONY

Foraging-trip durations were recorded at the Bass Rock every year from
1998 to 2003 except 1999. In four of these years (1998 and 2001–3) we
also assessed the diets of birds at the colony, from regurgitates from adults
(Box 16.2; Hamer et al. 2000, Lewis et al. 2003). In addition, we used satel-
lite telemetry to examine the foraging ranges of birds in 1998, 2002 and
2003. Dietary data were not available for 2000 and so we used data on trip
durations and diets in 2001 as a baseline for comparison with other years.
We first used the regression of trip duration on colony size (Fig. 16.4) to
calculate an equivalent population size for each year i, based on observed
 Use of gannets to monitor prey availability 245

Table 16.1. Foraging-trip duration and range, prey availability index

and diet of gannets at the Bass Rock in different years. Foraging range
was determined by satellite telemetry except in 2001, when range was
calculated from trip duration. See text for calculation of prey
availability index. Diet data are frequency of occurrence (%) of each
category of prey in regurgitates obtained from adults at the colony.

1998 2001 2002 2003

Mean trip duration (h) 32.2 23.1 40.2 23.6

Mean trip range (km) 232 (163) 303 157
Prey availability index 0.43 1 0.25 0.94
Sandeel 29.3 52.2 69.3 73.9
0-group 27.0 48.2 68.4 68.5
Older 2.3 4.0 0.9 5.4
Mackerel 31.6 45.7 21.9 17.2
Clupeidae 44.5 15.3 14.4 41.2
Gadidae 15.0 15.1 15.6 11.9
Others 3.7 9.6 4.5 6.5

Clupeidae were herring Clupea harengus and sprat Sprattus sprattus. Gadidae were
mainly haddock Melanogrammus aeglefinus, whiting Merlangius merlangus and cod
Gadus morhua. Other species were plaice Pleuronectes platessa, salmon Salmo salar,
trout S. trutta, grey gurnard Eutriglia gurnadus, garfish Belone belone, greater fork-
beard Phycis blennoides, dragonet Callionymus lyra and scad Trachurus trachurus.

mean trip durations, if there had been no change between years in prey
abundance or availability. We then calculated an index of relative per-capita
prey availability for each year with dietary information, as the equivalent
population size in year i expressed as a proportion of the population size in
2001 (Table 16.1). This indicated that in comparison with 2001, per-capita
prey availability was less than one-half as high in 1998, only one-quarter as
high in 2002 but very similar in 2003.
Accompanying these differences in trip duration, there was marked vari-
ation in diet between years. In particular, birds made relatively short trips
(about 160 km on average) in 2001 and 2003 and there was a high propor-
tion of sandeel (>50% by frequency) in the diet in both these years (Table
16.1). In 1998, birds made much longer trips (range = 232 km on aver-
age) and the proportion of sandeel in regurgitates was much lower than in
any other year. However, 2002 was apparently a year of very low prey avail-
ability judging from foraging-trip durations, yet the proportion of sandeel
in the diet in 2002 was more than twice that in 1998 and similar to that
in 2003, when the index of per-capita prey availability was >3 times that
246 K. C. Hamer et al.

in 2002 (Table 16.1). The mean foraging range in 2002 (303 km) was the
longest yet recorded for gannets, with the majority of trips heading north-
east of the colony and some extending as far as Bergen Bank, southwest
Norway (580 km from the Bass Rock). Hence despite the presence of a high
proportion of sandeel in the diet, sandeel availability close to the colony was
evidently low in 2002. That was also the only year when neither mackerel
nor Clupeidae (herring and sprats) made a major contribution (>25% by
frequency) to the diet (Table 16.1). In all four years, the majority of sandeels
in the diet (94% ± to 99%) were 0-group (Table 16.1).

DISCUSSION

There were clear differences in foraging and food-provisioning behaviour

between Great Saltee and the Bass Rock. In addition to making much longer
trips on average, birds breeding at the Bass Rock showed a high degree of
foraging-area fidelity whereas birds at Great Saltee did not. This may have
been a consequence of the difference in foraging ranges at the two colonies
but more probably resulted from a more uniform or less predictable distri-
bution of prey in the Celtic Sea than in the North Sea (Hamer et al. 2001).
Adults left chicks unattended only at the Bass Rock and this was a conse-
quence of the longer trip durations at the Bass Rock, because adults only left
chicks unattended after a long period of attendance at the nest. Unattended
chicks were frequently and severely attacked by adult conspecifics – proba-
bly non-breeders in search of nest sites – and this, together with increasing
food-requirements of chicks, presumably explains why the frequency and
duration of unattended trips increased as chicks grew older and so were
less vulnerable to being injured or killed.
Despite these major differences between colonies in foraging and food-
provisioning behaviour, the relationship between trip duration and for-
aging range at the two colonies was remarkably similar. Moreover, in a
comparison across nine different colonies, there was a strong relationship
between trip duration and the square root of colony size, which was very
similar within colonies between years and between colonies within a sin-
gle year. Seabirds appear to contribute relatively little to overall fish mortal-
ity (Furness & Tasker 1997) and the longer trips at larger colonies proba-
bly resulted from density-dependent disturbance of fish shoals rather than
differences in prey depletion (Lewis et al. 2001). Hence differences in trip
duration, whether between colonies or between years at a single colony,
could have reflected differences in prey availability as much as abundance.
Nonetheless, the relationship between population size and trip duration
 Use of gannets to monitor prey availability 247

could provide a powerful tool for gauging the importance of changes in trip
duration in terms of changes in per-capita prey availability. Further data are
now needed to determine the efficacy of this approach for other species and
different marine ecosystems. Annual variation in diet at the Bass Rock to
some extent reflected variation in trip durations, although 2002 was anoma-
lous and a combination of trip duration and diet provided a much more
complete picture than either did on their own.

ACKNOWLEDGEMENTS

Sincere thanks to Tracey Begg, Alan Bull, Jenny Bull, Francis Daunt, Cather-
ine Gray, Mike Harris, Chas Holt, Micky Maher, Linda Milne, Diana de
Palacio, Kelly Redman and Chris Roger for recording changeover rates at
gannet colonies. Thanks to Sir Hew Hamilton-Dalrymple, the Marr family,
Declan Bates, Paul Harvey (Scottish Natural Heritage), Oscar Merne and
Alyn Walsh (Duchas), the Neale family, Deryk Shaw (Fair Isle Bird Obser-
vatory Trust), Martin and Lynda Smyth, Bernie Zonfrillo, the Royal Soci-
ety for the Protection of Birds and the Scottish Seabird Centre for logis-
tical support. This work was funded by Natural Environment Research
Council CASE studentship GT4/99/55 and by grants from the European
Union (projects CEC 96-079 and Q5RS-2000-30864) and the Joint Nature
Conservation Committee (project F90-01-154).

REFERENCES
Anon. (1995). Review of sandeel biology. In Report of the ICES Workshop on Sandeel
Otolith Analysis. ICES CM 1995/9: 4. Copenhagen, Denmark: ICES.
Furness, R. W. & Ainley, D. G. (1984). Threat to Seabird Populations Presented by
Commercial Fisheries. ICBP Technical Publication 2. Cambridge, UK: ICBP,
pp. 701–708.
Furness, R. W. & Camphuysen, C. J. (1997). Seabirds as monitors of the marine
environment. ICES J. Mar. Sci., 54, 726–37.
Furness, R. W. & Tasker, M. L. (1997). Seabird consumption in sand lance MSVPA
models for the North Sea, and the impact of industrial fishing on seabird
populations. In Marine Ecosystems. Proceedings of the International Symposium on
the Role of Forage Fishes in Marine Ecosystems: Alaska Sea Grant College
Program Report Ak-SG-97–01. Fairbanks, Alaska: University of Alaska
Fairbanks, pp. 147–69.
Garthe, S., Gremillet, D. & Furness, R. W. (1999). At-sea activity and foraging
efficiency in chick-rearing northern gannets Sula bassana: a case study in
Shetland. Mar. Ecol. Prog. Ser., 185, 93–9.
Hamer K. C., Phillips, R. A., Hill, J. K., Wanless, S. & Wood, A. G. (2001).
Contrasting foraging strategies of gannets Morus bassanus at two North Atlantic
248 K. C. Hamer et al.

colonies: foraging trip duration and foraging area fidelity. Mar. Ecol. Prog. Ser.,
224, 283–90.
Hamer, K. C., Phillips, R. A., Wanless, S., Harris, M. P. & Wood, A. G. (2000).
Foraging ranges, diets and feeding locations of gannets in the North Sea:
evidence from satellite telemetry. Mar. Ecol. Prog. Ser., 200, 257–64.
Härkönen, T. (1986). Guide to the Otoliths of the Bony Fishes of the Northeast Atlantic.
Hellerup, Denmark: Danbiu ApS.
Lewis, S., Sherratt, T. N., Hamer, K. C. & Wanless, S. (2001). Evidence for
intra-specific competition for food in a pelagic seabird. Nature, 412, 816–19
Lewis, S., Benvenuti, S., Dall’Antonia, L. et al. (2002). Sex-specific foraging
behaviour in a monomorphic seabird. Proc. R. Soc. Lond. B, 269, 1687–93.
Lewis, S., Sherratt, T. N., Hamer, K. C., Harris, M. P. & Wanless, S. (2003).
Contrasting diet quality of Northern Gannets at two colonies. Ardea, 91,
167–76.
Lewis, S., Hamer, K. C., Money, L. et al. (2004). Brood neglect and contingent
foraging behaviour in a pelagic seabird. Behav. Ecol. Sociobiol., 56, 81–8.
Monaghan, P. (1992). Seabirds and sandeels: the conflict between exploitation and
conservation in the northern North Sea. Biodiversity Conserv., 1, 98–111.
Montevecchi, W. A. (1993). Birds as indicators of change in marine prey stocks. In
Birds as Monitors of Environmental Change, eds. R. W. Furness & J. J. D.
Greenwood. London: Chapman and Hall, pp. 217–65.
Nelson, J. B. (1978). The Sulidae: Gannets and Boobies. Oxford, UK: Aberdeen
University Press.
Wanless, S. & Harris, M. P. (2004). Northern gannet Morus bassanus. In Seabird
Populations of Britain and Ireland. Results of the Seabird 2000 Census
(1998–2002) eds. P. I. Mitchell, S. F. Newton, N. Ratcliffe & T. E, Dunn.
London: T. & A. D. Poyser, pp. 115–127.
Watt, J., Pierce, G. J. & Boyle, P. R. (1997). Guide to the Identification of North Sea
Fish using Premaxillae and Vertebrae. ICES Co-operative Research Report 220.
Copenhagen, Denmark: ICES, pp. 1–231.
 17
Population dynamics of Antarctic krill Euphausia
superba at South Georgia: sampling with
predators provides new insights
K . R E I D , E . J . M U R P H Y, J . P . C R O XA L L
A N D P . N . T R AT H A N

Variability in the Southern Ocean is often characterized by fluctuations in

the distribution and abundance of a single dominant zooplankton species,
Antarctic krill Euphausia superba. The ability to sample krill in the diet
of predators at temporal scales not available using conventional (i.e. ship-
based) sampling methods has provided the basis for a re-evaluation of the
role of high rates of growth and mortality, as well as recruitment variabil-
ity, in generating variability in krill abundance at South Georgia. In addi-
tion, the use of a consistent index of krill population size composition from
the diet of predators at South Georgia over the past decade has provided
evidence for a relationship between sea-surface temperature and the level
of krill recruitment. Predators that depend on krill not only show distinct
behavioural responses to changes in krill abundance but also provide dietary
data that help us to understand the mechanisms underlying the popula-
tion dynamics of krill. Where the diet of predators includes commercial
prey species, they can provide information on the key life-history variables
of these species that are fundamental to reducing uncertainty in fisheries
management models.

VARIABILITY IN THE SOUTHERN OCEAN

Understanding the causes and consequences of natural variability in

marine ecosystems is a prerequisite to determining the nature and extent of

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
250 K. Reid et al.

changes of anthropogenic origin and is a central component of ecosystem-

based approaches to fisheries management. Variability in the Southern
Ocean is often characterized by fluctuations in the distribution and abun-
dance of a single dominant zooplankton species, Antarctic krill Euphausia
superba (Croxall et al. 1988, Murphy et al. 1998) and has historically been
most frequently detected in the response of upper-trophic-level predators.
It was first noted in the whaling era when the numbers and species of whale
caught varied between years (Harmer 1931, Kemp & Bennett 1932). Since
the mid 1970s, monitoring of the population size and reproductive perfor-
mance of a range of predators that depend upon krill has revealed the peri-
odic occurrence of years of dramatically reduced breeding success across
almost all land-based krill predators (Croxall et al. 1999, Reid & Croxall
2001). In addition, data from krill-eating fish caught around South Georgia
indicate that their body condition (mass per unit length) shows low values
coincident with these periods of reduced breeding success in land-based
predators (Everson et al. 1997). During some of these periods of breeding
failure, estimates of krill biomass derived from shipboard acoustic surveys
have revealed unusually low levels of local krill abundance (Brierley et al.
1999). Thus there is a clear link between the performance of krill-dependent
predators and an independent assessment of the availability of their prey.

A SUITABLE SAMPLER

Determining changes in the krill population dynamics associated with

changes in abundance is limited by the ability to sample the krill popula-
tion over appropriate time scales both within and between years. The oper-
ational constraints of shipboard sampling in this large and inhospitable
region mean that obtaining samples of the size structure of the krill popu-
lation is typically limited to a short period, usually only 2 to 3 weeks during
January and/or February (Watkins et al. 1999). Sampling in this way pro-
duces a quantifiable ‘snapshot’ of the population size and structure but the
period of time over which such estimates can be extrapolated is difficult to
determine.
Krill in the diet of predators provides a potential sampling method
with which to obtain data on the size composition of the krill popula-
tion. Although this approach is unlikely to provide the same quantifiable
data as sampling with nets, it does have the advantage that it can be used
over time scales not available using conventional sampling techniques.
This is especially so at Bird Island, South Georgia where a year-round
 Population dynamics of Antarctic krill 251

Box 17.1 Seals as samplers of krill

During lactation (November to March), female Antarctic fur seals
are central-place foragers. They engage in repeated cycles of 3 to 4
days at sea followed by 1 to 2 days ashore feeding their pup before
returning to sea to feed once again (Boyd et al. 1991). Outside the
breeding season the population ashore is dominated by sub-adult
males that appear to alternate periods at sea with time spent ashore.
This regular presence of fur seals on land provides an opportunity
to regularly collect fresh scats (Reid 1995). The exoskeletons of krill
pass through the digestive tract sufficiently intact to allow the size
of the ingested krill to be estimated. This means that there is no
requirement to perform invasive procedures on the animals in order
to estimate their diet. Measuring the length of the krill carapaces in
the scats allows estimation of the total length of the krill consumed.
It is then possible to reconstruct the size–frequency composition of
the krill in the diet of the seal during its most recent period at sea
(Hill 1990, Reid & Measures 1998, Staniland 2002).

biological research programme, including dietary sampling, is conducted

on upper-trophic-level, krill-dependent predators (Croxall et al. 1988). Of
the species studied, the Antarctic fur seal Arctocephalus gazella is perhaps
the most suitable sampler since it is present ashore throughout the year
and its diet during both summer and winter is dominated by krill (Reid
1995, Reid & Arnould 1996; Box 17.1 and Fig. 17.1).
Assessing the size composition of krill in the diet of the Antarctic fur
seal has a number of potential biases relating to where and when the krill
were taken and to what extent fur seals feed selectively with respect to the
size of krill. Most foraging by female fur seals from Bird Island takes place
on, or at the margins of, the South Georgia continental shelf (Staniland
et al. (Chapter 9 in this volume)) and this is the region in which the major-
ity of the krill biomass is reported from acoustic surveys (Brierley et al.
1999). In addition, there are no differences in the size of krill in the diet
of Antarctic fur seals in relation to the areas in which they forage. Antarctic
fur seals, along with other krill predators, tend to preferentially select large,
particularly gravid, female krill probably because of a combination of their
high energy content and reduced escape abilities compared with male krill
252 K. Reid et al.

Fig. 17.1 Antarctic fur seal Arctocephalus gazella.

(Hill et al. 1996, Reid et al. 1996). Nevertheless, a multi-year comparison of

the length–frequency distribution of krill in the diet of Antarctic fur seals
and data from scientific nets taken at the same time from within the forag-
ing range of the seals indicates that these two methods of sampling provide
a very similar view of the overall size composition of the local krill popula-
tion (Reid et al. 1999).

KRILL ABUNDANCE AND POPULATION DYNAMICS

During January 1994, a period of low krill biomass and very poor repro-
ductive success of krill-dependent predators at South Georgia (Croxall et al.
1999), the krill population structure in simultaneous samples from scien-
tific nets and land-based krill predators both showed a distinctly bimodal
distribution (Reid et al. 1999). However, samples from Antarctic fur seals,
in the weeks prior to and after this period, revealed a progressive change
 Population dynamics of Antarctic krill 253

03-Jan-94

10-Jan-94

18-Jan-94

24-Jan-94 30 38 46 54 62

31-Jan-94

06-Feb-94

30 38 46 54 62
Krill length (mm) Krill length (mm)
Antarctic fur seal Scientific nets

Fig. 17.2 The length–frequency distribution of Antarctic krill at South Georgia

from weekly samples of the diet of Antarctic fur seals during January and
February 1994, and from scientific-net samples collected on 9 to 10 January
1994.

from a dominance of large krill with a complete absence of small krill; the
gradual appearance of small krill producing a short period of bimodality
(when the net samples were collected) followed by a complete dominance
of small krill with no large krill present (Fig. 17.2). In contrast to this change
in the krill population structure, the modal size of krill remained more or
less constant throughout the predator breeding seasons of 1993 and 1995,
when predator reproductive success (and by inference krill abundance) was
considered ‘normal’ (Reid et al. 1999). These findings provided the first
evidence that distinct changes in the population structure of krill accompa-
nied periods of low krill biomass and, in turn, suggested that the population
dynamics of krill is a major factor in driving the inter-annual variability
in abundance. This pattern of change in the length–frequency distribu-
tion of krill provided a fundamental insight into the population dynamics
254 K. Reid et al.

55
53
51

Krill length (mm)

49
47
45
43
41
39
37
35
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year

Fig. 17.3 The mean length of Antarctic krill in the diet of Antarctic fur seals
during the first 14 days of March 1990–2003.

of krill and would have gone undetected using conventional ship-based

sampling.
In order to describe the inter-annual variation in the krill population,
Reid et al. (1999) derived an annual index from the length of krill eaten
by fur seals during early March, as this was when the short-term (week to
week) variation in krill size was lower than at any other time during the sum-
mer. This index, from 1990 to 2003 (Fig. 17.3), showed a distinct pattern,
whereby a large mean size of krill (in 1990, 1993 and 1997) preceded years
of low krill abundance; at least as indicated by the low reproductive perfor-
mance of predators (Reid & Croxall 2001). The large mean size of krill in
these years was attributed to a failure of small krill to enter the population
(Box 17.2).

KRILL DEMOGRAPHY REVISITED

Information from the diets of Antarctic fur seals was used to examine the
population dynamics of Antarctic krill by Murphy and Reid (2001); they
showed that the absence of a single age class of krill could indeed produce
the changes in the length–frequency distribution observed in the diet of
predators. Moreover, the analysis suggested that the observed size compo-
sition of krill in samples from both predators and nets was consistent with a
relatively high rate of krill mortality at South Georgia compared with other
areas in the Scotia Sea. This increased rate of mortality also means that the
changes in length–frequency distribution associated with the failure of a
single age class to recruit into the population would also be consistent with
the observed reduction in biomass (cf. Priddle et al. 1988).
The growth of Antarctic krill is thought to be highly seasonal and the
work of Mackintosh (1974) indicated a relatively rapid increase in the mean
 Population dynamics of Antarctic krill 255

Box 17.2 A conceptual model of krill population dynamics at

South Georgia
In order to link the inter- and intra-annual changes in the size struc-
ture of the krill population a conceptual model of the annual devel-
opment of a krill population was produced by Reid et al. (1999). The
model consists of three size/age classes with modal sizes of 28, 42
and 54 mm and shows how the mean size of krill, as indicated by
the solid vertical bars in Fig. 17.4, changes in relation to the pres-
ence/absence of a single class of krill.
In Fig. 17.4a, the krill population is represented by only the first
age class in the population; in Fig. 17.4b, the initial age-class-1 krill
have progressed into the second age class and a new age class 1 has
recruited into the population; in Fig. 17.4c, age classes 1, 2 and 3 are
present – this may be considered the equilibrium population struc-
ture. In Fig. 17.4d, there are no age-class-1 krill, as a result of a recruit-
ment failure, and the population is comprised of age classes 2 and 3
and hence the mean size has increased.
At the beginning of the year following a recruitment failure only
age-class-3 krill are present with the gradual arrival of the newly
recruiting age-class-1 krill and the reduction in the relative frequency
of the age class 3 – as shown in the time series of length–frequency
distribution from the diet of Antarctic fur seals in 1994 in Fig. 17.2.
In each of the years following such a ‘recruitment failure’ the
same shift in the krill length–frequency distribution, as described
above for 1994, was observed. Thus by the following March the mean
length of krill was small and this is reflected in the marked decrease
in mean krill length in the years of reduced predator performance.

length of krill over the period October to December. However, this is a

period of the year when net samples are rarely collected and especially not
over the time scales required to evaluate growth rates. Since Antarctic fur
seal are present at South Georgia throughout the year, Reid (2001) anal-
ysed the length–frequency distributions from regular sampling over the
period October to December in several years and found a consistent pat-
tern of increase in size. Fitting these observations to a seasonally adjusted
Von Bertalanfy growth model indicated that krill at South Georgia had a
much higher growth rate than had been previously assumed on the basis of
growth measurements from elsewhere in the Scotia Sea (Siegel 1987).
256 K. Reid et al.

(a)

10 20 30 40 50 60 70
(b)

10 20 30 40 50 60 70
(c)

10 20 30 40 50 60 70
(d)

10 20 30 40 50 60 70
Krill length (mm)

Fig. 17.4 Changes in mean krill length in relation to the presence/absence of a

single class of krill. See Box 17.2 for explanation of panels a–d.

Although Antarctic krill are generally considered to be advected into

the South Georgia region from the Antarctic Peninsula, attempts to follow
the fate of individual cohorts across the Scotia Sea had been unsuccessful
(Murphy et al. 1998). On the basis of the re-evaluation of growth and mor-
tality rates, Reid et al. (2002) undertook a comparison of a time series of
length–frequency distributions of krill from the two regions. This analy-
sis used the length–frequency distribution of krill from net samples taken
in the South Shetland Islands and applied the demographic rates from
South Georgia to derive a new length–frequency distribution that was then
‘sampled’ by a fur seal using the selectivity ogive derived by Murphy and
Reid (2001). The raw length–frequency distribution from South Georgia
and the South Shetland Islands showed no overlap in the dominant modes.
 Population dynamics of Antarctic krill 257

However, having accounted for the differences in demographic parameters

there was almost complete overlap in the dominant modes; such that years
of good and bad recruitment of the first year class were reflected in the two
locations in the same years.

DRIVERS OF KRILL VARIABILITY

The relatively high growth and mortality rates at South Georgia places an
increased importance on the role of recruitment variability at South Geor-
gia – particularly the occurrence of years of recruitment failure – in driving
ecosystem variability. Since the analysis of population dynamics across the
Scotia Sea indicated a large-scale concordance in the pattern of recruitment,
this suggests a link between recruitment and some large-scale forcing fac-
tor. To further investigate any such environment–krill recruitment relation-
ship would require a quantifiable index of recruitment. Although no such
index exists, at least at South Georgia, the index of mean krill length in the
diet of Antarctic fur seals in March (Fig. 17.3) provides a relative index of
recruitment insofar as it indicates periods of recruitment failure.
A relationship between extensive sea-ice cover during the winter and
subsequent high recruitment of krill in the Antarctic Peninsula region has
been suggested by several authors (Loeb et al. 1997, Hewitt et al. 2003). In
addition, Trathan et al. (2003) found a relationship between krill biomass
and sea-surface temperature at South Georgia over the period 1997–9.
Therefore, since sea-ice does not reach as far north as South Georgia,
we have used an index of sea-surface temperature to examine the poten-
tial environment–recruitment relationship at South Georgia. A preliminary
examination of the relationship between the index of krill length and the
sea-surface temperature at the start of the summer suggested that the years
in which there was a recruitment failure were those of high sea-surface tem-
perature; or at least those years at the high point of the sea-surface temper-
ature cycle (Fig. 17.5).
Sea-surface temperature is frequently used as a proxy for physical
changes in the marine environment as it is relatively easy to measure, par-
ticularly using remote systems such as satellites (Yuan & Martinson 2000).
It is important to recognize that in a correlational analysis such as this we
are not necessarily suggesting that warm sea-surface temperatures cause
a krill recruitment failure; rather, there may be some other process that
causes recruitment failure – but sea-surface temperature is a suitable proxy
for this process. Indeed the time series shown in Fig. 17.5 suggests that
the relationship between sea-surface temperature and recruitment was less
well defined during the period 1999–2003. Similarly an analysis of the
258 K. Reid et al.

55 2.50
53
51 2.00

Krill length (mm)

49
47 1.50

SST (°C)
45
43 1.00
41
39 0.50
37
35 0.00
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year

Krill length SST

Fig. 17.5 The mean length of Antarctic krill in the diet of Antarctic fur seals
during the first 14 days of March 1990–2003 and the mean sea-surface
temperature (SST) at South Georgia (SST data are from Reynolds et al. 2002).

relationship between the number of leopard seals (Hydrurga leptonyx) and

indices of sea-surface temperature and sea-ice in the South Georgia region
(Jessopp et al. 2004) indicated that throughout the early 1990s sea-surface
temperature and sea-ice were positively correlated but this relationship
decayed markedly after 1999. Hence the potential to use sea-surface tem-
perature as a predictor of recruitment processes may be limited to those
periods where it provides a good proxy for the processes that actually drive
krill recruitment.

PREDATORS AS SAMPLERS

The long-term monitoring of land-based marine predators at South Georgia

over the past two decades has revealed changes in the relationship between
krill population structure/abundance and the reproductive performance of
a range of predators with very different life-history strategies and ecolo-
gies. An increase in the frequency of years in which there is insufficient
krill to support predator demand indicates an ecosystem-wide response and
suggests that there has been a fundamental change in the availability of
krill over a large scale (Reid & Croxall 2001). The relationship between
the recruitment patterns of krill and environmental-temperature indices
may have particular relevance given the increasing evidence for a long-
term reduction in the extent and duration of sea-ice (Murphy et al. 1995)
and the increase in winter air temperatures (King 1994, Broeke 2000) in
the Antarctic Peninsula region. In the most simplified scenario, a long-
term increase in temperature may be linked to an increased frequency of
years of recruitment failure and an increase in the frequency of years when
 Population dynamics of Antarctic krill 259

the amount of krill available to predators is insufficient to support demand

(Reid & Croxall 2001).
The potential for predators to provide quantifiable responses to changes
in prey availability means that they have the potential to act as barometers
of change. In this study we have shown that they can also provide estimates
of life-history variables for prey species; a relationship that has led to a re-
evaluation of the processes involved in krill population dynamics. These
life-history variables are also an essential part of the management of fish-
eries for the prey species, in this case krill. Using predators to sample prey
populations at different scales should be viewed as an important addition to
the information available from conventional sampling. It would be naive to
suggest that predators could entirely replace conventional sampling meth-
ods; but the information provided by predators could reduce the uncertainty
in fisheries management models.

ACKNOWLEDGEMENTS

We thank all of the British Antarctic Survey staff who have collected, sorted
and measured krill in the diet of predators at Bird Island and to Drs Jon
Watkins and Bill Montevecchi, and Professor Ian Boyd for their advice and
encouragement.

REFERENCES
Boyd, I. L., Lunn, N. J. & Barton, T. (1991). Time budgets and foraging
characteristics of lactating Antarctic fur seals. J. Anim. Ecol., 60, 577–92.
Brierley, A. S., Watkins, J. L., Goss, C., Wilkinson, M. T. & Everson, I. (1999).
Acoustic estimates of krill density at South Georgia, 1981 to 1998. CCAMLR
Sci., 6, 47–57.
Broeke, M. R. van den (2000). On the interpretation of Antarctic temperature
trends. J. Clim., 13, 3885–9.
Croxall, J. P., McCann, T. S., Prince, P. A. & Rothery, P. (1988). Reproductive
performance of seabirds and seals at South Georgia and Signy Island, South
Orkney Islands, 1976–1987: implications for Southern Ocean monitoring
studies. In Antarctic Ocean and Resources Variability, ed. D. Sahrhage. Berlin:
Springer-Verlag, pp. 261–85.
Croxall, J. P., Reid, K. & Prince, P. (1999). Diet, provisioning and productivity
responses of marine predators to differences in availability of Antarctic krill.
Mar. Ecol. Prog. Ser., 177, 115–31.
Everson, I., Kock, K. H. & Parkes, G. (1997). Interannual variation in condition of
the mackerel icefish. J. Fish Biol., 51, 146–54.
Harmer, S. F. (1931). Southern whaling. Proc. Linn. Soc. Lond., 142, 85–163.
Hewitt, R. P., Demer, D. A. & Emery, J. H. (2003). An 8-year cycle in krill biomass
density inferred from acoustic surveys conducted in the vicinity of the South
260 K. Reid et al.

Shetland Islands during the austral summers of 1991–1992 through

2001–2002. Aquat. Living Resour., 16, 205–13.
Hill, H. J. (1990). A new method for the measurement of Antarctic krill Euphausia
superba Dana from predator food samples. Polar Biol., 10, 317–20.
Hill, H. J., Trathan, P. N., Croxall, J. P. & Watkins, J. L. (1996). A comparison of
Antarctic krill Euphausia superba caught in nets and taken by macaroni
penguins Eudyptes chrysolophus: evidence for selection? Mar. Ecol. Prog. Ser.,
140, 1–11.
Jessopp, M. J., Forcada, J, Reid, K., Trathan, P. N. & Murphy E. J. (2004). Winter
dispersal of leopard seals (Hydrurga leptonyx): environmental factors
influencing demographics and seasonal abundance. J. Zool. Lond., 263, 251–8.
Kemp, S. & Bennett, A. G. (1932). On the distribution and movements of whales on
the South Georgia and South Shetland whaling grounds. Discovery Rep., 6,
165–90.
King, J. C. (1994). Recent climate variability in the vicinity of the Antarctic
Peninsula. Int. J. Climat., 14, 357–69.
Loeb, V., Siegel, V., Holm-Hansen, O. et al. (1997). Effects of sea-ice extent and
krill or salp dominance on the Antarctic food web. Nature, 387, 897–900.
Mackintosh, N. A. (1974). Sizes of krill eaten by whales in Antarctica. Discovery
Rep., 36, 157.
Murphy, E. J. & Reid, K. (2001). Modelling Southern Ocean krill population
dynamics: biological processes generating fluctuations in the South Georgia
ecosystem. Mar. Ecol. Prog. Ser., 217, 175–89.
Murphy, E. J., Clarke, A., Symon, C. & Priddle, J. (1995). Temporal variation in
Antarctic sea-ice: analysis of a long term fast-ice record from the South Orkney
Islands. Deep-Sea Res. I, 42, 1045–62.
Murphy, E. J., Watkins, J. L., Reid, K. et al. (1998). Interannual variability of the
South Georgia marine ecosystem: biological and physical sources of variation
in the abundance of krill. Fish. Oceanogr., 7, 381–90.
Priddle, J., Croxall, J. P., Everson, I. et al. (1988). Large-scale fluctuations in
distribution and abundance of krill: a discussion of possible causes. In
Antarctic Ocean and Resources Variability, ed. D. Sahrhage. Berlin:
Springer-Verlag, pp. 169–82.
Reid, K. (1995). The diet of Antarctic fur seals (Arctocephalus gazella Peters 1875)
during winter at South Georgia. Antarct. Sci., 7, 241–9.
(2001). Growth of Antarctic krill Euphausia superba at South Georgia. Mar. Biol.,
138, 57–62.
Reid, K. & Arnould, J. P. Y. (1996). The diet of Antarctic fur seals Arctocephalus
gazella during the breeding season at South Georgia. Polar Biol., 16, 105–14.
Reid, K. & Croxall, J. P. (2001). Environmental response of upper trophic-level
predators reveals a system change in an Antarctic marine ecosystem. Proc.
R. Soc. Lond. B, 268, 377–84.
Reid, K. & Measures, J. (1998). Determining the sex of Antarctic krill Euphausia
superba using carapace measurements. Polar Biol., 19, 145–7.
Reid, K., Trathan, P. N., Croxall, J. P. & Hill, H. J. (1996). Krill caught by predators
and nets: differences between species and techniques. Mar. Ecol. Prog. Ser.,
140, 13–20.
 Population dynamics of Antarctic krill 261

Reid, K., Watkins, J. Croxall, J. & Murphy, E. (1999). Krill population dynamics at
South Georgia 1991–1997, based on data from predators and nets. Mar. Ecol.
Prog. Ser., 117, 103–14.
Reid, K., Murphy, E. J., Loeb, V. & Hewitt, R. P. (2002). Krill population dynamics
in the Scotia Sea: variability in growth and mortality within a single population.
J. Mar. Syst., 36, 1.
Reynolds, R. W., Rayner, N. A., Smith, T. M., Stokes, D. C. & Wang, W. Q. (2002).
An improved in situ and satellite SST analysis for climate. J. Clim., 15, 1609–25.
Siegel, V. (1987). Age and growth of Antarctic Euphausiacea (Crustacea) under
natural conditions. Mar. Biol., 96, 483–95.
Staniland, I. J. (2002). Investigating the biases in the use of hard prey remains to
identify diet composition using Antarctic fur seals (Arctocephalus gazella) in
captive feeding trials. Mar. Mamm. Sci., 18, 223–43.
Trathan, P. N., Brierley, A. S., Brandon, M. A. et al. (2003). Oceanographic
variability and changes in Antarctic krill (Euphausia superba) abundance at
South Georgia. Fish. Oceanogr., 12, 569.
Watkins, J. L., Murray, A. W. A. & Daly, H. I. (1999). Variation in the distribution
of Antarctic krill Euphausia superba around South Georgia. Mar. Ecol. Prog. Ser.,
188, 149–60.
Yuan, X. J. & Martinson, D. G. (2000). Antarctic sea ice extent variability and its
global connectivity. J. Clim., 13, 1697–717.
 18
The functional response of generalist predators
and its implications for the monitoring
of marine ecosystems
C . A S S E B U R G , J . H A R W O O D , J . M AT T H I O P O U L O S
AND S. SMOUT

It is often suggested that changes in the population biology of higher preda-

tors can be used as proxies for other processes within marine ecosystems,
such as changes in the size of prey populations. However, such predators
are almost always generalists, which are likely to respond to changes in
the abundances of more than one prey species. Using data from a terres-
trial generalist predator, we show that the form of the relationship between
energy intake and the abundance of a focal prey species can vary greatly
depending on the abundance of alternative prey, and that such proxies may
have insufficient statistical power to detect even substantial changes in prey
abundance. We then consider whether alternative approaches to analysing
the data collected by higher-predator monitoring schemes might provide
more reliable information on ecosystem processes.
An increasing number of nations and intergovernmental organizations
have accepted the principle that the exploitation of living resources should
be conducted using an ecosystem-based approach (e.g. the 1996 amend-
ment to the US Magnuson-Stevens Fishery Conservation and Manage-
ment Act or the United Nations Fish Stocks Agreement (United Nations
1995), see also Aqorau (2003)). The objectives of such an approach are
rarely clearly defined, but they usually involve ensuring that the ‘health’ or
‘integrity’ of an ecosystem is maintained. For example, the United Nations
Fish Stocks Agreement states that one aim of fisheries management is
to ‘maintain the integrity of marine ecosystems and minimize the risk of

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 The functional response of generalist predators 263

long-term or irreversible effects of fishing operations’ (United Nations

1995). However, ecosystem ‘integrity’ is not a well-defined quantity and can-
not be monitored directly. Much research has therefore focused on identi-
fying relatively easily measured characteristics of ecosystems that might act
as proxies for ‘integrity’.
Considerable progress in this direction has been made by the Scien-
tific Committee of the Convention on the Conservation of Antarctic Marine
Living Resources (CCAMLR). The Convention requires that ‘ecological rela-
tionships between harvested, dependent and related populations of Antarc-
tic marine living resources’ should be maintained (CCAMLR 1982). A par-
ticular concern has been the impact of commercial exploitation of krill
(Euphausia superba) on the many predators that appear to rely on this
species. In 1985, members of the Convention adopted an Ecosystem Mon-
itoring Programme (CCAMLR 1985), which monitors a range of variables
associated with the population dynamics of nine bird and seal predators of
krill (Constable et al. 2000). These variables include: the weight of animals
on arrival at breeding colonies, diet and growth rates of seabird chicks and
seal pups, and changes in population size. The central thesis is that these
variables provide a realistic proxy for krill availability in the regions that are
being monitored (Reid & Croxall 2001, see also Davoren & Montevecchi
2003).
Although krill is the most important prey in the diet of the predators
chosen by CCAMLR’s Scientific Committee, all of them consume a range
of prey species – even the Antarctic fur seal Arctocephalus gazella which is
normally thought of as a krill specialist (see Reid & Arnould 1996). Accord-
ingly, in this group of predators, the consumption of krill is likely to be
affected by the availability of prey other than krill. If we are to interpret cor-
rectly the observed changes in the variables chosen for monitoring, we need
to understand the way that prey consumption is affected by changes in the
abundances of all potential prey species.
To the best of our knowledge, there are no published examples of multi-
species responses of this kind for any marine predators. In order to illustrate
the way in which changes in prey abundance may affect the diet and produc-
tivity of a generalist predator, we have used the results of a recent analysis
(Asseburg 2005) of the response of a terrestrial predator to changes in the
abundance of its three principal prey species. In this chapter we focus par-
ticularly on how changes in the abundances of different prey species affect
the quantity of food supplied by a predator to its young – one of the key
parameters in CCAMLR’s Ecosystem Monitoring Programme (CCAMLR
1985).
264 C. Asseburg et al.

FUNCTIONAL RESPONSES, ENERGY INTAKE AND

PREDATOR PERFORMANCE

The relationship between the dynamics of a predator population and

the abundance of its prey is often described as the predator’s numerical
response (Turchin 2003). There are three components, at least, of this
response: predator growth, predator reproduction and predator mortality
(Beddington et al. 1976). For the sake of convenience, we shall refer to these
components collectively as predator ‘performance’. Beddington et al. (1976)
showed that each component can be modelled in terms of the relation-
ship between prey intake and prey density, that is, the predator’s functional
response (Box 18.1). In order to take account of the effects of all the prey
species consumed by a generalist predator, we need to expand the classic
single-species functional response into a multispecies form that relates the
consumption of each prey species to the density of all prey species. One pos-
sible form for this multispecies functional response (MSFR) is presented in
Box 18.1. Gentleman et al. (2003) have provided a useful review of other for-
mulations, and their general properties.
To model the effects of consumption of different prey species on preda-
tor performance, we need to combine the benefits the predator gains by
consuming these prey. The calorific energy of each prey item is the obvi-
ous currency to use to achieve this combination (Ginzburg 1998). We have
therefore expressed total prey consumption in terms of energy by multi-
plying the number of prey items consumed by the average energy content
for that prey species (Box 18.1 and Fig. 18.1). This ignores any differences
between prey species in the energetic costs of capturing and digesting indi-
vidual prey items and in the other nutrients they provide.

INVERSE INFERENCE AND UNCERTAINTY

Estimating the availability of a prey from measurements of predator per-

formance requires a form of inverse inference: the energy intake – or some
related measure of performance – is known, and we want to find a prey den-
sity that may have given rise to this observed energy intake. However, the
functions involved in this inference are non-linear and plateau at high prey
densities. In addition, the relationship between predator performance and
prey intake may be limited through constraints imposed by the predator’s
physiology and demography. As a result of this non-linearity, some prey
densities cannot be estimated precisely because even large changes in prey
density have no appreciable effect on performance (Box 18.2).
 The functional response of generalist predators 265

Box 18.1 Single- and multispecies functional responses (MSFRs)

The relationship between prey density and the rate of consump-
tion by an individual predator is called the predator’s ‘functional
response’. For a specialist predator, the functional response may be
described by an equation such as
a Nm
C= (18.1)
1 + at N m
(Real 1977). This equation has been used to model prey consumption
by predators in the CCAMLR area (Thompson et al. 2000). C repre-
sents consumption rate; N represents prey density; and a, t and m
are parameters. If m = 1, then the equation represents a hyperbolic
relationship equivalent to the Holling disc equation. If m > 1, the
function takes a sigmoidal form.
For a generalist predator, which consumes more than one type
of prey, the consumption of one prey species may be affected by
changes in the density of alternative prey. For example, the predator
may eat more of a particular prey species – even though the abun-
dance of that prey is unchanged – because other prey have become
less abundant. To take account of this, we need an MSFR. One pos-
sible multispecies formulation of equation (18.1) (Yodzis 1998) is
ai Nimi
Ci = (18.2)

n
m
1+ a j t j Nj j
j =1

in which the subscripts i and j refer to individual prey species.

Consumption of a focal prey species by a generalist predator
The continuous line in Fig. 18.1 shows the predicted consumption
rate of a focal prey species when there are no alternative prey. As
the abundance of alternative prey is increased (dashed and dotted
curves) the consumption of focal prey at a particular prey density
decreases and the prey density at which the asymptotic prey con-
sumption rate is achieved increases.
We can expand equation (18.2) to provide a function that relates
the energy intake of a predator to the density of all prey species:

e i ai Nimi
Energy ingested =  m (18.3)
1 + t j a j Nj j
j

where ei is the energy content of one item of prey i.

266 C. Asseburg et al.

0.6

0.5

0.4
Consumption rate

0.3

0.2

0.1

0.0
0 20 40 60 80 100
Focal prey density

Fig. 18.1

A further source of uncertainty derives from the empirical nature of

our knowledge of the relationship between performance and prey density –
the functional response must be estimated from observational data which
will be subject to observational error and the effects of individual variation.
Even small uncertainties in the estimate of predator performance will trans-
late into substantial uncertainty in prey availability (Box 18.2). Yet another
source of uncertainty is the imprecise knowledge of the abundances of other
prey species because, for a generalist predator, energy intake will depend on
the abundances of all prey.
The interaction between uncertainty and non-linearity in the relation-
ship between predator performance and prey density implies that predator
performance may not be a reliable proxy for ecosystem health. For similar
reasons, Constable (2001) concluded that the abundance and overall pro-
ductivity of the predators chosen by CCAMLR is likely to be a relatively
insensitive indicator of the effects of fishing on the ecosystem.

IMPLICATIONS FOR ECOSYSTEM MONITORING

Our analysis of the MSFR of a terrestrial predator (Box 18.3) illustrates

how the performance of a generalist predator may be used in monitoring
 The functional response of generalist predators 267

Box 18.2 Uncertainty and the inference of prey density from

predator performance
When inverse inference is used to deduce prey density from preda-
tor performance, uncertainty in the measurement of performance
and uncertainty in the form of the predator’s response to prey den-
sity both affect the precision of the estimates of prey density.

(a) Measurement uncertainty

Even if the form of the relationship between predator performance
and prey density is known precisely, uncertainty in the performance
estimates results in credibility intervals for estimated prey density
that vary with prey density.
Figure 18.2 shows how the same level of uncertainty in three dif-
ferent estimates of consumption rate (y-axis) translates into uncer-
tainty (represented by credibility levels) in the estimates of prey
abundance (x-axis). This depends on the gradient of the function.
At high or low consumption rates, where the gradient is shallow,
uncertainty in the estimate of prey density is much greater than the
uncertainty in consumption.

0.6

0.5

0.4
Consumption rate

0.3

0.2

0.1

0.0
0 20 40 60 80 100
Focal prey density

Fig. 18.2
268 C. Asseburg et al.

0.8

0.6
Consumption rate

0.4

0.2

0
0 20 40 60 80 100
Focal prey density

Fig. 18.3

(b) Parameter uncertainty

In practice, the parameters of the function relating performance
and prey density must be estimated from field data, which are sub-
ject to process and observation error. Therefore, depending on the
amounts of data available, there will be substantial uncertainty about
the precise shape of the function, which we can account for using a
probability distribution.
Figure 18.3 shows the probability distribution for a MSFR fit-
ted to sparse data on prey consumption by a terrestrial generalist
predator (Asseburg 2005). Here alternative prey are fixed at an inter-
mediate abundance. The shading indicates the posterior probabil-
ity of the Bayesian model fit – the darker the shading, the greater
the likelihood of observing the corresponding consumption of the
focal prey. Figure 18.3 shows that even when performance (in this
case, consumption) can be estimated with high precision, there
will be considerable uncertainty (indicated by the bold lines on the
x-axis, which represent 95% credibility intervals) in the estimates of
prey density, particularly at high prey densities.
 The functional response of generalist predators 269

Box 18.3 The MSFR of a generalist predator: fitting and

implications
Fitting the MSFR
We used data on the consumption of three prey species – the red
grouse Lagopus lagopus scoticus, the meadow pipit Anthus pratensis,
and the field vole Microtus agrestis – by hen harriers Circus cyaneus
to estimate the predator’s MSFR (Asseburg 2005) using Bayesian
methodology. Consumption rates were measured by observing the
provisioning of harrier chicks by parent birds, and observations were
made for 11 different combinations of prey density (Redpath & Thir-
good 1999). Informative priors were derived from field observations
of handling time and from data on the relationship between attack
rate and prey densities. We then used this fitted MSFR to predict
how total energy intake by the predator varies as a function of prey
density. The energy values for these items were derived from pub-
lished values (Dierenfeld et al. 2002, Park et al. 2002).

20 000

19 000
Total energy intake (kJ)

18 000

17 000

16 000

15 000

14 000
0 500 1000 1500 2000
Vole density (individuals km −2)

Fig. 18.4

Example 1: voles as focal prey, other prey is scarce

Figure 18.4 shows mean total energy intake resulting from the con-
sumption of all three prey species calculated over a range of field
270 C. Asseburg et al.

18000

Total energy intake (kJ) 17500

17000

16500

16000

15500

15000
0 500 1000 1500 2000
Vole density (individuals km−2)

Fig. 18.5

vole densities. Alternative prey is present only at very low levels

(grouse, 15 per km2 ; pipits, 15 per km2 ).
The predator’s energy intake tracks the abundance of voles,
showing an almost linear increase with increasing vole density.
Example 2: voles as focal prey, grouse abundance varies
Figure 18.5 shows predicted mean total energy intake plotted against
vole density when pipits are at a medium abundance (50 per km2 ),
and grouse numbers are varied. At low grouse density (15 per km2 ,
continuous curve) energy intake tracks the vole density; however at
intermediate (70 per km2 , bold-dotted curve) and high (120 per km2 ,
dotted curve) grouse densities, energy intake is almost unaffected
by changes in vole density. This is because harriers increase their
consumption of grouse as vole density declines, thus maintaining
an almost constant energy intake.
Uncertainty in the predicted total energy intake
The two examples above assume that we have perfect knowledge of
the relationship between prey density and total energy intake. In fact,
we can only model this relationship and we should account for the
uncertainty expressed in the MSFR data (see Box 18.2).
 The functional response of generalist predators 271

26000

24000
Total energy intake (kJ)

22000

20000

18000

16000

14000

12000
0 500 1000 1500 2000
Vole density (individuals km−2)

Fig. 18.6

Figure 18.6 shows the 80% credibility interval around the rela-
tionship between mean energy intake and prey density from Exam-
ple 1. When we take account of this uncertainty, the positive rela-
tionship between vole density and total energy intake is no longer so
clear.

the density of a prey species. The predicted energy intake derived from
this MSFR does show the hoped-for simple monotonic relationship to the
density of a focal prey species when other prey species are rare or absent
(Example 1 in Box 18.3). However, when alternative prey are more abundant
(Example 2 in Box 18.3), there may be virtually no relationship between total
energy intake and prey density. In addition, because the MSFR was derived
using the relatively sparse data typical of field situations, the uncertainty
in our knowledge of this relationship results in imprecise estimates of prey
density, even when energy intake and prey density should be strongly corre-
lated. Clearly, we need to know the form of the MSFR, and the uncertainty
associated with it, in order to identify when predator performance might
provide a reliable index of prey density.
272 C. Asseburg et al.

If the main concern of a monitoring programme is to ensure the preda-

tor’s persistence, the concerns described above are not particularly impor-
tant. However, they are fundamental if predator performance is a compo-
nent in the ecosystem-based management of lower trophic levels (Stefans-
son 2003). When one prey species predominates in the predator’s diet, as is
the case with krill in the Southern Ocean, the monitoring of higher preda-
tors may provide useful information on this prey species. However, when
there are many potential prey species – as in the North Sea – this moni-
toring approach will be much less informative, because predator dynamics
are likely to be influenced by the abundances of many prey species and also
because large quantities of data will be required to specify the predators’
functional responses precisely.

UNDERSTANDING THE GENERALIST PREDATOR’S ROLE

IN AN ECOSYSTEM

Even when predator performance is only used as a qualitative proxy for

ecosystem health, some information on prey abundance is required to cali-
brate this relationship. However, the same data can also be used to parame-
terize the predator’s MSFR, which provides quantitative information about
the form of the relationship between predator performance and the avail-
ability of all prey species. The terrestrial example we have used (Box 18.3)
shows clearly that, in this system and with the data on prey consumption
that are currently available, inverse inference can only be used over a rela-
tively narrow range of prey densities.
However, this will not always be the case. For example, Croxall et al.
(1999) observed a 90% reduction in the mass of krill in the diets of seabirds
and fur seals on South Georgia following a four-fold decrease in local krill
biomass, and an equivalent reduction in the rates at which their offspring
were provisioned. This suggests that the slope of the relationship between
energy intake and prey density for these predators is much steeper than in
our terrestrial example.
Our suggestion that the MSFR of higher predators, a complex equa-
tion fitted to empirical data, should be a central component of ecosystem-
based management seems to ignore the growing realization (Punt &
Hilborn 1997) that the use of simple control laws based on easily col-
lected data results in a more effective and precautionary management of
marine resources than procedures based on detailed biological models of
the exploited stocks (Geromont et al. 1999, Punt & Smith 1999). However,
we are not proposing that MSFRs should form part of the control laws used
 The functional response of generalist predators 273

in management. Rather, we suggest that they are an essential component

of the operating models that are used to evaluate the effectiveness of differ-
ent control laws (Harwood & Stokes 2003). These computer-based models
attempt to capture the full complexity of the system that is being managed.
They have proved to be a particularly effective way of accounting explic-
itly for uncertainty in the available knowledge on that system (Harwood &
Stokes 2003). The approach we are advocating makes it possible to use a
wide range of information (in the form of prior distributions and new data)
to determine the general form of the interactions between ecosystem com-
ponents and to document the uncertainty associated with these relation-
ships.

ACKNOWLEDGEMENTS

We thank Simon Thirgood and Steve Redpath for sharing with us their data
and insights on hen harrier foraging, and Carmen Fernández for statistical
advice.
Christian Asseburg and Sophie Smout were supported by studentships
from the Scottish Higher Education Funding Council and the Natural Envi-
ronment Research Council respectively.

REFERENCES
Aqorau, T. (2003). Obligations to protect marine ecosystems under international
conventions and other legal instruments. In Responsible Fisheries in the Marine
Ecosystem, eds. M. Sinclair & G. Valdimarsson. Wallingford, UK: CABI
Publishing, pp. 25–41.
Asseburg, C., (2005). Modelling uncertainty in multi-species predator–prey
interactions. Unpublished Ph.D. thesis, University of St Andrews.
Beddington, J., Hassell, M. P. & Lawton, J. H. (1976). The components of arthropod
predation. II. The predator rate of increase. J. Anim. Ecol., 45, 165–86.
CCAMLR (1982). Convention on the Conservation of Antarctic Marine Living
Resources. http://www.ccamlr.org/pu/e/pubs/bd/pt1p2.htm.
(1985). CCAMLR Ecosystem Monitoring Program: Standard Methods (revised
2004). http://www.ccamlr.org/pu/e/pubs/std-meth04.pdf
Constable, A. J. (2001). The ecosystem approach to managing fisheries: achieving
conservation objectives for predators of fished species. CCAMLR Sci., 8, 37–64.
Constable, A. J., de la Mare, W. K., Agnew, D. J., Everson, I. & Miller, D. (2000).
Managing fisheries to conserve the Antarctic marine ecosystem: practical
implementation of the Convention on the Conservation of Antarctic Marine
Living Resources (CCAMLR). ICES J. Mar. Sci., 57, 778–91.
Croxall, J. P., Reid, K. & Prince, P. (1999). Diet, provisioning and productivity
responses of marine predators to differences in availability of Antarctic krill.
Mar. Ecol. Prog. Ser., 177, 115–31.
274 C. Asseburg et al.

Davoren, G. K. & Montevecchi, W. A. (2003). Signals from seabirds indicate

changing biology of capelin stocks. Mar. Ecol. Prog. Ser., 258, 253–61.
Dierenfeld, E. S., Alcorn, M. L. & Jacobson, K. L. (2002). Nutrient Composition of
Whole Vertebrate Prey (Excluding Fish) Fed in Zoos. Beltsville, MD: AWIC, US
Department of Agriculture.
Gentleman, W., Leising, A., Frost, B., Strom, S. & Murray, J. W. (2003). Functional
responses for zooplankton feeding on multiple resources: a review of
assumptions and biological dynamics. Deep-Sea Res. II, 50, 2847–75.
Geromont, H. F., de Oliveira, J. A. A., Johnston, S. J. & Cunningham, C. L. (1999).
Development and application of management procedures for fisheries in
southern Africa. ICES J. Mar. Sci., 56, 952–66.
Ginzburg, L. R. (1998). Assuming reproduction to be a function of consumption
raises doubts about some popular predator–prey models. J. Anim. Ecol., 67,
325–7.
Harwood, J. & Stokes, K. (2003). Coping with uncertainty in ecological advice:
lessons from fisheries. Trends Ecol. Evol., 18, 617–22.
Park, K. J., Hurley, M. M. & Hudson, P. J. (2002). Territorial status and survival in
red grouse Lagopus lagopus scoticus: hope for the doomed surplus? J. Avian
Biol., 33, 56–62.
Punt, A. E. & Hilborn, R. (1997). Fisheries stock assessment and decision analysis:
the Bayesian approach. Rev. Fish Biol. Fish., 7, 35–63.
Punt, A. E. & Smith, A. D. H. (1999). Harvest strategy evaluation for the eastern
stock of gemfish (Rexea solandri). ICES J. Mar. Sci., 56, 860–75.
Real, L. A. (1977). The kinetics of functional response. Am. Nat., 111, 289–300.
Redpath, S. & Thirgood, S. (1999). Numerical and functional responses in
generalist predators: hen harriers and peregrines on Scottish grouse moors. J.
Anim. Ecol., 68, 879–92.
Reid, K. & Arnould, J. (1996). The diet of Antarctic fur seals Arctocephalus gazella
during the breeding season at South Georgia. Polar Biol., 16, 104–14.
Reid, K. & Croxall, J. P. (2001). Environmental response of upper trophic-level
predators reveals a system change in an Antarctic marine ecosystem. Proc. R.
Soc. Lond. B, 268, 377–84.
Stefansson, G. (2003). Multi-species and ecosystem models in a management
context. In Responsible Fisheries in the Marine Ecosystem, eds. M. Sinclair & G.
Valdimarsson. Wallingford, UK: CABI Publishing, pp. 171–88.
Thompson, R. B., Butterworth, D. S., Boyd, I. L. & Croxall, J. P. (2000). Modeling
the consequences of Antarctic krill harvesting on Antarctic fur seals. Ecol.
Applic., 10, 1806–19.
Turchin, P. (2003). Complex Population Dynamics. Princeton, NJ: Princeton
University Press.
United Nations (1995). The United Nations Agreement for the Implementation of the
Provisions of the United Nations Convention on the Law of the Sea of 10 December
1982 relating to the Conservation and Management of Straddling Fish Stocks and
Highly Migratory Fish Stocks (in force as from 11 December 2001)
(http://www.un.org/Depts/los/convention agreements/convention overview
fish stocks.htm)
Yodzis, P. (1998). Local trophodynamics and the interaction of marine mammals
and fisheries in the Benguela ecosystem. J. Anim. Ecol., 67, 635–58.
 19
The method of multiple hypotheses and the
decline of Steller sea lions in western Alaska
N. WOLF, J. MELBOURNE AND M. MANGEL

In recent years, enormous effort has been expended to explain the cause
of the precipitous decline of the western population of Steller sea lions
(Eumatopias jubatus) since the late 1970s; however, despite these efforts and
the proposal of a wide variety of hypotheses, the decline has proven to be
very difficult to explain. The authors of a recent comprehensive review of the
problem emphasized repeatedly that the system is in dire need of a mod-
elling approach that takes advantage of the data available at small spatial
scales (at the level of the rookery). We view this as an opportunity for eco-
logical detection, a process in which multiple hypotheses simultaneously
compete and their success is arbitrated by the relevant data. We describe
ten hypotheses for which there are sufficient data to allow investigation, a
method that allows one to link various sources of data to the hypotheses and
the conclusions from this approach.
The decline of the western Alaska population of Steller sea lions has
proven to be very difficult to explain, in part because most aspects of the
population and the environmental variables proposed to explain its decline
involve a combination of high spatial and temporal variability, and limited
data. Consequently, most researchers pooled data across rookeries or across
time, obscuring spatial and/or temporal patterns (Fig. 19.1). Some of these
previous studies are described below.
(1) Construction of a Leslie matrix model for a stable population, followed
by perturbation of various transition rates to find the most
parsimonious way to produce a trajectory matching the observed
decline. York (1994) determined that the initial decline could be

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
276 N. Wolf et al.

(a)
120000

100000

80000
Population

60000

40000

20000

0
1975 1980 1985 1990 1995 2000 2005
Year

(b)
2002 8506
6512
4785
3323
2127
1196
532
133
0
No data

Year

1973
West SSL rookeries East

Fig. 19.1 (a) Composite time series of the Steller sea lion (SSL) population in
western Alaska. (b) Space–time plot of the counts of non-pups at 38 individual
rookeries from 1973 to 2002. The abscissa indicates rookeries from west to east,
with each column representing a different rookery. The ordinate indicates time,
with each row representing a single year. The area of each circle indicates the
observed number of sea lions at that rookery in that year. The arrows indicate
years in which a ‘synoptic’ survey of the entire population was taken. Notice,
however, that the dataset is much richer than the synoptic survey (panel a, with
10 points total) would suggest.
 The method of multiple hypotheses 277

explained most easily by a 10% to 20% decrease in juvenile survival.

Pascual and Adkison (1994) estimated the effective mortality and
fecundity rates for six individual rookeries. Adkison et al. (1993) and
Pascual and Adkison (1994) also modified the matrix model by
allowing vital rates to vary according to alternative hypotheses.
(2) Assumption of a fixed set of underlying vital rates and calculation of
the number of animals that would have to be removed in order to
match the observed census data (e.g. Blackburn 1990 (cited in
Castellini 1993), Loughlin & York 2002, NRC 2003.
(3) Construction of a simulation model that includes the hypothesized
effect (e.g. Barrett-Lennard et al. 1995, NRC 2003).
(4) Analysis of trends across space rather than across time. Merrick et al.
(1997) observed a correlation between population growth rate and diet
diversity among different rookeries.

None of these studies made use of both spatial and temporal variation in sea
lion counts and environmental data. Researchers either pooled data across
space, combining all rookery censuses within each year and performing
their analyses using a composite time series representing the entire west-
ern population (Fig. 19.1); or else pooled data across time, treating only the
overall trend of the census at each rookery. A recent review of the problem
(NRC 2003) emphasized that the system requires a modelling approach that
takes advantage of the data available at small spatial scales. Here we give a
précis of the results of such a study (Wolf & Mangel 2004)

POPULATION BIOLOGY OF THE STELLER SEA LION

Genetic (Bickham et al. 1998) and behavioural (Raum-Suryan et al. 2002)

evidence suggest that 144◦ W longitude separates distinct populations and
that the rookeries within each region qualify as a meta-populations (York
et al. 1996). The eastern population is estimated to have been growing
slowly since survey methods were standardized in the 1970s, but the west-
ern population declined by more than 80% (Fig. 19.1). The decline appears
to have begun in the eastern Aleutians and spread from there, with asso-
ciated changes in size-at-age and condition (Castellini 1993, Calkins et al.
1998, Sease & Loughlin 1999, Andrews et al. 2002). The western stock was
listed as ‘Endangered’ in 1997.
Steller sea lions are largely opportunistic foragers. Walleye pollock
(Theragra chalcogramma) are currently the principal diet component; they
have also been the most abundant prey species since the mid 1970s,
278 N. Wolf et al.

when the North Pacific climate switched to a warm regime favouring

pollock (Alverson 1992, Anderson & Blackburn 2002). Other important
prey species include Atka mackerel (Pleurogrammus monopterygius), Pacific
cod (Gadus macrocephalus) and Pacific herring (Clupea pallasi).

RELEVANT FISHERIES

Walleye pollock, Atka mackerel and Pacific cod are harvested primarily
using groundfish trawling gear. The largest groundfish harvests in the area
occur in the Bering Sea. The peak catch occurred in 1972, followed by a
decline in the late 1970s and a recovery in the mid 1980s. Pollock comprise
over 76% of the groundfish caught in the Bering Sea (NRC 2003). In the
Gulf of Alaska, the pollock catch peaked between 1976 and 1985. A fishery
for groundfish developed in the Aleutian Islands in the late 1970s.

OTHER MARINE MAMMALS

The decline of the Steller sea lion was preceded by declines in populations
of northern fur seal (Callorhinus ursinus) and Pacific harbour seal (Phoca
vitulina) occupying the same region. The causes of these declines remain
similarly unexplained (Merrick 1997, Springer et al. 2003). The range of the
western population is also home to a large population of killer whales (Orci-
nus orca); some of these are ‘transients’, whose diet is thought to consist
mainly of marine mammals, including Steller sea lions (Barrett-Lennard
et al. 1995, Matkin et al. 2002). Unfortunately, very little is known about the
spatial or temporal distribution of these whales. Steller sea lions may com-
prise 5% to 20% of their diet (Matkin et al. 2002). The stomach of one killer
whale that washed up on a beach in British Columbia contained flipper tags
from 14 different Steller sea lion pups, all of which had been tagged at the
Marmot Island rookery 3 to 4 years before (Saulitis et al. 2000).

COMPETING HYPOTHESES

Competing hypotheses have been proposed to explain the decline of the

western population. Sufficient data exist to explore ten hypotheses with
alternative models and rank them according to their explanatory power
(Hilborn & Mangel 1997).

H1 to H3: food limitation hypotheses

We assume that fecundity (H1), pup recruitment (H2) or non-pup sur-

vival probability (H3) is a positive function of the local encounter rate with
 The method of multiple hypotheses 279

groundfish prey. Specifically, starvation (H2, H3) or termination of preg-

nancy (H1) occur if an animal experiences a long series of unsuccessful for-
aging attempts. Under poor foraging conditions, animals may lose condi-
tion because they consume less prey, spend more time and energy hunting,
or both. Body condition, in turn, is a significant determinant of the probabil-
ity that a pregnant female actually completes her pregnancy and produces a
pup (Pitcher et al. 1998). Poor foraging conditions also increase the proba-
bility of starvation and expose the animals to additional predation risk dur-
ing any extra time spent foraging, leading to elevated mortality rates. The
probability of pup recruitment may be linked indirectly to prey availability
if mothers are more likely to abandon pups under poor foraging conditions,
or directly when the inexperienced pups begin foraging for themselves near
the end of their first year.

H4 to H6: ‘Junk-food’ hypotheses

We assume that fecundity (H4), pup recruitment (H5) or non-pup sur-

vival probability (H6) is a positive function of the local encounter rate
with groundfish prey other than walleye pollock. Specifically, starvation
(H5, H6) or termination of pregnancy (H4) occur with higher probability
when alternative prey are scarce. The fact that a pollock-intensive diet might
lead to poor body condition and depressed vital rates was first proposed by
Alverson (1992), and was supported by evidence from captive sea lions that
lost weight on a diet of pollock alone (Rosen & Trites 2000) and from correl-
ative studies of diet diversity and population decline (Merrick et al. 1997).
Pups, with limited dive depth, may be especially sensitive to the species
composition of the prey base because some prey types are probably inacces-
sible to them.

H7 and H8: fishery-related mortality hypotheses

We assume that survival probability of pups (H7) or non-pups (H8) is a

declining function of the local encounter rate with groundfish trawling
operations. Incidental mortality, usually entanglement of sea lions in fish-
ing gear, is now estimated to be killing less than 100 animals per year (Perez
& Loughlin 1991, Loughlin & York 2002), but in the past it was much higher
(NRC 2003). It was legal to shoot sea lions in defence of gear until 1990,
and there are anecdotal reports suggesting that shooting (even unrelated to
defence of gear) may still occur (NRC 2003). It may be very difficult to deter-
mine whether incidental or deliberate mortality is the problem, since both
might scale with fishing effort. However, it seems likely that entanglement
280 N. Wolf et al.

would be a bigger problem for naı̈ve pups (Loughlin et al. 1983), whereas
adults are more likely to be targeted by shooters.

H9 and H10: predation–mortality hypotheses

Transient killer whales are predicted to prey upon sea lions or their pups
when the whales’ preferred prey, harbour seals, are scarce (Springer et al.
2003, Mangel & Wolf in press). Therefore, survival probability of sea lion
pups (H9) or non-pups (H10) is predicted to decline if local harbour seal
density falls below a threshold value.

THE APPROACH

The hypotheses and data were linked by the following procedure (all the
details can be found in Wolf and Mangel (2004)).

(1) We formulated the alternative hypotheses as one-parameter scaling

functions that modify vital rates according to local conditions.
(2) We sorted the data concerning ‘local conditions’ by rookery and year.
(3) We formulated a suitable population model, with appropriate process
uncertainty and observation error (described below).
(4) We calculated the likelihood for all possible values of all parameters
(one per hypothesis) simultaneously.
(5) We ranked the effects in terms of statistical support and strength of the
effect.

In contrast to all previous modelling approaches, this procedure does not

require a complete dataset, and makes use of variation both between rook-
eries and across time.
In the population model (step 3), we assume two age classes: pups and
non-pups. The population dynamics (Box 19.1) can then be characterized
by the survival and fecundity of non-pups and the survival of pups to the
non-pup stage (recruitment). There are thus three parameters that charac-
terize how population size changes from one year to the next. Because each
of these parameters must be between 0 and 1, process uncertainty (sensu
Hilborn and Mangel (1997)) is captured using binominal transitions. In
particular, three binomial distributions represent all possible transitions in
the model: recruitment (the number of pups that survive from the previous
year, with rate parameter ρ), survival (the number of non-pups that sur-
vive, with rate parameter σ ) and fecundity (the number of non-pups that
give birth to new pups, with rate parameter φ). The probability distributions
 The method of multiple hypotheses 281

Box 19.1 The population model

The population model uses two age classes: pups and non-pups. The
number of pups in a particular year at a particular rookery, J(i, t), is
determined by the per-capita probability of reproduction by female
non-pups, N(i, t), in that year at the same rookery. The number of
non-pups in a particular year is determined by the survival of non-
pups from the previous year and the recruitment (survival to age
1) of pups produced in the previous year. These transitions capture
process uncertainty through a set of nested binomial distributions.
Although the true number of pups can be estimated by aerial survey,
the observed number of non-pups, Nobs (i, t), is smaller than the true
number – so a beta-binomial model is used to account for observa-
tion error (see Fig. 19.2).

for the numbers of non-pups (N) and pups (J) at time t, given the numbers
at time t − 1, are computed directly from the associated binomial distribu-
tions.
Although pups can be accurately counted during aerial surveys, some
fraction of the non-pups are likely to be at sea foraging during the surveys.
Thus, the observed number of non-pups is smaller than the true number
of non-pups. We characterize this observation uncertainty using a beta-
binomial distribution (Martz & Waller 1982, Evans et al. 2000, Wolf &
Mangel 2004).
We set fixed background values for the life-history parameters (denoted
by ρ 0 , σ 0 and φ 0 ) that are modified by local conditions to reflect a particular
hypothesis. We chose background values from observed values measured
on the Marmot Island rookery by Calkins and Pitcher (1982) and used by
York (1994), and Pascual and Adkison (1994). The annual growth rate of a
population using these values is 0.4% (Pascual & Adkison 1994). Fecundity
estimates also account for only about 50% of the pups being female (York
1994, NRC 2003) and for about 50% of the non-pup population being juve-
nile (Holmes & York 2003).
The background values for life-history parameters are then modified
to take account of local conditions. For the case of food limitation (H1 to
H3), the modification is that low abundance of groundfish and other prey
causes local fecundity (hypothesis 1), pup recruitment (hypothesis 2), or
non-pup survival probability (hypothesis 3) to be diminished. These are
282 N. Wolf et al.

J(i, t −1) N true(i, t −1)

Time

J (i, t ) N true(i, t ) N obs(i, t )

J (i, t +1) N true(i, t + 1)

J (i, t +2) N true(i, t + 2) N obs(i, t + 2)

Fig. 19.2

characterized by a Holling Type III functional response (Holling 1959;

Box 19.2). The choice of a sigmoid function implies that sea lions are
unlikely to find enough to eat when prey are scarce, and unlikely not to
find enough to eat when prey are abundant.
For the case of the junk-food hypotheses (H4 to H6), the modification of
the background life-history parameters is that a high proportion of walleye
pollock among the available prey causes local fecundity (hypothesis 4), pup
recruitment (hypothesis 5), or non-pup survival probability (hypothesis 6)
to be diminished. We model this as the fraction of non-pollock food raised
to an unknown power (Box 19.2).
For the case of fishery-related mortality (H7 and H8), we assume that
the survival probability of pups (hypothesis 7) or non-pups (hypothesis 8)
is diminished in areas where there is more commercial fishery activity
(Box 19.2 and Fig. 19.3c).
 The method of multiple hypotheses 283

Box 19.2 Hypotheses to explain the decline

We formulated ten different hypotheses as functions relating life-
history parameters of Steller sea lions to local conditions. Function i
(i = 1, 2, 3, . . ., 10) is parameterized with a single unknown constant,
ci , in such a way that a zero value for the constant indicates no effect
for hypothesis i. Hypotheses 1 to 3 are related to food; specifically,
they propose that low prey abundance leads to diminished fecundity,
pup recruitment or non-pup survival rates. These are captured as
separate Holling Type III functional-response curves, with the half-
saturation values denoted by c1 , c2 and c3 respectively. Hypotheses 4
to 6 propose that an elevated fraction of pollock among available prey
in the environment leads to diminished fecundity, pup recruitment
or non-pup survival, respectively, for the sea lions. These effects are
captured by raising the fraction of non-pollock in the environment
to the power ci (i = 4, 5, 6), where ci ≥ 0. Hypotheses 7 and 8
describe the possibility that increased fishing activity near a rook-
ery leads to diminished pup recruitment (hypothesis 7) or non-pup
survival (hypothesis 8). We assume that this is characterized by a
negative exponential distribution with parameter ci ≥ 0. Hypotheses
9 and 10 are related to predation pressure by killer whales. They are
investigated by using optimal diet theory (Mangel & Wolf in press)
in which it is assumed that predators consume sea lions (depress-
ing recruitment or survival rates) when the density of the more prof-
itable prey (harbour seals) falls below a critical value. The survival
rate is depressed by ci (i = 9, 10) (see Fig. 19.3).

Finally, for the case of the predation-mortality hypotheses (H9 and

H10), we assume that killer whales broaden their diet to include Steller
sea lions when harbour seals are scarce, so that the survival probability
of sea lion pups (hypothesis 9) or non-pups (hypothesis 10) decline when
local harbour seal populations fall below a critical value (Mangel & Wolf,
in press). In particular, we set the critical harbour seal density by com-
paring the densities of harbour seals around sea lion rookeries with rising
and falling populations, and choosing a point between these distributions.
(If killer whale predation is a significant source of mortality for Steller sea
lions, then the densities of harbour seals around sea lion rookeries with
284 N. Wolf et al.

1 1
(a) (b)

Multiplier
Multiplier
c = 400 000 c = 0.04
c = 2 000 000 c = 0.2
c = 10 000 000 c=1
c = 50 000 000 c=5

0 0
Prey density Pollock fraction

1 1
(c) (d)

Multiplier
Multiplier

c = 0.000 000 000 08 1−c

c = 0.000 000 000 4
c = 0.000 000 002
c = 0.000 000 01

0 0
Fishing activity Harbour seal density

Fig. 19.3

rising and falling populations should tend to be above and below the criti-
cal value respectively.)
Thus, the local value of each vital rate is calculated by multiplying the
background rate by all the relevant scaling factors. Note that there are ten
unknown parameters, one per hypothesis. In each case, a parameter value
of zero indicates that the corresponding hypothesis has no effect.

SOURCES OF DATA

In order to test the various hypotheses using the model, we acquired

data on relevant prey species from triennial groundfish survey results
collected by the National Marine Fisheries Service/Alaska Fisheries Sci-
ence Center (NMFS/AFSC) in the Gulf of Alaska/Aleutian Islands
(http://www.afsc.noaa.gov/race/groundfish/default gf.htm). We calculated
estimates of fishing activity in minutes per year from the NMFS groundfish
fishery observer database. This database covers foreign and joint-venture
groundfish fisheries from 1973 to 1991 and domestic fisheries from 1986
to 2001.
Estimates of harbour seal density came from online NMFS/AFSC
marine mammal stock assessments and reports (Withrow et al. 2000,
2001, 2002, Angliss & Lodge 2002), a Marine Mammal Commission
report (Hoover-Miller 1994), and eight journal articles (Bailey & Faust
 The method of multiple hypotheses 285

1980, Everitt & Braham 1980, Pitcher 1990, Frost et al. 1999, Mathews &
Pendleton 2000, Jemison & Kelly 2001, Boveng et al. 2003, Small et al.
2003). The Steller sea lion counts were from the NMFS/AFSC/National
Marine Mammal Laboratory (NMML) online database (http://nmml.
afsc.noaa.gov/AlaskaEcosystems/sslhome/stellerhome.html). We limited
our consideration to year–rookery combinations in which counts from June
or July were available for both pups and non-pups, and to rookeries for
which such censuses from at least two different years were available. When
more than one count was available for a particular rookery in a single year,
we took the average. Several sets of adjacent rookeries were censused as one
large rookery early in the dataset and as separate rookeries in later years. In
some of these cases, we combined the counts from the separate rookeries
in later years in order to extend the time series for the ‘joint’ rookery. When
prey abundance or harbour seal density estimates were missing for certain
area–year combinations, we used linear interpolation to estimate the miss-
ing value from reported values in earlier and later years for the same area.
Further details about the acquisition and treatment of data are found in Wolf
and Mangel (2004).

RESULTS

We estimated the unknown parameters by comparing the predictions of

the stochastic population model with the observed counts. To do this we
started with the beta-binomial observation error distribution and a two-life-
stage stochastic population model employing the local vital rates, and cal-
culated the probability of observing the sequence of reported pup and non-
pup counts at a particular rookery, given: (a) relevant local conditions and
(b) a particular set of parameter values in the hypothesized equations. We
then computed the maximum-likelihood estimate (MLE) of each parameter
and constructed ten one-dimensional profile likelihoods (Hilborn & Mangel
1997) so that we could examine the support for each parameter, holding the
others at their MLEs. For each parameter, we computed a profile-likelihood
interval by finding the area under the curve that contains 95% of the total
area.
The result of these computations is shown in Fig. 19.4. Each column
represents one class of hypothesized impacts (overall abundance of food,
fraction of food that is pollock, fishing activity or predation) and the rows
represent the hypothesized effects on fecundity, recruitment or survival.
Each plot shows the profile likelihood for the relevant parameter in the
functional form. A peak at a non-zero value indicates support for the
286 N. Wolf et al.

0 0
Fecundity

1 4
0 0 0 0
Recruitment

2 5 7 9
0 0 0 0
Survival

3 6 8 10
Food Pollock Boats Seals
Fig. 19.4 Likelihood profiles for the parameters associated with the functional
forms for the four classes of hypotheses and their effects on the relevant
life-history parameters. See text for further details.

corresponding effect. We therefore find: strong support for hypotheses 1,

5 and 10; weak support for hypothesis 4; and no support for the other six
hypotheses.
We also computed Aikaike’s information criterion (AIC) weights
(Burnham & Anderson 1998) from the likelihood information. Hypothe-
ses 1, 4, 5 and 10 account for >99% of the AIC weight (and hypothesis 4
only provides 5% of that total). In Fig. 19.5, we show the functional forms
associated with the different hypotheses, evaluated at the MLE values of the
parameters, and in Fig. 19.6 we show the lost production of Steller sea lions
due to the effects in hypotheses 1, 4, 5 and 10.

DISCUSSION

The strong message of ecology is that the world changes and that the rea-
sons for change are manifold. Thus, rather than trying to ‘prove’ one mecha-
nism, we should recognize that multiple mechanisms will almost always be
at work, and we should ask how to weigh the importance of different mech-
anisms. It is this approach that we have taken in understanding the decline
 The method of multiple hypotheses 287

(a) 1 (b) 1

Relative fecundity
Relative fecundity

c1 = 900000 c4 = 0.04

0 0
0 10 000000 20000000 30000000 0 0.2 0.4 0.6 0.8 1
'Food' (CPUE of all prey species in kg km−3) Pollock fraction (by CPUE weight)

(c) 1 (d) 1
Relative pup recruitment

Relative non-pup survival

c 5 = 5.0 c10 = 0.01

0 0
0 0.2 0.4 0.6 0.8 1 0 5000 10000 15000 20000
Pollock fraction (by CPUE weight) Harbour seals within 300 km

Fig. 19.5 Vital rate functions corresponding to MLE parameter values. (a)
Hypothesis 1: food availability affects fecundity (magnitude of effect: medium).
(b) Hypothesis 4: pollock fraction affects fecundity (magnitude of effect: weak at
best). (c) Hypothesis 5: pollock fraction affects pup recruitment (magnitude of
effect: very strong). (d) Hypothesis 10: harbour seals (via predation) affect
non-pup survival (magnitude of effect: weak but persistent).

of the western population of Steller sea lions. There is good evidence for
two strong effects: H1, total prey availability affects fecundity; H5, pollock
fraction in the environment affects pup recruitment. One moderate effect
was found: H10, harbour seal density (predation) affects non-pup survival.
There was also marginal evidence for one weak effect: H4, pollock fraction
in the environment affects fecundity. No evidence was found for any of the
other hypotheses. What our work has done is to guide the weight of the evi-
dence, when all plausible hypotheses are competing, towards those that win
the competition.
Although we used the word mechanism, we recognize that a study such
as this one cannot demonstrate causality. It would seem that H1 is a rela-
tively clear and simple mechanism: lower abundance of all prey types leads
to lower fecundity through the direct effect of reduced resource accumu-
lation by adults and thus reduced storage for reproduction. H10 also has
288 N. Wolf et al.

2000–01 0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0
No data
Year

1980–81
West SSL rookeries East
Fig. 19.6 Lost production of the Steller sea lion population due to hypotheses 1,
4, 5 and 10. The diameter of each circle represents the fraction of potential
production lost.

a clear mechanism, but note that its MLE is about 0.01, so that the effect
of changes in the breadth of the diet of killer whales leads to only a 1%
reduction in non-pup survival, and then only in cases where harbour seal
numbers are sufficiently low.
On the other hand, H4 and H5 are more complicated. A high pollock
fraction can result either from high pollock or from low non-pollock, and
either of these could be the underlying factor. Furthermore, the mechanism
might be something completely different for which pollock fraction is only
a correlate. For example, juveniles may require some easily caught subset
of prey species because they are unable to dive deep enough or swim fast
enough to catch anything else. (In the current dataset, the fish biomass is
not broken down by size class of fish. However, the really small fish and the
really big ones are probably not useful to sea lions. Thus, some additional
thinking is required about how to modify the survey data to address this
question.)
Our results also suggest an adaptive management plan in which one
designates the areas around some of the rookeries as experimental zones
in which to make fishery quotas contingent upon the results of pre-fishing-
season survey trawls. We envision a series of treatments:

r rookeries around which fishing is not affected by the pre-season survey

information (control type 1);
r rookeries around which no fishing occurs (control type 2);
 The method of multiple hypotheses 289

r rookeries around which fishing is reduced or prohibited if the total

prey biomass in the pre-season zone is below a critical threshold
(determined by c1 ).
r rookeries around which a directed pollock fishery occurs if the
pre-season survey suggests pollock fraction is above a critical threshold
(determined by c5 ).

This combination of rookery types would allow sufficient variation in treat-

ment, which is crucial in adaptive management. Sea lion vital rates would
be monitored in the same areas to see if the management plan was having
a positive effect. Before organizing any adaptive management, it would be
possible to use our model to simulate forward and suggest a time scale over
which results might be expected to appear.
Our results also suggest a form of ‘adaptive observation’: identify rook-
eries with high numbers and low numbers of harbour seals (regardless of
the number of sea lions). The prediction of H10 is that the per-capita attack
rate of killer whales on sea lions will be higher around rookeries where har-
bour seal densities are low. Careful monitoring of killer whale attack rates
would provide a natural test of hypothesis 10: if low harbour seal numbers
are associated with declining sea lions, but not with elevated killer whale
attack rates, then there must be some other factor to explain the observed
correspondence between low harbour seals and declining sea lions.
The question ‘Is it food?’ has been asked a number of times in the con-
text of the decline of Steller sea lions. As with most questions in biology, we
shall never be able to ‘prove’ that it is lack of food. However, we conclude
that the weight of the current evidence is that it is indeed food – and both
the quantity and quality of the food matters. The more recent question ‘Is
it killer whale predation?’ can be answered too: sometimes, if harbour seal
populations are sufficiently low; however, the predation does not cause a
large reduction in survival, but has a persistent annual effect.

ACKNOWLEDGEMENTS

This research was supported in part by the Alaska Fisheries Science Cen-
ter of NOAA Fisheries, Department of Commerce under Contract AB133F-
02-CN-0085: ‘Alternative Management Strategies for the Recovery of the
Stellar Sea Lion’; and in part by the Center for Stock Assessment Research,
University of California Santa Cruz. We thank Dan Goodman and Dan
Hennen for providing the commercial fishery and NMFS trawl data. For
various helpful discussions we thank Doug DeMaster, Anne York, Lowell
290 N. Wolf et al.

Fritz, Tim Ragen, Gunnar Steffanson, Jim Estes and the entire Mangel lab-
oratory, but especially Steve Munch. We appreciate the use of a computer
from the laboratory of Ingrid Parker.

REFERENCES
Adkison, M. D., Pascual, M. A., Hilborn, R. et al. (1993). Modeling the trophic
relationships between fish and marine mammal populations in Alaskan
waters. In Is it food? Addressing Marine Mammal and Seabird Declines, ed. S.
Keller. Fairbanks, Alaska: University of Alaska Sea Grant College Program,
pp. 54–6.
Alverson, D. L. (1992). A review of commercial fisheries and the Steller sea lion
(Eumetopias jubatus): the conflict arena. Rev. Aquat. Sci., 6, 203–56.
Anderson, P. J. & Blackburn, J. E. (2002). Status of demersal and epibenthic
species in the Kodiak Island and Gulf of Alaska region. In Steller Sea Lion
Decline: Is it Food II, eds. D. DeMaster & S. Atkinson. Fairbanks, Alaska:
University of Alaska Sea Grant College Program, pp. 57–60.
Andrews, R. D., Calkins, D. G., Davis, R. W. et al. (2002). Foraging behavior and
energetics of adult female Steller sea lions. In Steller Sea Lion Decline: Is it Food
II, eds. D. DeMaster & S. Atkinson. Fairbanks, Alaska: University of Alaska Sea
Grant College Program, pp. 19–22.
Angliss, R. P. & Lodge, K. L. (2002). Alaska Marine Mammal Stock Assessments,
2002. NOAA Technical Memorandum NMFS AFSC-133. Seattle, WA: US
Department of Commerce.
Bailey, E. P. & Faust, N. H. (1980). Summer distribution and abundance of marine
birds and mammals in the Sandman Reefs, Alaska, USA. Murrelet, 61, 6–19.
Barrett-Lennard, L. G., Heise, K. A., Saulitis, E., Ellis, G. & Matkin, C. (1995). The
Impact of Killer Whale Predation on Steller Sea Lion Populations in British
Columbia and Alaska. Vancouver, BC: North Pacific Universities Marine
Mammal Research Consortium.
Bickham, J. W., Loughlin, T. R., Calkins, D. G., Wickliffe, J. K. & Patton, J. C.
(1998). Genetic variability and population decline in Steller sea lions from the
Gulf of Alaska. J. Mammal., 79, 1390–5.
Blackburn, C. (1990). Sea Lion Briefing Paper. Kodiak, Alaska: Alaska Groundfish
Data Bank.
Boveng, P. L., Bengtson, J. L., Withrow, D. E. et al. (2003). The abundance of
harbor seals in the Gulf of Alaska. Mar. Mammal Sci., 19, 111–27.
Burnham, K. P. & Anderson, D. R. (1998). Model Selection and Inference: A Practical
Information-Theoretic Approach. New York: Springer-Verlag.
Calkins, D. & Pitcher, K.W. (1982). Population assessment, ecology and trophic
relationships of Steller sea lions in the Gulf of Alaska. In Environmental
Assessment of the Alaskan Continental Shelf, Final Report. Washington, DC: US.
Department of Commerce and US Department of the Interior, pp. 447–546.
Calkins, D. G., Becker, E. F. & Pitcher, K. W. (1998). Reduced body size of female
Steller sea lions from a declining population in the Gulf of Alaska. Mar.
Mamm. Sci., 14, 232–44.
Castellini, M. (1993). Report of the Marine Mammal Working Group. In Is it Food?
Addressing Marine Mammal and Seabird Declines, ed. S. Keller. Fairbanks,
Alaska: University of Alaska Sea Grant College Program, pp. 4–11.
 The method of multiple hypotheses 291

Evans, M., Hastings, N. & Peacock B. (2000). Statistical Distributions, 3rd edn. New
York: John Wiley.
Everitt, R. D. & Braham, H. W. (1980). Aerial survey of pacific harbor seals, Phoca
vitulina richardsi, in the southeastern Bering Sea. Northw. Sci., 54, 281–8.
Frost, K. J., Lowry, L. F. & Ver Hoef, J. M. (1999). Monitoring the trend of harbor
seals in Prince William Sound, Alaska, after the Exxon Valdez oil spill. Mar.
Mamm. Sci., 15, 494–506.
Hilborn, R. & Mangel, M. (1997). The Ecological Detective: Confronting Models with
Data. Princeton, NJ: Princeton University Press.
Holling, C. S. (1959). Some characteristics of simple types of predation and
parasitism. Can. Entomol., 91, 385–98.
Holmes, E. E. & York, A. E. (2003). Using age structure to detect impacts on
threatened populations: a case study with Steller sea lions. Conserv. Biol., 17,
1794–806.
Hoover-Miller, A. A. (1994). Harbor Seal (Phoca vitulina) Biology and Management
in Alaska. Washington, DC: Marine Mammal Commission.
Jemison, L. A. & Kelly, B. P. (2001). Pupping phenology and demography of harbor
seals (Phoca vitulina richardsi) on Tugidak Island, Alaska. Mar. Mamm. Sci., 17,
585–600.
Loughlin, T. R. & York, A. E. (2002). An accounting of the sources of Steller sea
lion mortality. In Steller Sea Lion Decline: Is it Food II, eds. D. DeMaster & S.
Atkinson. Fairbanks, Alaska: University of Alaska Sea Grant College Program,
pp. 9–13.
Loughlin, T. R., Consiglieri, L., DeLong, R. L. & Actor, A. T. (1983). Incidental
catch of marine mammals by foreign fishing vessels, 1978–81. Mar. Fish. Rev.,
45, 44–9.
Mangel, M. & Wolf, N. Predator diet breadth and prey population dynamics:
mechanism and modeling. In Whales, Whaling and Ocean Ecosystems, ed. J.
Estes. Berkeley, CA: UC Press, in press.
Martz, H. F. & Waller, R. A. (1982). Bayesian Reliability Analysis. New York: John
Wiley.
Mathews, E. A. & Pendleton, G.W. (2000). Declining Trends in Harbor Seal (Phoca
vitulina richardsi) Numbers at Glacial Ice and Terrestrial Haulouts in Glacier Bay
National Park, 1992–1998. Final Report to Glacier Bay National Park and
Preserve, Resource Management Division. Cooperative Agreement
9910–97–0026. Bedford, VA: National Park Service, US Department of the
Interior.
Matkin, C. O., Barrett-Lennard, L. G. & Ellis, G. (2002). Killer whales and
predation on Steller sea lions. In Steller Sea Lion Decline: Is it Food II, eds. D.
DeMaster & S. Atkinson. Fairbanks, Alaska: University of Alaska Sea Grant
College Program, pp. 61–6.
Merrick, R. L. (1997). Current and historical roles of apex predators in the Bering
Sea ecosystem. J. Northw. Atl. Fish. Sci., 22, 343–55.
Merrick, R. L., Chumbley, M. K. & Byrd, G. V. (1997). Diet diversity of Steller sea
lions (Eumetopias jubatus) and their population decline in Alaska: a potential
relationship. Can. J. Fish. Aquat. Sci., 54, 1342–8.
NRC (National Research Council) (2003). Decline of the Steller Sea Lion in Alaskan
Waters: Untangling Food Webs and Fishing Nets. Washington, DC: National
Academy of Sciences Press.
292 N. Wolf et al.

Pascual, M. A. & Adkison, M. D. (1994). The decline of the Steller sea lion in the
northeast Pacific: demography, harvest, or environment? Ecol. Applic., 4,
393–403.
Perez, M. A. & Loughlin, T. R. (1991). Incidental Catch of Marine Mammals by
Foreign and Joint Venture Trawl Vessels in the U.S. EEZ of the North Pacific,
1973–88. Seattle, WA: US Department of Commerce, National Oceanographic
and Atmospheric Administration, National Marine Fisheries Service.
Pitcher, K. W. (1990). Major decline in the number of harbor seals, Phoca vitulina
richardsi, on Tugidak Island, Gulf of Alaska. Mar. Mammal Sci., 6, 121–34.
Pitcher, K. W., Calkins, D. G. & Pendleton, G. W. (1998). Reproductive
performance of female Steller sea lions: an energetics-based reproductive
strategy? Can. J. Zool., 76, 2075–83.
Raum-Suryan, K. L., Pitcher, K. W., Calkins, D. G., Sease, J. L. & Loughlin, T. R.
(2002). Dispersal, rookery fidelity, and metapopulation structure of Steller sea
lions (Eumetopias jubatus) in an increasing and decreasing population in
Alaska. Mar. Mammal Sci., 18, 746–64.
Rosen, D. A. S. & Trites, A. W. (2000). Pollock and the decline of Steller sea lions:
testing the junk-food hypothesis. Can. J. Zool., 78, 1243–50.
Saulitis, E. L., Matkin, C. O., Heise, K. A., Barrett-Lennard, L. G. & Ellis, G. M.
(2000). Foraging strategies of sympatric killer whale (Orcinus orca) populations
in Prince William Sound, Alaska. Mar. Mammal Sci., 16, 94–109.
Sease, J. L. & Loughlin, T. R. (1999). Aerial and Land-based Surveys of Steller Sea
Lions (Eumetopias jubatus) in Alaska, June and July 1997 and 1998. NOAA
Technical Memorandum NMFS-AFSC-100. Washington, DC: US Department
of Commerce.
Small, R. J., Pendleton, G. W. & Pitcher, K. W. (2003). Trends in abundance of
Alaska harbor seals, 1983–2001. Mar. Mammal Sci., 19, 344–62.
Springer, A. M., Estes, J. A., Vliet, G. B. van et al. (2003). Sequential megafaunal
collapse in the North Pacific Ocean: an ongoing legacy of industrial whaling?
Proc. Natl. Acad. Sci. U.S.A., 100, 12 223–8.
Withrow, D. E., Cesarone, J. C., Jansen, J. K. & Bengston, J. L. (2000). Abundance
and distribution of harbor seals (Phoca vitulina) along the Aleutian Islands
during 1999. In Marine Mammal Protection Act and Endangered Species Act
Implementation Program 1999, eds. A. L. Lopez & D. P. DeMaster. AFSC
Processed Report 2000–11. Seattle, WA: Alaska Fisheries Science Center,
pp. 91–115.
(2001). Abundance and distribution of harbor seals (Phoca vitulina richardsi) in
Bristol Bay and along the north side of the Alaska peninsula during 2000. In
Marine Mammal Protection Act and Endangered Species Act Implementation
Program 2000, eds, A. L. Lopez & R. P. Angliss. AFSC Processed Report
2001–06. Seattle, WA: Alaska Fisheries Science Center.
Withrow, D. E., Cesarone, J. C., Hiruki-Raring, L. & Bengston, J. L. (2002).
Abundance and distribution of harbor seals (Phoca vitulina) in the Gulf of
Alaska (including the south side of the Alaska Peninsula, Kodiak Island, Cook
Inlet and Prince William Sound) during 2001. In Marine Mammal Protection
Act and Endangered Species Act Implementation Program 2001, eds. A. L. Lopez &
S. E. Moore. AFSC Processed Report 2002–06. Seattle, WA: Alaska Fisheries
Science Center.
 The method of multiple hypotheses 293

Wolf, N. & Mangel, M. (2004). Understanding the Decline of The Western Alaskan
Steller Sea Lion: Assessing the Evidence Concerning Multiple Hypothesis. Report of
MRAG Americas. Tampa, FL.: MRAG Americas.
http://www.soe.ucsc.edu/∼msmangel/Stellerfinal.pdf.
Figures at http://www.soe.ucsc.edu/∼msmangel/FiguresFinal.pdf.
York, A. E. (1994). The population dynamics of northern sea lions, 1975–1985. Mar.
Mammal Sci., 10, 38–51.
York, A. E, Merrick, R. L. & Loughlin, T. R. (1996). An analysis of the Steller sea
lion metapopulation in Alaska. In Metapopulations and Wildlife Conservation, ed.
D. R. McCullough. Washington, DC: Island Press, pp. 259–92.
 20
Modelling the behaviour of individuals
and groups of animals foraging in
heterogeneous environments
J . G . O L L A S O N , J . M . Y E A R S L E Y, K . L I U A N D N . R E N

We present an individual-based model of an animal that forages in a spa-

tially explicit environment for food which it uses to maintain itself. The
model subsumes optimal foraging theory as a special case of a general
dynamical theory of foraging, capable of predicting both the transient
behaviour and the steady-state behaviour of the forager in heterogeneous
environments that vary with time. It also predicts aspects of the social struc-
turing of populations of competing foragers, and can do so in environ-
ments containing food that is ingested continuously or as individual par-
ticles. The model has been elaborated to represent the collection of food by
diving seabirds, treating the collection of oxygen between dives as the col-
lection of a second nutrient from a continuous patch. The models provide
the basic building blocks of individual-based models of populations of ani-
mals which can predict the spatial disposition of populations of animals in
environments in which the resources necessary for life are not uniformly
distributed.
In order to understand the relationship between the spatial distribution
of animals and the spatial distribution of their food supply, it is insufficient
simply to assume that there will be a correlation between the standing crop
of prey and the standing crop of the predators that feed on it. What matters
to the individual animal is availability of food, and that will be determined by
its own biological properties and the biological properties of its prey. If the
movements of populations are primarily caused by changes in the spatial

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Modelling foraging behaviour of individuals and groups 295

distribution of food, these movements will be caused by the movements

of individual animals making decisions in the light of their requirements
for food, and their experience of finding, or failing to find it in a partic-
ular place. To understand how such decisions are made, one approach is
to construct a model of an individual forager, and to explore – by simula-
tion – the effects that the decisions of individual members of a population
have on the spatial disposition of that population. To model an individual
forager it is necessary to include sufficient information about the individ-
ual’s need for food, to provide appropriate rules for making decisions, and
to simulate a population of such individuals to determine the kinds of deci-
sions that lead to the observed spatial disposition of individuals in relation
to their observed supply of food. The models described in this chapter were
developed to predict the time-budgets of diving seabirds preying on partic-
ulate food distributed in the water column. The models are elaborations of
a simple model based on a minimal set of assumptions that must be sat-
isfied for the construction of a coherent model. The set contains assump-
tions about the collection and utilization of food, about the effect of foraging
behaviour on the food in the environment, and about the indirect effects
of animals foraging together on the same source of food. These assump-
tions are instantiated into two classes of model: one predicting the foraging
behaviour of an animal that can feed continuously at sources of food, as for
example a bee or humming bird feeds on nectar from a flower; the other
predicting the foraging behaviour of an animal that feeds discontinuously
on particles, as sticklebacks feed on Daphnia. The two classes of model are
combined to produce a model of a seabird that forages for particulate food,
at depth in the water column, and is constrained by the need to surface
to breathe. Evidence will be provided to show that the foraging behaviour
emerging from the assumptions approximates closely to that observed in
real animals, and that the approach is readily extended to model the for-
aging behaviour of a named seabird, and by implication the foraging of a
population of seabirds exploiting a spatially heterogeneous distribution of
prey.

FOUNDATIONS

The fundamental properties of the models are defined by a set of six

assumptions which we regard as the minimal set of assumptions permit-
ting the construction of a coherent model of a foraging individual (Ollason &
Yearsley 2001).
296 J. G. Ollason et al.

(1) Animals eat food, part of which is incorporated into the animal
(requirement of consumption).
(2) To maintain itself an animal utilizes body tissue, the equivalent of
stored food, and the rate of utilization is dependent on the mass of the
animal (requirement of utilization).
(3) An animal will reject food if the rate at which the food can be eaten
is less than the rate of utilization of body tissue (requirement of
hunger).
(4) By eating, an animal depletes the standing crop of food (requirement
of depletion).
(5) The rate at which an animal feeds is a function of the standing crop of
food (requirement of ingestion).
(6) The rate at which an animal feeds is a function of the density of all of
the animals feeding on the same food (requirement of competition).

The set of assumptions defines the class of minimal models of forag-

ing individuals. This approach differs from optimal foraging theory in rec-
ognizing that animals use the food they collect. The assumptions are also
conservative in the sense that food consumed by the individual is explicitly
removed from the environment. All of the models that follow satisfy the
assumptions; but they represent the mathematically simplest cases, in that
all the functional relationships are linear.
The requirement of consumption, when feeding is continuous, is rep-
resented by the following differential equation:
dm dφ
= − rm (20.1)
dt dt
where m(t) is the mass of the animal’s body, dφ/dt is the rate of feeding of
the animal and r is the specific metabolic rate, the rate of use of food per
unit mass for maintenance.
The requirement of hunger implies that the animal will not feed unless
dφ/dt > r m. The simplest case to be considered is feeding at a constant rate.
Such a system can be represented by the following coupled equations:
dF
= −v (20.2)
dt
dm
= v − rm (20.3)
dt
where v is the rate of feeding, and dF /dt is the rate of change of food. In
the following discussion animals engage in separate phases of activity such
as feeding, or searching for the next item of food, or travelling; and these
activities take place sequentially. At the beginning of each activity, time is
 Modelling foraging behaviour of individuals and groups 297

rescaled to zero so that the timing of the current activity starts when the
activity begins.

EXPLOITING A PATCHY ENVIRONMENT

Suppose that an animal forages in an environment containing patches of

food in two-dimensional space, which regenerate at a constant rate. The
feeding can be represented by the following equations:
dF
= a − vF (20.4)
dt
dm
= vF − r m (20.5)
dt
where a is the rate of regeneration of the patch. Notice that the requirement
of competition is implicit in equation (20.4). If n individuals are feeding
together at the patch, each individual feeds at the rate of vF and the total
rate of removal of food is nvF. The standing stock of food is reduced more
rapidly when n is large than when n is small, and this amounts to exploita-
tion competition at the patch.
Notice that this implies that the animal has a Holling Type I functional
response. In an environment containing just two types of patch, patch 0
and patch 1, regenerating at the rates a0 and a1 , separated by a travelling
time of tt , Ollason and Yearsley (2001) proved the following results:
tc0 a0
lim = (20.6)
r →0,v →0 tc1 a1
tc0 a0
lim = (20.7)
r →0,tt →0 tc1 a1
tc0
lim =1 (20.8)
r →0,tt →∞ tc1

where tc0 and tc1 are the steady-state times spent foraging at patch 0 and
patch 1 respectively. The model predicts that the time spent by an animal
foraging at each of the patches is proportional to the rates of regeneration
at the patch provided that the parameters r, v and tt all tend to zero.
When these parameters are increased the proportion of time spent
feeding is decreased, the proportion of time spent travelling between the
patches is increased causing the undermatching of the ideal free distribu-
tion observed by Kennedy and Gray (1993). Equations (20.7) and (20.8) pre-
dict the increasing degree of mismatching observed as the distance between
the feeding patches is increased. These results are also true of environments
containing many patches. Table 20.1 shows the results of simulating the
298 J. G. Ollason et al.

Table 20.1. The proportions of the population of 150

individuals occupying the 4 × 4 matrix of patches. Ideal free
distribution is shown in bold type

0.052 0.103 0.059 0.022

0.052 0.089 0.061 0.028
0.096 0.044 0.007 0.074
0.097 0.063 0.030 0.074
0.015 0.037 0.118 0.066
0.029 0.042 0.105 0.067
0.088 0.081 0.029 0.110
0.077 0.073 0.035 0.081

distribution that arises when 150 individuals are free to forage in an envi-
ronment containing 16 patches arranged in a 4 × 4 matrix. When it leaves,
the individual is free to move at random to any adjacent patch. Each individ-
ual applies the requirement of hunger to determine if it will stay at a patch.
For realistic values of r, v and tt the predictions approximate closely to the
ideal free distribution.
Simulation shows that a Holling Type II functional response, and allo-
metrically scaled metabolic costs, have little qualitative effect on the model
(Ollason & Yearsley 2001). Facilitation, the specific feeding rate of the
individual increasing with the number of foragers at a patch, causes indi-
viduals to travel in flocks, but the average numbers occupying the patches
approximate to the ideal free distribution nevertheless (Fig. 20.1).
It is possible to model animals with differing dominance, with dom-
inants feeding more rapidly than subordinates. Populations made up of
competitors of varying dominance divide so that dominants forage in the
company of dominants, subordinates in the company of subordinates, but
despite this the ideal free distribution is approximated (Fig. 20.2). For fur-
ther details see Ollason and Yearsley (2001). Notice that the decision mak-
ing depends only on m and r. The animal does not need to know how
many patches are present in the environment or the pay-offs at the different
patches. All it needs to do is to leave the current patch at which it is feeding
when its rate of intake is less than its total metabolic rate. All aspects of its
time budget emerge from this single rule.

FEEDING ON PARTICLES OF FOOD

The optimality of the foraging of great tits was defined by a model of con-
tinuous feeding, despite the patches containing a small number of large
 Modelling foraging behaviour of individuals and groups 299

30
25
No. of animals

20
15
10
5
0
2000 3000 4000 5000 6000 7000 8000
Time

Fig. 20.1 The numbers of foragers at patch 0, a0 = 0.01 (solid line) and at
patch 1, a1 = 0.02 (dashed line). The corresponding mean values are shown by
the straight lines. The animals facilitate each other’s feeding rate; consequently,
when one animal leaves a patch, the feeding rates of all the others decrease,
causing most of the remaining animals to leave almost at once. The expected
mean proportion of feeding animals is 1/3 and 2/3 at patch 0 and at patch 1
respectively. The observed means were 0.373 and 0.626 respectively,
representing a small degree of undermatching. From Ollason and Yearsley
(2001).

30

25
Mean rank

20

15

10
6 8 10 12 14 16
No. of animals at patch 0

Fig. 20.2 Individuals with large numerical ranks can feed faster than those with
smaller ranks. When the mean rank of the occupants is large, only a small
number can occupy a patch, driving the subordinates away to the other patch.
The patches regenerated as in Fig. 20.1. The expected proportion of foragers at
patch 0 and at patch 1 is 1/3 and 2/3 respectively, the observed proportions were
0.327 and 0.673. From Ollason and Yearsley (2001).

particles (Cowie 1977). Oaten (1977) showed that the optimal foraging strat-
egy in a stochastic environment does not necessarily converge to the optimal
strategy in the corresponding deterministic environment. Iwasa et al. (1981)
proved that the optimal strategy in a stochastic environment is dependent
on the statistical distribution of the food, and that no single strategy was
300 J. G. Ollason et al.

always optimal. A possible mechanism underlying the observed foraging

behaviour, depends on four state variables: m, the body mass; mn , the
remembered number of particles eaten; mtw , the total remembered wait-
ing time to find the next particle; and mth , the remembered handling time
of a particle. Following the timing convention described above m(0), mtw (0),
mn (0) and mth (0) all refer to states at the beginning of the particular activity,
whether travelling between patches, waiting on arrival at a patch or feed-
ing on a particle. This amounts to rescaling t to zero each time the activ-
ity changes. All of these quantities decay at the same specific rate r. It is
assumed that r is the metabolic rate of the animal, measured as the rate of
decay of body mass required to maintain a unit mass of animal, and this
implies that m encodes both the memory of feeding and the current state of
the animal, in other words a unit quantity of memory is nothing more and
nothing less than a unit of body mass. The other quantities need not have
exactly the same values of r, provided r is small, but the simplest assump-
tion is that r is the same for each process. Each particle of food that is eaten
is eaten at a constant rate, hence a particle containing f units of food, with
a handling time of th is eaten at the rate of f /th units of food per unit time.
The effect of depletion is to increase the waiting time for the next particle.
Patches containing np particles are searched at a constant speed, at ran-
dom, by the forager, each patch separated from the next by a travelling time
of tt time units. The corresponding continuous case represents the feed-
ing of the animal by the following equations (Ollason 1980, Ollason & Ren
2004):

dF
= −vF
dt
dm
= vF − r m
dt

and at the steady state it can be shown that

F (0)(1 − e−v t c )
lim F (0)v e−v t c =
r →0 (v − r )

where F(0) is the mass of food at each patch when the animal arrives, and tc
is the steady-state staying time. As r → 0 the staying time, tc , converges to
that predicted by the marginal value theorem (Charnov, 1976). Feeding on
particles of food requires that three states have to be considered: feeding,
waiting to find the next particle and travelling between patches. The require-
ment of hunger is restated as follows: Leave the patch when the time spent
 Modelling foraging behaviour of individuals and groups 301

waiting to find the next particle is greater than the remembered time to find
a particle. In symbolic terms:
mtw (t)
t¯w (t) =
mn (t)
where t¯w (t) is the remembered waiting time. The rule, then, is leave when
tw ≥ t¯w (t). Between patches, m, mtw and mn all decay exponentially. On
arrival at the patch the animal is waiting so m and mn continue to decay
while mtw is increasing:
m(t) = m(0)e−r t
mn (t) = mn (0)e−r t
dmtw
= 1 − r mtw
dt
(The time spent waiting passes at the rate of 1 time unit per unit time.) If
the animal encounters a particle of food it feeds. This increases both m and
mn , but while it is feeding it is not waiting so mtw decays away:
dm f
= − rm
dt th
dmn 1
= − r mn
dt th
mtw (t) = mtw (0)e−r t
where f is the mass of food in a particle and th is the handling time. A
problem with this approach is that the calculation of mn depends on th which
cannot be known before the particle has been handled. The problem can be
resolved by assuming that th is small, giving the following approximation:
mn (t) ≈ 1 + mn (0)e−r t
From the continuous case of feeding in a patch that is searched at a constant
rate and at random, the amount of food removed as a function of time is
F (t) = F (0)(1 − e−v t )
Hence
 
1 F (0)
t= ln
v F (0) − F (t)
and a simple discretization of this process involving np particles gives
 
1 np
twi = ln
v np − i
302 J. G. Ollason et al.

1.02
1.00
0.98
0.96
Matching

0.94
0.92
0.90
0.88
0.86

220
200
8 180
9 160
10 140
11 120
12 100
13
80 t s1
t a1 60
14 40
15 20

Fig. 20.3 In 16 time units 16 particles of food are added to two patches. At
constant intervals ta1 defines the ratio of the times at which the particles are
added. ta1 = 8 signifies that particles are added to each patch at the rate of
16/8 = 2 time units, ta1 = 15 signifies that at one patch a particle is added every
16 time units, and at the other, every 16/15 time units. In this way, although the
rates of addition of the particles vary the overall rate of regeneration of the
system is constant. ts1 is the time taken to find a particle when there is one
particle in the patch, the smaller the value of ts1 the faster the patch may be
searched. Perfect matching is indicated by the dashed grid. Undermatching
increases as the difference in the rates of regeneration increases, and when ts1 is
small. The undulations of the response surface are aliasing artifacts. From
Ollason and Ren (2002).

where twi is that time at which the ith particle has been removed. In the
limits r → 0, np → ∞, the equilibrium value of twi converges to the equi-
lbrium value of tc in the continuous case, and to the prediction of the
marginal value theorem (Ollason & Ren 2002). A simulation of 16 animals
foraging at 2 patches of regenerating food is shown in Fig. 20.3. The par-
ticles are added one by one to each patch such that a total of 16 particles
were added to the patches in 16 time units. The particles were added to
the patches in the ratios 15:1, 14:2, 13:3, . . . , 8:8. It is assumed that the
time to find the next particle in a patch containing n particles, ts (n) = ts1 /n.
Figure 20.3 shows that matching approximates closely to the ideal free dis-
tribution and is weakly affected by the ratios of delivery, undermatching
being more pronounced as the ratio of regeneration at the patches becomes
more disparate.
 Modelling foraging behaviour of individuals and groups 303

MODELLING THE DIVING SEABIRD

The two models are drawn together modelling the foraging of the Common
guillemot, Uria aalge, which is represented as two interdigitated processes
of foraging at patches, involving travelling between patches of particles of
food and between continuous patches of oxygen. This work extends a model
developed by Liu (2002).
The model depends on three state variables: O, the volume of oxygen
contained in the body; g, the mass of food in the gut; and m, the body mass
(wet weight) of the animal. The environment is a patch of particles of food,
each of which the animal eats at the constant rate f. The density of particles
is not depleted. The aim of the model is to predict the time budget of the
guillemot in terms of the time spent diving and feeding, and the time spent
at the surface replenishing oxygen used in diving. The guillemot is in one
of five possible states:

(1) waiting (at night when it is too dark to feed);

(2) recovering from a previous dive, the recovery being terminated by one
of several possible states;
(3) descending to the depth at which it feeds;
(4) feeding on consecutively encountered particles until the bout is
terminated by one of several possible states;
(5) ascending to the surface.

Because costs of travelling involve the mass transported – not only of the
body tissue, but also the unincorporated contents of the gut – the wet weight
of gut contents is explicitly represented. The equations of the new model are
shown below.

DIVING AND ASCENDING

dg
= −r g g
dt
dm
= αr g g − r d m − (r d − r b )g
dt
dO
= −ρr d m − ρ(r d − r b )g
dt
The equations for ascending are the same as those for diving except that r a
is substituted for r d throughout. Because the full gut can make a significant
contribution to the total mass of the animal, the cost of carrying the unin-
corporated food may not be insignificant. It is assumed that the difference
304 J. G. Ollason et al.

Table 20.2. Parameters of digestion

Parameter Definition Units

rb The specific basal metabolic rate g s−1 g−1

rg The specific rate of clearance of the gut g s−1 g−1
rd The specific metabolic rate while diving g s−1 g−1
ra The specific metabolic rate while ascending g s−1 g−1
α The assimilation efficiency Dimensionless
ρ The stoichiometric equivalent of oxygen 1 g−1

between r d and r b is the net specific cost of moving a unit mass of material.
The residue is assumed to be the cost of maintaining the living tissue. The
specific cost, therefore, of moving the contents of the gut – which being
unassimilated, do not need to be maintained – while descending will be
r d − r b . The parameters are defined in Table 20.2.

FEEDING

The equations are:

dg
= f − rgg
dt
dm
= αr g g − r f m − (r f − r b )g
dt
dO
= −ρr f m − ρ(r f − r b )g
dt

The two differences between feeding and diving or ascending involve the
parameters f and r f . See Table 20.3 for parameters of feeding.

RECOVERING, WAITING AND PROVISIONING

The equations are:

dg
= −r g g
dt
dm
= αr g g − r r m − (r r − r b )g
dt
dO
= u(Omax − O/m)m − ρr r m − ρ(r r − r b )g
dt
 Modelling foraging behaviour of individuals and groups 305

Table 20.3. Parameters of feeding

Parameter Definition Units

rf The specific metabolic rate while feeding g s−1 g−1

f The rate of feeding g (wet weight) s−1

Table 20.4. Parameters of recovery

Parameter Definition Units

rr The specific metabolic rate while recovering g s−1 g−1

rp The specific metabolic rate while provisioning g s−1 g−1
Omax a The maximum concentration of oxygen in the tissue 1 g−1
u The time constant of re-oxygenation s

a
This is the hypothetical maximum concentration of oxygen in the tissue of an
animal without metabolic costs.

It is assumed that costs of recovering and of waiting are the same so the
relevant metabolic parameter, r r , is common to both processes. Table 20.4
shows the parameters of recovery.

ENDING FEEDING AND STARTING TO DIVE

On diving the animal contains O(0) units of volume of oxygen. The animal
will stop feeding and return to the surface when the concentration of oxygen
contained falls to Omin . The total metabolic cost of an activity is a function of
the mass m of the animal and of the activity in which the animal is engaged.
Associated with each activity is a linear scaling factor that expresses the cur-
rent metabolic cost as a proportion of some basal metabolism. When tis-
sue is catabolized there is a stoichiometrically equivalent usage of oxygen.
The four activities to be considered are: descending, feeding, ascending and
recovering. The scaling factors of the costs are s d , s f , s a and s r . The basal
specific metabolic cost is r b , and the stoichiometric equivalent of oxygen
is ρ.
The animal has a maximum and a minimum body mass, mmax and mmin
respectively. If m ≤ mmin the animal is dead from starvation. If m ≥ mmax
the animal is replete. The contents of the gut lie in the range 0 ≤ g ≤ g max ;
and a second bound is established, g min , the gut being effectively empty.
The guillemot has to stop feeding when the gut is filled.
306 J. G. Ollason et al.

RULES FOR DECISIONS

The bird has to stop feeding when the oxygen is running out, when O =
Omin m. It has to stop feeding when g = g max , or when m = mmax . The
resumption of feeding may be determined by the mass of food remaining in
the gut, for example, when g = g min ; it may also be determined by the mass
of oxygen in the tissue. For example, the animal might choose to dive again
when it contains the local maximum volume of oxygen obtained while it is
at the surface; or alternatively it might choose to dive again when the body
mass falls to mmax . The bird will dive when g = g min or when dO/dt = 0.
As noted above, the bird will stop feeding when m = mmax , but when it does
so, its gut will contain food, and as the food passes from the gut to the body
mass, the mass of the bird will increase and exceed mmax before decreasing
as the costs exceed the contribution of food from the emptying gut. The bird
may be regarded as being replete until the costs of metabolism reduce the
mass to mmax , and the conditions for resuming diving are m = mmax , and
dm/dt < 0.

FEEDING ON PARTICULATE FOOD

This is represented as described above except that the rate constants for the
different memories are allowed to differ, and the handling time is remem-
bered in addition. Hence
dmn 1
= − r n mn
dt th
dmtw
= 1 − r w mtw
dt
The third state variable, mth , is the memory of the handling time involved
in consuming the average particle of food, and is represented as follows:
dmth
= 1 − r h mth
dt
and as with the other memories, when the animal is not handling the
food the memory decays exponentially. Using these state variables, together
with m, the body mass of the animal, it is possible to derive the variables
t¯w (t) = mtw (t)/mn (t), where t¯w (t) is the expected waiting time to find the
next particle, t¯h (t) = mth (t)/mn (t), where t¯h (t) is the expected handling time
of the next particle, and f¯p (t) = m/mn , where f¯p (t) is the expected net
mass of food in the the next particle. These derived variables can be used to
construct additional rules to determine both the selection of the diet while
 Modelling foraging behaviour of individuals and groups 307

Table 20.5. Definitions of variables and parameters required for the

model of the guillemot foraging on particulate food

Definition Units

State variables
mtw The remembered waiting time s
mth The remembered handling time s
mn The remembered number of particles eaten Dimensionless
Derived variables
t¯w Expected waiting time, mtw /mn
t¯h Expected handling time, mth /mn
f¯p Expected net mass of food in a particle, m/mn
Parameters
tw Time to find next particle s
th Handling time of particle s
fp Mass of food in particle g (wet weight)
rw The specific rate of decay of the memory of s−1
waiting times
rh The specific rate of decay of the memory of s−1
handling times
rn The specific rate of decay of the memory of s−1
numbers

feeding and the total duration of the dives. It is assumed that the equations
representing the state variables g, m and O all apply, where relevant, as in
the continuous case.
Using the new-state and the derived-state variables, additional rules can
be developed to control the behaviour of the diving bird. A candidate rule
for feeding is:
Accept the newly encountered particle of food if

f p m̄th (t) r f m(t)

>
th [m̄tw (t) + m̄th (t)] α

where f p /th is the average feeding rate, ignoring the time spent waiting to
find the particle. The right-hand side represents the remembered rate of
feeding of the animal while it is feeding. The term m̄th (t)/[m̄tw (t) + m̄th (t)]
is the proportion of time spent handling the particle, so the left-hand side
of the equation represents an estimate of the feeding rate (including both
waiting time and handling time). The right-hand side represents the cur-
rent metabolic cost in terms of the wet mass of food ingested. Table 20.5
summarizes the new-state variables and parameters.
308 J. G. Ollason et al.

The following conditions lead to the termination of a dive. They include:

(1) when ts ≥ t¯w ;
(2) when O/m ≤ Omin ;
(3) when g ≥ g max ;
(4) when m ≥ mmax .

The chosen time is the earliest of all of these. The model has been param-
eterized to represent the foraging and diving behaviour of the guillemot
(Ollason et al. 2003). This document contains the solutions of all equations,
the simulation programs, the documentation of the programs and sam-
ple runs of the simulation of the individual guillemot feeding on sandeels
at depth. It is important to note that although the seabirds use oxygen,
they are never abundant enough to cause local depletion. The spatial dis-
tribution of the foragers will be determined by the properties of the prey
only. If significant depletion of the prey occurs as a result of the feed-
ing by the foragers, this will cause the distribution of foragers to approx-
imate to the ideal free distribution as predicted by both the continuous-
feeding and the particulate-feeding models in their simple forms. However,
even without the depletion of prey, interference increasing as a function of
local density of predators can also permit the ideal free distribution to arise
(Ollason 1987).

CONCLUSION

The models developed here subsume optimal foraging theory as a spe-

cial case of a more general account of the dynamics of foraging in hetero-
geneous environments containing continuously and discontinuously dis-
tributed resources. In summary, they are models of decision making to
reduce hunger, both hunger for food and hunger for oxygen; hunger being
defined by the forager’s current physiological needs. The models predict
the observed spatial distributions of foragers arising as individuals interact
searching for the requirements needed to maintain life. They provide foun-
dations for developing realistic models of foraging individuals that can be
allowed autonomously to forage together providing the atoms from which
the properties of populations can emerge.

ACKNOWLEDGEMENTS

We gratefully acknowledge the financial support of the Commission of

the European Community (to J. G. Ollason) under the MIFOS project
 Modelling foraging behaviour of individuals and groups 309

(CFP 96-079) and the IMPRESS project (Q5RS-2000-30864). We wish to

thank J. C. Ollason for reading and commenting on an earlier version of
this paper.

REFERENCES
Charnov, E. L. (1976). Optimal foraging: the marginal value theorem. Theor. Popul.
Biol., 9, 129–36.
Cowie, R. J. (1977). Optimal foraging in great tits (Parus major). Nature, 268, 137–9.
Iwasa, Y., Masahiko, H. & Yamamura, N. (1981). Prey distribution as a factor
determining the choice of optimal foraging strategy. Am. Nat., 117, 710–23.
Kennedy, M. & Gray, R. D. (1993). Can ecological theory predict the distribution of
foraging animals? Oikos, 68, 158–66.
Liu, K. (2002). Modelling the physiology, behaviour, and ecology of dive foraging
seabirds. Unpublished Ph.D. thesis, University of Aberdeen, Scotland, UK.
Oaten, A. (1977). Optimal foraging in patches: a case for stochasticity. Theor. Popul.
Biol., 12, 263–85.
Ollason, J. G. (1980). Learning to forage: optimally? Theor. Popul. Biol., 18, 44–56.
(1987). Learning to forage in a regenerating patchy environment: can it fail to be
optimal? Theor. Popul. Biol., 31, 13–32.
Ollason, J. G. & Ren, N. (2002). Taking the rough with the smooth: foraging for
particulate food in continuous time. Theor. Popul. Biol., 62, 313–27.
(2004). A general dynamical theory of foraging in animals. Discrete Contin. Dyn.
System Ser. B, 4, 713–20.
Ollason, J. G. & Yearsley, J. M. (2001). The approximately ideal, more or less free
distribution. Theor. Popul. Biol., 59, 87–105.
Ollason, J. G., Ren, N., Scott, B. E. & Daunt, F. (2003). A Model of the Foraging
Behaviour of a Diving Predator Feeding Underwater at a Patch of Food (Revision
2004.3.0). IMPRESS Technical Report 2003-14 (Available from The
Netherlands Institute for Sea Research, Texel, the Netherlands.)
 21
The Scenario Barents Sea study: a case of
minimal realistic modelling to compare
management strategies for marine ecosystems
T. SCHWEDER

Scenario modelling to evaluate management strategies was originally devel-

oped for whaling, but is now increasingly applied in fisheries. The basic idea
is to establish a minimal but realistic model for computer simulation of the
system, with removals governed by a management strategy, predation and
additional natural mortality. The system is projected forward under com-
peting strategies for a number of years, with replications to capture the sta-
tistical uncertainties surrounding the system. Strategies are compared in
terms of their simulated long-term performance.
The purpose of the present study is to evaluate the effect on the cod,
capelin and herring fisheries of managing minke whaling and harp sealing
in the Barents Sea in this way. The study is funded by the Norwegian Min-
istry of Fisheries. In a recent White Paper (Stortingsmelding nr 27 2003–4)
on the management of marine mammals in Norwegian waters, the Min-
istry plans ‘to establish a scientific basis for changing to ecosystem-based
management where marine mammal stocks are managed in conjunction
with the other living marine resources’.
To manage a marine ecosystem is to play a game with Nature, and per-
haps also with other players such as fishermen and industries which are
causing the system to become increasingly polluted. The game we consider
is played by the Agency (government body) and Nature.
The concept of strategy is central to game theory. A management strat-
egy is a feedback rule that specifies the action to be taken, given the history

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 The Scenario Barents Sea study 311

of the process as observed by the player. For our purpose, the actions open
to the Agency are to set TACs for removals of harp seals and minke whales.
The Agency also sets TACs for the fisheries. TACs are assumed to be exactly
filled if abundance allows. Fishermen also behave according to economic
realities, and may discard or take more than the TAC. These economic real-
ities are disregarded here.
The strategy of Nature is determined by the internal dynamics of the
ecosystem, and its dynamic reaction to the removals caused by the fisheries.
The better the Agency knows Nature’s strategy, the better it can assess the
quality of a given management strategy for whaling and sealing. Our study
is an attempt to assemble the available knowledge and data pertinent to the
relevant dynamics of the upper trophic level of the ecosystem.
Despite more than 100 years of marine research in the area and despite
the fact that the system is less complex than many other ecosystems and
is comparably well known, our knowledge of the Barents Sea ecosystem is
rather limited. The quality of understanding varies considerably depending
on the topic. For features of the ecosystem about which little is known, the
model must be simple in order not to spread the available information too
thinly. Other better known features might be less relevant for the interaction
between the modelled species, and are therefore modelled in less detail than
is possible. Our aim is not a detailed description of the ecosystem represent-
ing all available knowledge, but rather a practical and reasonably realistic
model tailored to the purpose of the study. Borrowing a term from Punt
and Butterworth (1995), a model that balances realism and uncertainty,
and that is operational and practical to use, is called a ‘minimal realistic
model’.
We consider the Barents Sea, including the spawning grounds for north-
east Arctic cod in Lofoten, and also a residual area where Norwegian spring-
spawning herring and minke whales are found when they are not present in
the Barents Sea. Only young herring migrate to the Barents Sea. The minke
whales feed in the study area and breed elsewhere during winter. With
our limited and pragmatic purpose, we will include only harp seals, minke
whales, cod, capelin and herring in the model. Hamre (1994) describes the
main features of the Barents Sea–Norwegian Sea ecosystem.
A strategy should be evaluated in terms of its anticipated performance
in the long run. The system needs more than 100 years to become station-
ary, due to the slow dynamics of minke whales. We will therefore simulate
catches and stock status over at least a 100-year period, and study perfor-
mance by statistics summarizing the simulation results. It is beyond the
312 T. Schweder

scope of this chapter to give weights to the various objectives for manage-
ment, and to interpret results in financial terms.
The model has yearly stochastic variation, mainly in fish recruitment
and in the stock abundance estimates on the basis of which TACs are set.
This last uncertainty applies also to abundance estimates for minke whales
and harp seals, and translates into stochastic variation in TACs for these
species as well.
The Agency is faced with uncertainties with respect to the population
dynamics of the key stocks, and also to the interactions caused by preda-
tion between stocks. These uncertainties are handled by drawing parame-
ters from statistical distributions for each run of the simulation model, but
uncertainties with respect to functional forms are not addressed.

HISTORICAL BACKGROUND

The present study (2002–4) is a sequel to a previous Scenario Barents Sea

study. The aim of that earlier study was to compare management strate-
gies for cod, capelin and herring (Hagen et al. 1998). The model was previ-
ously extended to study the effects on the fisheries of retuning the Revised
Management Procedure (RMP) of the International Whaling Commission
(IWC) (IWC 1994) for minke whales (Schweder et al. 1998, 2000). In addi-
tion to including harp seals in the current model, the structural forms of
the predation models are changed, and they are re-estimated.
Our study, and that of Punt and Butterworth (1995) – which was devel-
oped to reflect recommendations reported in Butterworth and Harwood
(1991) – has been inspired by the simulation approach taken when the
RMP of the IWC was developed for managing single baleen whale stocks
(Kirkwood et al. 1997). The resulting RMP is robust against the type and
extent of uncertainties that usually surrounds stocks of baleen whales, as
demonstrated in a scenario study focusing on uncertainties represented
by variability and bias in survey data, and in variations in the population
dynamics, e.g. in the shape and strength of density-dependent recruitment,
changes in carrying capacity, episodic events, etc.
In the early 1970s the Club of Rome based its gloomy predictions on
scenarios simulated on computers to illustrate environmental effects of
population growth and industrial activity. This work was inspired by Buck-
minster Fuller, who educated many Americans in strategic thinking and in
the use of simulation through computers and otherwise to improve envi-
ronmental and other policies. He conducted military war games to save
 The Scenario Barents Sea study 313

80° N

Svalbard

6
75°
Bear Island

4
3 5
2 Barents Sea
70° Norwegian Sea
Lofoten
Vesteralen Norway Russia

Vestorden

1 White
Sea
65°

60° Karmay

10° 20° 30° 40° 50° 60° E

Fig. 21.1 Spatial division of the Barents Sea, adapted for MULTISPEC (Bogstad
et al. 1997).

Spaceship Earth from destruction (http://www.bfi.org) throughout Amer-

ican university campuses, and was an inspiration to the environmental
movement.

THE PRESENT SCENARIO BARENTS SEA MODEL

Overview

Cod, capelin, herring, harp seals and minke whales are distributed over
the seven areas used in previous studies (Bogstad et al. 1997) plus a resid-
ual area (Fig. 21.1), and over age and month. Fish are also distributed
314 T. Schweder

over lengths. The chosen subdivision of the study area and the time
step is regarded as the coarsest possible temporal and spatial stratifi-
cation respecting gradients in seasonal overlaps between predators and
prey in the Barents Sea system (S. Tjelmeland, personal communication,
2004).
Moving from one month to the next, surviving fish of the same length
are allocated to new length groups according to an individual growth sched-
ule which depends on season and species. Individual growth in cod depends
on the supply of capelin, but is independent of prey availability for capelin
and herring. A fixed length–weight relationship for each fish species is
used for calculating the biomass of the species at any time. This is needed
because fish TACs and predation are calculated in biomass terms.
The population-dynamics model for minke whales is taken from a report
of the IWC (1993), and is identical to that in Schweder et al. (1998). The
model for harp seal is taken from Skaug and Øien (2003). Recruitment
in cod, capelin and herring is modelled by Beverton–Holt functions aug-
mented with stochastic variability.
Minke whales and harp seals are top predators. Their population dynam-
ics is modelled as being unaffected by fish stock abundance. While this
is not quite realistic, it is nevertheless considered adequate for present
purposes because the modelled population dynamics broadly reflects prey
abundances similar to those expected over the simulation period. The
marine mammals prey on all three species of fish, and also on other organ-
isms, as do cod which are cannibals; herring prey on capelin larvae; capelin
are at the base of the model.
The interactions between populations are modelled solely as mortality
due to predation. These interactions are additive in the sense that they act
only to cause removals. Hjermann et al. (2004) found the mortality factors
to be additive and not compensatory for Barents Sea capelin. Fishing is also
assumed to cause removals but not to have indirect effects. In addition to
mortality due to modelled removals, there is a component of fixed residual
natural mortality.
Predation is modelled in two steps. The daily biomass consumed is cal-
culated for each group of predators from energetic considerations and from
other sources,. This consumption is then split between the groups of prey
according to prey choice probabilities. The prey choice probabilities are esti-
mated from stomach-content data and estimated prey availability by time
and area.
Key equations of the model are discussed below, together with some
comments on the estimation of their parameters. Additional information
 The Scenario Barents Sea study 315

is the Norwegian Computing Center website (http://www.nr.no/files/

samba/emr/scenario document.pdf).

Recruitment and yearly dynamics

Juvenile capelin are recruited as 1 year olds on 1 October. Juvenile cod and
herring are recruited as 1 year olds on 1 January. Given a spawning stock
with biomass B, the number of recruits, R, one year later is calculated from
the Beverton–Holt function

R = L µB/(B + η)

with multiplicative log-normal stochastic variation L (with median 1 and

coefficient of variation σ ), maximal recruitment in median years µ, and
half-value parameter η. There are stochastically chosen abnormal years,
where the recruitment parameters are switched to more productive values.
Newly recruited fish are distributed by length group.
Recruitment in minke whales follows the age- and sex-structured model
used by the IWC. The median age at maturity is 7 years, and the fecundity
drops as the number of mature females N approaches a carrying capacity K
according to the Pella Tomlinson formula

R = Nβ{1 + α[1 − (N/K )z]}

where β is the mean number of live births per year for mature females at
carrying capacity, α is the resilience parameter and z is the degree of com-
pensation. The last two parameters determine the stock level, here 60%
of carrying capacity, corresponding to maximum sustainable yield. Recruit-
ment to the exploitable stock happens one year after birth. The recruitment
parameters are estimated using catch data, abundance data and a series of
relative abundance indices based on catch and effort data (Schweder et al.
1998).
Recruitment in harp seals is also modelled by the Pella Tomlinson for-
mula above, but with z = 1 degree of compensation, making the maximum
sustainable yield stock level around 50% (Skaug & Øien 2003).

MIGRATION AND MONTHLY DYNAMICS

Monthly age-dependent transition matrices between areas (migration

matrices) based on survey data were developed by Bogstad et al. (1997). The
stationary distributions related to these matrices are used to distribute the
various populations over areas, by month and age.
316 T. Schweder

Mortality in fish has three components: removals due to fishing,

removals due to modelled predation and residual natural mortality.
Removals due to fishing are calculated from the yearly TAC, and distributed
according to a specified scheme over areas and months.
TACs for cod and herring are calculated from yearly abundance esti-
mates obtained from a simple virtual population analysis (VPA) assess-
ment by applying fixed fishing mortalities, and based on catch data and
yearly survey data simulated with noise. Capelin is managed according to
the approach described in Gjøsæter et al. (2002) which, approximately, is to
set aside the estimated amount of capelin needed to feed the cod, and also
500 000 t of capelin to spawn.
The removals due to predation in a given month and area are calculated
from the number of predators present and the biomass of all their prey
species, as described below. From the number of individuals in a given cat-
egory at the beginning of a month, the number consumed by predators and
the number harvested are subtracted, and the remainder are reduced fur-
ther as a result of the residual mortality rate.

Predation

Predation models are estimated anew for cod, minke whales and harp seals.
They consist of two components: total amount eaten per day, and diet choice
probabilities depending on prey abundance. The quantity consumed is esti-
mated from energetic considerations for harp seals and minke whales, and
also from experimental and observational studies for cod. The choice prob-
ability models have a common multiple logistic format, and are estimated
by comparing the type of prey found in sampled stomachs to estimated prey
abundance at the time and in the area of sampling.
Food takes time to digest, and sampled stomachs often have mixed
contents. Each stomach is therefore regarded as representing three feed-
ing events, and a rule codes them by food item. Minke whales, for exam-
ple, can choose between capelin, herring, cod and other food – indexed
from 1 to 4. If only one item is found in a stomach, all three feed-
ing events are coded by this item. For a covariate vector x represent-
ing area-specific abundance of cod, capelin and herring, and for param-
eter vectors β k related to food item k, the model consists of the linear
predictors


θk = βki xi
i
 The Scenario Barents Sea study 317

leading to the logistic choice probabilities



4
pk = exp(θk ) exp(θj )
j=1

Zhu (2003) estimated the following diet choice model for minke whales

θk = −1.82Iherring − 2.52Icapelin − 3.63Icod

+ 0.000 231Bk + 0.265 log(Bk ),

where Bk is the abundance (in thousands of tonnes) of fish species k

in the area, and Iherring indicates whether herring is the target species
k or not, and similarly for cod and capelin. Other food is the reference
category, θother = 0 (i.e. pother = 1/(1 + θcod + θcapelin + θherring )). All coeffi-
cients, except that for linear abundance, are highly significant.
For harp seals, stomachs are sampled mainly on the ice north in the Bar-
ents Sea in summer (Nilssen et al. 2000). The logistic diet choice model we
use is estimated from these data stomach (http://www.nr.no/files/samba/
emr/scenario document.pdf). Folkow et al. (L. P. Folkow, E. S. Nordøy and
A. S. Blix, unpublished observations, 2004) found, however, from satellite
tagging that Greenland Sea harp seals regularly migrate to the Barents Sea
in summer and autumn, and spend about half their time there in open
water, presumably feeding on capelin. The stomach contents found in sam-
pled harp seals are therefore not representative of the diet in the important
summer feeding period in the Barents Sea. The same presumably applies
to harp seals from the White Sea, which constitutes the dominant harp seal
stock in the Barents Sea.

Management

Fisheries are modelled to be managed by fixed and simple rules. Minke

whaling is managed by the RMP of the IWC, tuned at three different lev-
els; while harp sealing is managed by three different rules taken from the
International Council for the Exploration of the Seas (ICES 2001). There are
thus nine different combinations of management procedures for whaling
and sealing to compare.
318 T. Schweder

DISCUSSION

Current management

Fish
TACs for cod are set to target a fishing mortality of F = 0.4 when spawning-
stock biomass is estimated above precautionary stock biomass, Bpa . The
stock is assessed by the VPA-method XSA (see ICES 2004). The capelin
and herring fisheries are essentially managed as modelled.

Minke whaling
Currently, Norway has a fleet of some 33 fishing vessels which harvest
minke whales in the summer months in Norwegian waters. Whaling
accounts for about 25% of the income for this fleet, and is economically sus-
tainable. Minke whaling resumed in 1993 after it was stopped in 1987. The
catch has increased from some 250 whales in 1993–5 to some 600 whales
in recent years. This is slightly below the TAC calculated from the man-
agement strategy. Vessel quotas are given to licensed whalers. The catch is
closely controlled with skilled observers on board.
The TAC for minke whales is calculated by the RMP of the IWC.
The input data to the RMP is the catch series and the series of absolute
abundance estimates obtained from double-platform, line-transect surveys
(Skaug et al. 2004).
The RMP can be tuned to target various population levels under cautious
standard assumptions. In the early years, a target level of 72% of carrying
capacity was used to set Norwegian minke whale TACs. This target level is
that which would be reached after 100 years of application of the RMP to
a stock, originally at its pristine level, with the lowest productivity rate con-
sidered plausible. In 2003 the target level was reduced to 66% of carrying
capacity, and for 2004 to 62%, resulting in higher TACs.

Harp sealing
Norwegian sealing is currently a small, heavily subsidized industry carried
out with old specialized sealing vessels. In the 1990s the yearly average
number of harp seals taken in the Barents Sea by Norway was 6200, and
has declined in recent years.
Norway manages harp sealing in the Barents Sea together with Russia,
and according to advice from ICES which calculates TACs to keep the stock
at its current level. The White Sea–Barents Sea stock is now estimated to
comprise nearly 2 million adults, and the TAC was 53 000 adult equivalents
 The Scenario Barents Sea study 319

(one adult equals 2.5 pups) in 2001–3, well above the catch. Pups become
adult when they are 1 year old.
The economics of sealing has improved somewhat recently, but sealing
will probably need substantial subsidies to raise the catch to what is cur-
rently regarded as sound from an economic perspective for the fishery as a
whole (Stortingsmelding nr 27 2003–4).

Management issues

The Norwegian Government wants to develop a management strategy for

sealing and whaling as part of a management regime for the ecosystem
containing fish and marine mammals. For this to have scientific support,
substantial research is needed, as acknowledged in the White Paper (Stort-
ingsmelding nr 27 2003–4). Since harp seals, cod and capelin are managed
jointly by Norway and Russia – while herring is managed by five parties
(Norway, Russia, the Faroes, Iceland and the European Union) – manage-
ment of this ecosystem is far from being a national issue for Norway alone.
The model described above is an improvement of the model used by
Schweder et al. (1998), who studied the effect on the fisheries of retuning
the RMP of the IWC for minke whaling. The revision is not yet complete,
and no results are available. To illustrate the type of results anticipated, our
earlier tentative conclusion was that on retuning the RMP from targeting
72% of carrying capacity down to 60%, annual catches increase by some
300 minke whales, and 0.1 million tonnes of cod on average. The effect on
the herring fishery was unclear. The effect on the long-term minke whale
abundance was a reduction from about 96 000 whales to 87 000 whales.
Whether the previously estimated effects on fisheries from increased
minke whaling will be maintained in our current study remains to be seen.
It also remains to be seen whether increased sealing will influence the fish-
eries positively to the degree currently expected by the Government and
fishermen.

Modelling considerations

The various parts of the model were first estimated on relevant data, but
in isolation from the rest of the model and from other data. The model
as a whole is however more than the sum of its parts, and this piecewise
estimation has turned out to be unsatisfactory. To balance the model it has
been necessary to fit the model simultaneously to historical data. This is
work in progress.
320 T. Schweder

The approach is the following. Assume abundance estimates with asso-

ciated uncertainty measures to be available for years 1, . . . , T. From esti-
mated stock status in year t, the model predicts the stock status in year
t + 1 when all parameters have given values. Comparing the abundance
estimates and the one-year predicted stock status in year t + 1, accounting
for observed catches, leads to a likelihood component. In addition to these
likelihood components, each piecewise fitting results in a likelihood com-
ponent. The total likelihood is then obtained by combining all the likelihood
components, and parameters are estimated by maximizing this combined
likelihood.
The log-likelihood function is expected to be approximately quadratic
in the parameter vector, perhaps after a parametric transformation. The
(transformed) parameter vector then has a multivariate normal confidence
distribution representing the uncertainty in estimated model parameters
(Schweder & Hjort 2002).

CONCLUSION

Our model rests on a number of assumptions of varying degrees of real-

ism. Feedback from fish to marine mammals is disregarded. Little is actu-
ally known of this feedback, which presumably is of importance at least for
harp seals. The modelled populations interact only by removals due to pre-
dation. Other interactions are indeed conceivable. The residual mortality
rate is assumed constant, but is presumably variable. The category ‘other
food’ has been modelled as always available in sufficient supply. There are
also a number of assumptions regarding structure and data that are open to
criticism. The predation model for harp seals is, for example, based on selec-
tive and weak data. Making the model more realistic will on most counts
entail more complexity.
The uncertainties surrounding the model are indeed substantial. They
are handled by drawing parameter values from statistical distributions in
each simulation run. These distributions will also rest on assumptions, and
will broadly represent sampling variability in the data used to estimate coef-
ficients. There are additional uncertainties not taken into account, particu-
larly regarding the structure of the model and functional forms.
To establish a minimal realistic model for the upper trophic levels of the
ecosystem in the Barents Sea, also as regards the handling of uncertainty,
is quite a challenge to address adequately. The alternative to scenario mod-
elling is, however, not to model, and thus to leave the choice of management
 The Scenario Barents Sea study 321

strategy to qualitative judgment. This will probably lead to a bad strategy,

perhaps based on political compromises without much consideration for
ecosystem dynamics. From this point of view, our model might in fact be
criticized for being too complex relative to the available knowledge and data,
and relative to the purpose of the exercise.
Our model, programmed in 25 000 lines of C-code, might actually be too
complex. Having struggled to estimate recruitment and mortality parame-
ters by fitting the model to historical data, we have not yet managed to bal-
ance the model to our satisfaction. When simulating the model, stock trajec-
tories in some of the runs are implausible, and they vary too much between
runs compared with historical data. Our small project is thus unsuccess-
ful in that no comparative simulation experiments could be performed that
would be useful for management.
As mentioned, the Norwegian Government wants to move towards
ecosystem management of marine mammals based on science. This com-
mitment to science is encouraging. Our project might have been useful as
a learning exercise, but more research is needed to establish a sufficiently
credible model – as is needed for ecosystem management. In addition to the
research needs created by the Norwegian commitment to ecosystem man-
agement, we think the current research, aiming at a scenario model for the
Barents Sea, should be carried forward also for general reasons. The Barents
Sea system is in fact less complex than other marine systems of comparable
size, in terms of biomass in the upper trophic level, and if marine ecosys-
tem management is possible at all, it should indeed be so in the Barents
Sea.

ACKNOWLEDGEMENTS

Thanks to Gro Hagen of the Norwegian Computing Center for keeping the
computer program under control, and to Bjarte Bogstad, Ulf Lindström and
Sigurd Tjelmeland of the Marine Institute, for providing data and biological
insight.

REFERENCES
Bogstad, B., Hiis Hauge, K. & Ulltang, Ø. (1997). MULTISPEC: a multispecies
model for fish and marine mammals in the Barents Sea. J. Northw. Atl. Fish.
Sci., 22, 317–41.
Butterworth, D. S. & Harwood, J. (1991). Report on the Benguela Ecology Programme
Workshop on Seal–Fishery Biological Interactions. Report of the Benguela Ecology
Programme, South Africa, 22.
322 T. Schweder

Gjøsæter, H., Bogstad, B. & Tjelmeland, S. (2002). Assessment methodology for

Barents Sea capelin, Mallotus villosus (Müller), ICES J. Mar. Sci., 59, 1086–95.
Hagen, G. S., Hatlebakk, E. & Schweder, T. (1998). Scenario Barents Sea. A tool
for evaluating fisheries management regimes. In Models for Multi-species
Management, ed. T. Rødseth. Berlin: Physica-Verlag, pp. 173–226.
Hamre, J. (1994). Biodiversity and exploitation of the main fish stocks in the
Norwegian Sea–Barents Sea ecosystem. Biodiversity Conserv., 3, 473–92.
Hjermann, D. Ø., Ottersen, G. & Stenseth, N. C. (2004). Competition among
fishermen and fish causes the collapse of Barents Sea capelin. Proc. Natl. Acad.
Sci. U. S. A, 101, 11 679–84.
ICES (International Council for the Exploration of the Sea) (2001). Report of the
Joint ICES/NAFO Working Group on Harp and Hooded Seals. ICES CM
2001/ACFM:08. Copenhagen, Denmark: ICES.
(2004). Report of the Arctic Fisheries Working Group. ICES CM 2004/ACFM:28.
Copenhagen, Denmark: ICES.
IWC (International Whaling Commission) (1993). Specification of the North
Atlantic minke whaling trials. Rep. Int. Whal. Commn, 43, 189–96.
(1994). The Revised Management Procedure (RMP) for Baleen Whales. Rep. Int.
Whal. Commn, 44, 145–52.
Kirkwood, G. P., Hammond, P. & Donovan, G. P. (eds.) (1997). Development of the
Revised Management Procedure. Special Issue 18. Cambridge, UK: International
Whaling Commission.
Nilssen, K. T., Pedersen, O.-P., Folkow, L. P. & Haug, T. (2000). Food
consumption estimates of Barents Sea harp seals. In Minke Whales, Harp and
Hooded Seals: Major Predators in the North Atlantic Ecosystem, eds. G. A.
Vikingsson & F. O. Kapel. NAMMCO Scientific Publication 2. Tromsø,
Norway: NAMMCO, pp. 9–27.
Punt, A. & Butterworth, D. S. (1995). The effects of future consumption by the
Cape fur seal on catches and catch rates of the Cape hakes. 4. Modelling the
biological interaction between Cape fur seals Artocepalus pusillus pusillus and
the Cape hake Merluccius capensis and M. paradoxus. S. Afr. J. Mar. Sci., 16,
255–85.
Schweder, T. & Hjort, N. L. (2002). Confidence and likelihood. Scand. J. Statistics,
29, 309–32.
Schweder, T., Hagen, G. S. & Hatlebakk, E. (1998). On the effect on cod and
herring fisheries of retuning the Revised Management Procedure for Minke
whaling in the Greater Barents Sea. Fish. Res., 37, 77–95.
Schweder, T., Hagen, G. S. & Hatlebakk, E. (2000). Direct and indirect effects of
minke whale abundance on cod and herring fisheries: a scenario experiment
for the Greater Barents Sea. In Minke Whales, Harp and Hooded Seals:
Major Predators in the North Atlantic Ecosystem, eds. G. A. Vikingsson & F. O.
Kapel. NAMMCO Scientific Publication 2, Tromsø, Norway: NAMMCO,
pp. 121–32.
Skaug, H. J. & Øien, N. (2003). Historical assessment of harp seals in the White
Sea/Barents Sea. Working Paper 2. Workshop held by Joint ICES/NAFO
Working Group on Harp and Hooded Seals, Woods Hole, MA, USA, February
2003.
 The Scenario Barents Sea study 323

Skaug, H. J., Øien, N., Bøthun G. & Schweder, T. (2004). Abundance of minke
whales (Balaenoptera acutorostata) in the Northeast Atlantic: variability in time
and space. Can. J. Fish. Aquat. Sci., 61, 870–86.
Stortingsmelding nr 27 (2003–4). Norsk sjøpattedyrpolitikk. Oslo 19 March 2004
(summary in English at http://odin.dep.no/filarkiv/202967/marine mammal
summary final.pdf).
Zhu, M. (2003). Minke whale: a rational consumer in the sea. Unpublished M.Sc.
thesis in economics, University of Oslo, Norway.
 22
Setting management goals using information
from predators
A. J. CONSTABLE

This chapter examines how goals and reference points might be set for
higher trophic levels – such as marine mammals, birds and fish. It briefly
explores the general characteristics of objectives for higher trophic levels
within the context of ecosystem-based management, noting that the empha-
sis for managing the effects of human activities on higher trophic levels is
biased towards fisheries-based approaches rather than approaches that take
into account the maintenance of ecosystem structure and function. Follow-
ing this, the precautionary approach developed in the Commission for the
Conservation of Antarctic Marine Living Resources (CCAMLR) for taking
account of higher trophic levels in setting catch limits for target prey species
is described. The last section considers indicators of the status of preda-
tors with respect to establishing target and limit/threshold reference points
that can be used directly for making decisions. These indicators include
univariate indices summarizing many multivariate parameters from preda-
tors, known as composite standardized indices, as well as an index of preda-
tor productivity directly related to lower trophic species affected by human
activities.
Ecosystem-based management encapsulates notions of conservation
and wise use of ecosystems (Mangel et al. 1996). Managers are now expected
(a) to maintain ecosystem properties and, in some cases, (b) to restore
ecosystems when they are judged to be impacted (caused to be altered),
directly or indirectly, by human activities. With appropriate scientific sup-
port, they need to define how ecosystems might be judged to be impacted
and to determine mechanisms for reducing or eliminating such impacts.

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
 Setting management goals using information from predators 325

Objectives for higher trophic levels, and ecosystems as a whole, need to

be set in quantitative terms, often called operational objectives. Also, criti-
cal values for indicators need to be specified for triggering action to achieve
those operational objectives; they would be couched in rules to facilitate
decision making by managers. This chapter examines how goals and refer-
ence points might be set for higher trophic levels – such as marine mam-
mals, birds and fish. It briefly explores the general characteristics of objec-
tives for higher trophic levels within the context of ecosystem-based man-
agement. Following this, indicators of the status of predators will be consid-
ered with respect to establishing target and limit/threshold reference points
that can be used directly for making decisions.

GENERIC GOALS FOR HIGHER TROPHIC LEVELS

Principles for the conservation of wild living resources (terrestrial and

marine) arose three decades ago (Holt & Talbot 1978, updated by Man-
gel et al. 1996; Box 22.1). These have been incorporated in various ways
into management (Box 22.2) in terms of four general objectives, which
are not mutually exclusive: (a) maintain abundance at productive levels, (b)
minimize the threats of extinction, (c) maintain ecosystem processes, and
(d) maintain biodiversity. To a large extent, managers have focused on the
first two general objectives, probably because the theory surrounding pro-
ductivity and maintenance of individual populations has been well articu-
lated in the fisheries literature (e.g. Beverton & Holt 1957, Quinn & Deriso
1999).
The latter two objectives on ecosystem dynamics and biodiversity are
less well formulated and suffer probably because of the considerable debate
over the relative importance and/or existence (or not) of ecological proper-
ties of ecosystems such as species diversity, resilience, resistance to pertur-
bation and ecosystem recovery (e.g. Pimm & Gilpin 1984, Chesson & Case
1986, Wilson & Peter 1988, Petraitis et al. 1989, Underwood 1989, Yodzis
1994, 1996, Tilman 1999). However, the potential direct or indirect effects
of human activities on higher predators is not in dispute. Nor is the potential
for higher predators to sometimes have a disproportionately large influence
on the structure of lower trophic levels through trophic cascades and key-
stone predation (Fairweather 1990). Thus, it should be possible to articulate
these wider ecosystem objectives in relation to higher trophic levels despite
the more general ecological debates.
At present, too much attention is given to elaborating the fisheries-
based, single-population approach to the management of higher trophic
326 A. J. Constable

Box 22.1 Principle for conserving wild living resources

The second principle for conservation (including utilization of
species) of wild living resources updated in 1994 (Mangel et al. 1996)
from principles published in Holt and Talbot (1978) is:

II. The goal of conservation should be to secure present and future

options by maintaining biological diversity at genetic, species,
population, and ecosystem levels; as a general rule neither the resource
nor other components of the ecosystem should be perturbed beyond
natural boundaries of variation.

Seven mechanisms to help implement this principle were described

by Mangel et al. (1996):
(1) manage total impact on ecosystems and work to preserve
essential features of the ecosystem;
(2) identify areas, species and processes that are particularly
important to the maintenance of an ecosystem, and make
special efforts to protect them;
(3) manage in ways that do not further fragment natural areas;
(4) maintain or mimic patterns of natural processes, including
disturbances, at scales appropriate to the natural system;
(5) avoid disruption of food webs, especially removal of top or basal
species;
(6) avoid significant genetic alteration of populations;
(7) recognize that biological processes are often non-linear, are
subject to critical thresholds and synergisms, and that these
must be identified, understood and incorporated into
management programmes.

levels and too little consideration has been given to considering the wider
ecosystem consequences of changes to the composition of higher trophic
levels.
Higher trophic levels may be affected by human activities in many
ways, including direct exploitation, incidental mortality in fisheries, pollu-
tion, coastal works and mining, shipping and tourism. Direct interactions
can result in death (removal of individuals from populations), altered well-
being (change in individual growth and/or reproduction, change in health
or physiological condition, or displacement to other areas), or accumulated
 Setting management goals using information from predators 327

Box 22.2 Key international objectives for the conservation of

higher marine predators
Convention on the Conservation of Antarctic Marine Living
Resources (1980), Article II, Paragraph 3
3. Any harvesting and associated activities in the area to which this
Convention applies shall be conducted in accordance with the
provisions of this Convention and with the following principles
of conservation:
(a) prevention of decrease in the size of any harvested
population to levels below those which ensure its stable
recruitment. For this purpose its size shall not be allowed to
fall below a level close to that which ensures the greatest net
annual increment;
(b) maintenance of the ecological relationships between
harvested, dependent and related populations of Antarctic
marine resources and the restoration of depleted populations
to the levels defined in sub-paragraph (a) above; and
(c) prevention of changes or minimization of the risk of change
in the marine ecosystem which are not potentially reversible
over two or three decades, taking into account the state of
available knowledge of the direct and indirect impact of
harvesting, the effect of the introduction of alien species, the
effects of associated activities on the marine ecosystem and
of the effects of environmental changes, with the aim of
making possible the sustained conservation of Antarctic
marine living resources.

United Nations Convention on the Law of the Sea (1982),

Article 119, Paragraph 1
Conservation of the living resources of the high seas
1. In determining the allowable catch and establishing other
conservation measures for the living resources in the high seas,
States shall:
(a) take measures which are designed, on the best scientific
evidence available to the States concerned, to maintain or
restore populations of harvested species at levels which can
produce the maximum sustainable yield, as qualified by
328 A. J. Constable

relevant environmental and economic factors, including the

special requirements of developing States, and taking into
account fishing patterns, the interdependence of stocks and
any generally recommended international minimum
standards, whether subregional, regional or global;
(b) take into consideration the effects on species associated with
or dependent upon harvested species with a view to
maintaining or restoring populations of such associated or
dependent species above levels at which their reproduction
may become seriously threatened.
Convention on Biological Diversity (1992)
Article 1. Objectives
‘‘The objectives of this Convention . . . are the conservation of bio-
logical diversity, the sustainable use of its components and the fair
and equitable sharing of the benefits arising out of the utilization of
genetic resources . . .
Article 2. Use of Terms
For the purposes of this Convention:

‘‘Biological diversity” means the variability among living organisms from

all sources including, inter alia, terrestrial, marine and other aquatic
ecosystems and the ecological complexes of which they are part; this
includes diversity within species, between species and of ecosystems . . .

individual and/or population effects that could ultimately affect genetic

diversity and lower population viability (Fig. 22.1).
Indirect effects on higher trophic levels arise from the loss of biomass
from the ecosystem or habitat caused by the factors above. These losses
result in changes to the trophic structure in the food web or displacement
or movement of higher predators from an area because of the loss of criti-
cal habitat (Fig. 22.1). This is most easily envisaged in terms of the removal
of prey by fisheries but could also arise if lower trophic levels are directly
affected by any of the other non-fishing factors listed above. Alternatively,
direct effects on some higher trophic taxa might result in changes to prey
assemblages, which then result in changes to other higher trophic taxa
indirectly through alteration of competitive interactions or through vari-
ous trophic cascades and distant trophic linkages (Fairweather 1990, Yodzis
2000, Constable 2004).
 Setting management goals using information from predators 329

Local extinction

Decline in local Regime

genetic diversity shift

Limit
Death
Altered spatial and
temporal relationships

Condition Displacement

Effect
Target - containment of effects

Habitat
Time

Fig. 22.1 Schematic showing the potential pathways of effects of human

activities on higher trophic levels over time. The target of ecosystem-based
management is to contain the effects within reasonable bounds shown by the
darker grey area, which relates to (a) indirect effects on higher predators caused
by direct effects on lower trophic levels or (b) sublethal direct effects of some
activities. The limit relates to what needs to be avoided as a product of all human
activities.

The maintenance of productive exploitable populations is the goal of

fisheries managers. It is well known that productivity cannot necessarily
be maximized simultaneously for all harvested populations, particularly for
exploited populations from different trophic levels; exploitation at lower
trophic levels reduces the maximum possible production at higher trophic
levels (May et al. 1979, Beddington & May 1982). Natural inter-annual vari-
ability in population demographic parameters also makes an ecosystem
sufficiently variable that population and ecosystem objectives need to be
couched in terms of probability distributions rather than in a deterministic
form. The probabilities of given population states at a specified time will
depend on trajectories of the interlinked populations in a food web along
with variation in the physical environment.
Often, the expectation of long-term productivity of an exploited popu-
lation implies that the rest of the ecosystem will remain in some compar-
atively ‘stable’ state. It also implies that the harvest strategy used to con-
trol the population will not have an impact on the ecosystem structure
and function. What confidence is there that these assumptions are true?
More importantly, are the target and threshold (limit) levels for populations
330 A. J. Constable

appropriately set to achieve these expectations? Also, is it correct to assume

that populations will be able to recover to target or ‘pristine’ levels if declines
are arrested before the limit level is reached? What changes to the ecosystem
might have occurred that might prevent such recovery?
Limits or thresholds for populations and ecosystems need to incorpo-
rate two concepts: (a) the maintenance of genetic diversity within popula-
tions to enable those populations to be evolutionarily robust against future
environmental changes, and (b) the maintenance of populations to avoid
human-induced regime shifts in food webs and ecosystems (Fig. 22.1).
The approach to maximize productivity in fisheries aims, for many top
predators, to reduce the population to a level at which density-dependent
effects are almost negligible. However, populations experiencing intraspe-
cific competition may well have greater controlling influences on lower
trophic structures as keystone predators (Paine 1980) than those not expe-
riencing such competition. To that end, reducing the role of one or many
higher trophic species may have more distant ramifications to food-web
structure and function than expected (e.g. Paine 1980, Yodzis 2000), even
before higher trophic levels are substantially reduced or depleted (Pauly
et al. 1998). Thus, ecosystem objectives may require thresholds for exploited
or impacted populations to be higher than single-species models would
suggest.

MANAGING DIRECT EFFECTS ON POPULATIONS

Fisheries terminology for operational objectives now revolve around target

and threshold (limit) levels for populations. For exploited species, target lev-
els have been discussed relative to maximum, now optimum, sustainable
yields (Larkin 1996); while threshold levels are usually considered as those
levels below which recruitment becomes impaired. Rules of thumb have
been somewhere around 50% and 20% of pre-exploitation levels respec-
tively (see Beddington and Cooke 1983). In recent years, uncertainty and the
natural dynamics of populations and ecosystems have being considered in
the development of management strategies (FAO 1996, de la Mare 1998,
Cooke 1999, Sainsbury et al. 2000) such as that achieved in the precau-
tionary approach of the CCAMLR (see Constable et al. (2000) for review;
Box 22.3). This requires that probabilities of given outcomes be incorpo-
rated into management objectives and rules governing the decisions to con-
trol human activities, such as fishing or the disposal of waste (pollution).
The assessments of the CCAMLR advanced the work of Beddington
and Cooke (1983). They use the usual fisheries assessment data – such as
 Setting management goals using information from predators 331

Box 22.3 Precautionary approach of the CCAMLR for setting

catch limits of Antarctic krill, Euphausia superba
The CCAMLR was the first international commission to imple-
ment a precautionary approach for setting catch limits for the impor-
tant prey Antarctic krill, Euphausia superba. It agreed on target and
threshold levels for a target population, and set operational objec-
tives and decision rules incorporating variability and uncertainty in
population processes, notably recruitment. These objectives are set
in reference to a median expectation of the pre-exploitation (prior to
the effects of fishing) spawning biomass. Target levels for escape-
ment are governed by the median expectation for the spawning
biomass relative to the pre-exploitation median following the full
implementation of the harvest strategy over one generation time of
the target species. The threshold level is in reference to the prob-
ability of the stock declining below the limit, such as 20% of the
pre-exploitation median (see Constable et al. (2000) for review). This
approach advanced the work of Beddington and Cooke (1983) to esti-
mate a long-term annual yield based on a survey of the biomass
of the target species, in this case Antarctic krill, and estimates of
the life-history parameters of krill including recruitment variabil-
ity. Monte Carlo simulations are used to integrate across uncertain-
ties in the input parameters and natural (recruitment) variability
to determine the annual catch that could be taken while satisfying
a three-part decision rule aimed at satisfying the objectives of the
CCAMLR, including ecosystem objectives (Butterworth et al. 1994,
de la Mare 1996). In this case, the requirements of krill predators
were allowed for by setting the target escapement of krill at 75% of
the median pre-exploitation biomass.
Yield is calculated as a proportion (γ ) of an estimate of the pre-
exploitation biomass (B0 ). The three-part rule for krill is:

(a) choose γ 1 , so that the probability of the spawning biomass

dropping below 20% of its pre-exploitation median level over a
20-year harvesting period is 10%;
(b) choose γ 2 , so that the median krill escapement in the spawning
biomass over a 20-year period is 75% of the pre-exploitation
median level;
332 A. J. Constable

1.75
A B C
1.50
Relative biomass
1.25

1.00

0.75

0.50

0.25

0
0 10 20 30
Years

Fig. 22.2 Monte Carlo projection of a population model. See text for further
details.

(c) select the lower of γ 1 and γ 2 as the level for calculation of krill
yield.
Figure 22.2 (after Constable et al. 2000) illustrates the results of
Monte Carlo projections of a population model and how the derived
statistical distributions of krill abundance are used to determine the
lowest γ (criterion 3) from the two estimates of γ derived from
the first two criteria. Distribution A is the potential unexploited
biomasses derived from the model, which takes into account the
effects of uncertainties in krill demography. Distribution B is the
statistical distribution of lowest population biomasses over 20 years
of simulation under a specified catch limit of γ 1 B0 . Distribution C is
the statistical distribution of krill abundance at the end of 20 years
of exploitation under a specified catch limit of γ 2 B0 . Distributions
B and C incorporate uncertainties in the estimate of B0 . The dot–
dash line represents the median pre-exploitation biomass derived
from distribution A. The large-dashed line is the critical point of
the ‘recruitment criterion’ relative to the median pre-exploitation
biomass against which the tenth percentile of distribution B is com-
pared. The dotted line is the critical point of the ‘predator criterion’
relative to the median pre-exploitation biomass against which the
median of distribution C is compared.
 Setting management goals using information from predators 333

estimates of biomass, age structure and demographic parameters – in a

Monte Carlo simulation framework, which is used to determine the prob-
abilities of meeting these operational objectives given specified harvest
strategies (see Constable (2002, 2004) for review; Box 22.3).
At present, the CCAMLR has accepted that target levels for the spawn-
ing biomass of exploited top predators, such as the Patagonian toothfish
(Dissostichus eleginoides), should be at 50% of the pre-exploitation median
spawning biomass. It has not yet considered whether such a target, and the
threshold level of 20%, is appropriate for maintaining ecosystem processes
at lower trophic levels.

MANAGING INDIRECT EFFECTS

Data on the status and function of marine ecosystems are mostly centred
on targeted fish species in fisheries. Consequently, this discussion focuses
on the indirect effects of fishing even though the term ‘fished species’ could
be used to refer also to species directly affected (killed) by pollution or other
human activities.

Accounting for indirect effects in assessments of target species

In its precautionary approach, and given its ecosystem-based objectives,

the CCAMLR has adopted a different target level for fished species if they
are considered to be important prey for higher predators. In the case of
the Antarctic marine ecosystem, the fishery for Antarctic krill (Euphausia
superba) could potentially have an impact on many of the fish, bird and
marine mammal populations in the region because of its primary impor-
tance in the food web (Everson 1984). In this case, the requirements of krill
predators were allowed for by setting the target escapement of krill at 75%
of the median pre-exploitation biomass. As described in Constable et al.
(2000), this level of escapement split the difference between no fishing
(allow completely for the requirements of predators) and the usual ‘rule-
of-thumb’ level of 50% for maximizing productivity of target species (not
allowing any escapement for predators) until more information became
available on predator requirements (see Thomson et al. (2000) for initial
work in this regard). Work is continuing in the CCAMLR to refine this and
other ecosystem-based approaches (Constable 2002).
The indirect effects of exploitation of top predators, such as large preda-
tory fish, on lower trophic levels could be managed in a similar way. At
present, the target escapement for an exploited species is at some level for
334 A. J. Constable

optimizing yield, a level at which intraspecific competition is minimized. In

the absence of information on the influence of a top predator on controlling
the abundances of prey species (A. J. Constable & N. Gales, unpublished
observations), it may be prudent to allow a greater escapement of a target
predator until a better understanding of its ecological role is obtained.

Managing indirect effects directly

Management objectives of top predators have been very much influenced by

the approaches to fisheries, couched in terms of individual populations or,
at most, a few interacting populations. This implies that targets and thresh-
olds for individual populations can be correctly assessed and all met simul-
taneously. This is unlikely to be achieved because of the mulitivariate nature
of ecosystems. A more ‘assemblage-oriented’ approach is desirable. Emer-
gent properties need not be identified to incorporate the multivariate nature
of effects into objectives.
The degree to which changes in any indicator can be attributed to a
particular human activity or environmental parameter will depend on the
design of the monitoring programme. Firstly, changes in indicators need
to be highly and unambiguously correlated to changes in the target species
affected by human activities. Secondly, the temporal and spatial scales of
the monitoring need to match the spatial and temporal scales of the pop-
ulation and ecosystem process of interest in order that the indicators are
correctly estimated (see Constable (2002) and (2004) for review).
The design of the monitoring programme will also be influenced by
whether the overall objective is (a) to maintain the ecosystem within some
specified multivariate bounds or (b) to limit the effects such that the ecosys-
tem does not depart too far from that which would be expected in the
absence of the activity/activities. The relative merits of these alternative
objectives have been discussed in detail elsewhere (e.g. Underwood 1990,
Constable 2002, 2004, Downes et al. 2002).
The second objective requires closed reference (control) areas of suf-
ficiently large enough scale without the activities in order to provide the
contrast needed to detect the ecosystem effects of the activities in the open
(impact) areas when they arise. Clearly, the latter experimental approach
could also be used to help identify the cause of changes if such an under-
standing is needed.
In this chapter, I consider the development of indicators and their ref-
erence levels in relation to wider ecosystem objectives for managing the
 Setting management goals using information from predators 335

indirect effects on top predators. In particular, I focus on how to identify

significant departures from a natural reference state estimated prior to the
introduction of the activity. Consideration is not given here to the use of
control areas, although that would be a preferable approach because they
would help identify how the natural environment might be changing and,
therefore, potentially influencing ecosystem function overall.
A desirable management strategy for detecting indirect effects on top
predators would be to detect changes prior to the occurrence of long-term
adverse effects on the abundance of the population. For top predators, mon-
itoring and other scientific research provides estimates of local abundances
of species, reproductive rates and various measures of individual condition
and diet (species and quantity). For many marine mammals and birds, mon-
itoring is mostly restricted to accessible, localized breeding colonies rather
than across whole populations. In these monitoring programmes, eg. the
CCAMLR Ecosystem Monitoring Program (Agnew 1997), many attributes
of these predators will be measured at some regular interval, usually annual
(see Croxall (Chapter 11 in this volume)).
These measures may be directly related to the availability of a harvested
species to the predator at the time of sampling (diet) or may integrate the
effects of a number of factors over time prior to measurement (mortality
and body mass) (Murphy et al. 1988). The utility of each measure for iden-
tifying the effects of human activities will be contingent on the degree to
which changes in a parameter reflect the immediate effects of that activity
or the potential for that activity to have an effect in the future (de la Mare &
Constable 2000).
Decisions are most easily made when considering only one or a few
quantities and that action is triggered when critical values are detected.
A great difficulty in this regard is to be able to synthesize many variables
from a monitoring programme into management advice and/or decisions
when they may not be varying in a consistent way. In addition, a multivariate
‘envelope’ that defines a natural or reference state is very difficult to quan-
tify for consistent use over many years. A method for creating a single index
from many related parameters was proposed by de la Mare and Constable
(2000); and further developed by Boyd and Murray (2001) as the composite
standardized index (CSI) (see Croxall (Chapter 11 in this volume) and Box
22.3). Criteria for the inclusion of predator parameters are also developed
in de la Mare and Constable (2000), who indicate that parameters should
be highly correlated and related to the target prey species for which indirect
effects are being monitored.
336 A. J. Constable

The CSI is not intended to capture emergent ecological properties of a

predator–prey complex but to indicate whether predator ‘performance’ is
changing over time. A CSI can be related to indices of abundance of target
species (Boyd & Murray 2001, Boyd 2002) and/or to other environmen-
tal factors. De la Mare and Constable (2000) indicate a work programme
important for validating the use of such an index in management.
Such univariate indices developed from multivariate data can be used to
indicate general departures of predator performance from the norm. Criti-
cal values that bound the ‘natural’ state could be determined from a base-
line dataset with any values falling outside of those bounds considered to
be anomalous (de la Mare & Constable 2000). This is akin to the approach
adopted in the CCAMLR for predator indices considered to be related to
krill availability, where anomalies are values of parameters outside of the
95% confidence intervals (see Constable (2002) for review). Alternatively,
the values could be compared with areas without fishing to detect if fishing
has caused a change in predator performance. How this approach will be
used in a decision framework by CCAMLR has not been resolved, although
an increased frequency of departures would be expected to result in action
to change the harvest strategies, in this case of krill.
Boyd and Murray (2001) showed that a combined index of summer
parameters for land-based krill predators at South Georgia in the southwest
Atlantic can be related to estimates of krill density from the local area using
a non-linear asymptotic function (Fig. 22.3). Boyd (2002) argued that the
index was related to the fitness of those predators. He proposed that overall
predator fitness could be retained at approximately existing levels if the fish-
ery only had access to the foraging grounds when krill was at such a density
that the fishery would cause little change to fitness. Thus, it was proposed
to identify the critical density of krill below which substantial reduction in
fitness might arise. If the krill density was below this level then the fishery
was proposed to be closed (see Constable (2002) for a discussion of this
proposal).
An important consideration in developing these relationships is to deter-
mine the measure that most reflects the availability of krill to predators. This
is likely to be governed by the density of krill most commonly observed in
the foraging area by the predator; if this were to be represented by a sin-
gle quantity then the median density may be the most appropriate measure
(rather than the mean).
Monitoring the availability of a prey species could be costly compared
with monitoring the attributes of land-based predators. The non-linear
 Setting management goals using information from predators 337

Critical density

CSI

50% fitness

Krill density
Fig. 22.3 A functional relationship between predator fitness (approximated by a
composite standardized index, CSI) and local krill density in the local foraging
area; this relationship is used to identify a critical density below which predator
fitness would be expected to be significantly below the norm. After Boyd
(2002).

nature of the relationship between a CSI for top predators and the availabil-
ity of krill provides the foundation for considering what might constitute
anomalies in a CSI without having to measure krill availability associated
with a value of the CSI (Constable & Murphy 2003). In this case, an anomaly
is a value of the CSI below which predator performance would be considered
to have been affected by a reduction in krill availability. The method to iden-
tify an anomaly, with high statistical power, will need to take account of vari-
ability in the CSI for a given level of krill density. Therefore, an anomaly for
the CSI could be classified as values less than, say, the lower 0.1 percentile
of CSI values in the baseline dataset recorded for krill densities greater than
the critical level of krill availability. This would probably be a CSI value
greater than the 0.1 percentile for all values of the CSI, which is the cur-
rent approach (Fig. 22.3).
One of the difficulties in the approach to monitor higher predator per-
formance is to weight appropriately (statistically) the different parameters
in the analysis. They should be highly correlated to changes in the abun-
dance of the species being directly affected, such as targeted fish species,
with the degree of correlation remaining independent of the characteristics
(e.g. abundance) of the target fish species or other characteristics of the food
338 A. J. Constable

web or environment. If not, then estimated changes in the parameter con-

dition or in the multivariate index may not effectively signal when action is
or is not required.
An alternative approach might be to determine, through monitoring,
whether the change in abundance of fished or other impacted species
causes a predictable change in the production of their predators, irrespective
of the abundance of predators. Constable (2001, 2004) discusses the formu-
lation of objectives based on productivity arising from fished species (Box
22.4), which is that part of total production of the predator estimated to
have come from the consumption of fished species. This provides a form
of statistical weighting of the information from the different predators used
in the CSI, taking account of the predators’ abundance and dependence on
the fished species.
In the general case, the assumption is that the catch limits are derived
with sufficient confidence that the median annual production of preda-
tors arising from fished species will not be reduced by more than the
expected reduction in median biomass of the target species, although this
may be modified according to the expected increased production of recov-
ering species. In terms of maintaining ecological relationships, is there a
minimum level of production arising from fished species necessary to pro-
vide relative stability in or maintenance of the food web?
Constable (2001) proposes an operational objective that aims to main-
tain the production of predators arising from the consumption of fished
species (an index, W) at or above some limit reference point. A subsidiary
objective would be to ensure that the productivities of individual predator
species are not disproportionately affected even though the overall objective
is satisfied. The expected outcome of these objectives is that the contribu-
tions of different species to the food-web structure would remain largely
unaltered through fishing, thereby maintaining ecological relationships in
the system. This will result in attention being given to the primary interac-
tion between fished species and their predators (discussed in Andrewartha
and Birch (1984)) rather than examining the consequences of secondary
and other indirect interactions distant in the food web from the fished
species.
This approach takes account of the hierarchy of objectives relating to
the effects of fishing on the productivity of a system and the potential
for changes to the food web. It can easily be made general for systems
much more complex than the Antarctic and for which fisheries are already
present. For example, the Eastern Bering Sea food web (Trites et al. 1999)
 Setting management goals using information from predators 339

Box 22.4 Operational objectives based on predator productivity

Production of a species in a given year is related to the accumu-
lation of biomass through growth of individuals and reproduction.
Production Pp,y (in mass) of a predator, p, in a given year, y, can be
represented by the following equation, which includes variation in
some parameters with age, a:


1

Pp,y = R p,y B̂ p,0,y + Np,a,y (t) B̂ p,a,y (t)G p,a,y (t) dt

a>0
0

where Np,a,y is the number at age in that year, B̂ p,a,y is the individual
mass of the predator at age in that year, G p,a,y is the age-specific
growth rate of individuals, R p,y is the number of offspring in that
year and t is the proportion of the year passed.
Constable (2001) proposed an operational objective for higher
trophic levels focusing on limit reference points for production aris-
ing from fished species rather than overall production of predators of
fished species. In a given year, y, the production arising from fished
species, P̃ p,y , can be specified as a fraction of the overall production
of the predator, such that
 F
d p,y, j A p, j
j =1
P̃ p,y = Pp,y

D
d p,y,i A p,i
i =1

where F is the number of species that are caught in fisheries, i.e.

fished species, D is the number of species and d is the overall pro-
portion of a given species in the diet.
After fishing begins, the acceptable degree of change (limit ref-
erence point/threshold) in production arising from fished species
would depend on the objectives for individual species. For a given
predator, the average production arising from fished species dur-
ing the baseline period for Y years could be the reference level. The
limit reference points for individual predators would be a propor-
tion of this, ap , and need not be the same for all predators, depend-
ing on the conservation requirements and target level of recovery
for that species as well as the dependency of those predators on the
fished species; e.g. ap may need to be greater than 1 for predators
that require recovery.
340 A. J. Constable

1.0

0.8

0.6
W
0.4

0.2

0.0
0 10 20 30 40 50 60
Year

Fig. 22.4 Hypothetical time series of W over 60 years. After Constable (2001,
2004).

Thus, the limit reference point for average production arising

from fished species and combined for all predators during the fish-
ing period would be


y
P̃ p,y

y =1
Wref = ap
p Y

This formulation of a threshold status of the food web relative to

the fished species explicitly provides for both dependent species as
well as for the recovery of species. If important, the subsidiary objec-
tive for an individual predator could be derived from the component
of this equation related to that predator. The index may need to be
restricted to a subset of predators for logistical reasons. In such a
case, the robustness of the index to the choice of predators would
need to be considered.
Figure 22.4 shows a hypothetical time series of W over 60 years.
No fishing occurs in the first 20 years, over which time a baseline
W series was used to estimate the baseline mean, Wo (solid circle)
and calculate Wref ( a = 0.8 for all species) (filled diamond). The box
and whisker plots show the frequency distribution of W during that
baseline period. The box and whiskers adjacent to W0 show the rela-
tionship between the mean and the distribution of values. This rela-
tive distribution is applied as the expected distribution around Wref
 Setting management goals using information from predators 341

during the management period. The upper and lower dotted lines
show possible critical upper and lower range limits, WH and WL ,
for a case when the critical acceptable frequency outside the limit is
0.25. The dashed line refers to a possible interim upper range limit,
WiH , during the period when the system is adjusting to the fishing
activity. In this example, the trend for W to remain below the lower
range limit after 16 years of fishing would signal that a reduction in
fishing was required.

could be considered in this way (Fig. 22.5). The important part of this assess-
ment is to divide the system into a number of groups:

(1) Fished species. This managed system is where all species presumably
have target levels and/or threshold reference points applicable to them.
(2) Dependent predators of fished species. The effect of fishing on this
group can be considered as a whole – i.e. the effect of lost production
in the system – or could be subdivided to explore the effects on
individual species or groups of species.
(3) Prey of fished species and/or alternative prey of those predators. These
taxa might assume greater importance in the diet of predators and/or
might increase their productivity as a result of reductions in
abundance of their predators and competitors.
(4) Predators of the non-fished prey species in the third group. The
response of these predators would be difficult to foreshadow without
good knowledge of the function of the food web.

If the relationship between predator fitness and the abundance of fished

species is non-linear then the threshold for production arising from fished
species would be expected to relate directly to the optimal fitness levels dis-
cussed above.

CONCLUDING REMARKS

Many current discussions surrounding marine-ecosystem management

seem to be in pursuit of defining target states for ecosystems. This implies
a capacity for us to engineer desirable attributes of whole ecosystems. The
science of ecology is not sufficiently mature to evaluate whether such a
proposition is tractable and some would question whether assessment and
management practice will ever be up to the task. To that end, the precau-
tionary approach would seem to be prudent in trying to deliver successful
342 A. J. Constable

Pacific walrus
Birds Toothed whales Sperm whales Beaked whales Bearded seals Baleen whales

Steller sea lions Seals

Deepwater fish

Pelagics Cephalopods

Large zoopl. Herb. zoopl.

Large flatfish Adult pollock
Juv. pollock
Crabs and
Small flatfish O. dem. fish shrimps Infauna Epifauna

Fig. 22.5 Food web of the Eastern Bering Sea showing primary interactions
with fished species along with other marine mammals and birds (based on data
and taxonomic groups from Trites et al. (1999)). Fished taxa are indicated by the
dark-grey boxes (O. dem., other demersal). Predators of fished taxa are in
light-grey boxes. Other taxa are in white boxes (Herb. zoopl., herbivorous
zooplankton; Juv., juvenile). Arrows indicate direction of prey to predators. Solid
lines indicate predation on fished species, dotted lines indicate predation on
non-fished species. The heavier weighted lines indicate where prey make up at
least 50% of the diet, lighter lines are where prey make up at least 20% but less
than 50% of the diet. Interactions where prey make up less than 20% of the diet
are not shown. After Constable (2001).

ecosystem management in the longer term; the revised principles of Mangel

et al. (1996) indicate what needs to be achieved.
As an example, alternative stable states can occur naturally for some sys-
tems resulting in vastly different assemblage and food-web structures (e.g.
regime shifts, Steele 1998). The mechanisms and conditions that might
trigger shifts of large-scale ecosystems to other stable states are not well
known, nor are the characteristics of potential alternative states. Manipula-
tion of species might result in the system being triggered to shift into an
alternative stable state. Further, the potential for return to historical, per-
haps more desirable states may be impaired by the reduction in abundance,
species or genetic diversity of those assemblages. In the first instance,
it would be desirable to manage activities in such a way that the effects
of those activities are consistent with the scale of natural disturbances
and that we establish appropriate monitoring and research to learn more
about the ecosystem consequences of these activities. This will help assist
 Setting management goals using information from predators 343

in developing appropriate ecosystem-based management practices in the

future.
It is now commonly considered that fishing is a large-scale experiment
(Beddington & de la Mare 1985, Walters 1986, Ludwig et al. 1993, Man-
gel et al. 1996). The principles for designing a monitoring programme to
adequately test for the effects of an environmental perturbation, such as
fishing, are well described in the literature (Underwood 1990, Stewart-
Oaten et al. 1992, Mangel et al. 1996, Downes et al. 2002). Ultimately the
power of the programme for management purposes will be dictated by
the spatial and temporal relationships between the predator populations,
the harvested population, the harvesting activities and the monitoring activ-
ities, as well as the procedures that have been put in place to translate
the observed changes into management action. The issues concerned with
designing field monitoring programmes in large-scale marine ecosystems
are discussed in detail by Constable (2002, 2004).
An important consideration in using field data concerns what indicators
can be used to signal when, and by how much, population-wide and regional
effects are likely to occur and when action might need to be taken to alter
the human activities. The overall process of specifying operational objec-
tives, decision rules, monitoring indicators and assessments is the process
of defining management procedures or strategies. Once defined, there are
simulation procedures that can be used to evaluate whether the manage-
ment procedure would be expected to be successful despite uncertainties
in knowledge of how the ecosystems function, as well as uncertainties (bias
and precision) in the data being collected (e.g. de la Mare 1986, 1998, Smith
1993, Cooke 1999, Sainsbury et al. 2000).
Returning to the general objectives described by Mangel et al. (1996),
our task is to find the management systems that conserve the overall natu-
ralness of ecosystems, including top predators, to ensure those ecosystems
are robust and resilient to the changes and pressures of our future world.
In this way, the desirable services of the ecosystems currently enjoyed by
our generation have the greatest chance of being retained into the future
despite natural variability and uncertainties in our knowledge.

ACKNOWLEDGEMENTS

Many thanks to Bill de la Mare, Ian Boyd, Campbell Davies, John Croxall,
Inigo Everson, Keith Reid, Eugene Murphy, Phil Trathan, Marc Mangel,
Graeme Parkes, David Agnew, Denzil Miller, Roger Hewitt, Keith Sains-
bury and Tony Smith for many stimulating and challenging discussions
344 A. J. Constable

over the years about ecosystem-based management. Also, thanks to Camp-

bell Davies, Di Erceg, Ian Boyd and an anonymous referee for comments
on the manuscript. Lastly, I would like to thank my colleagues in CCAMLR
and at the Australian Antarctic Division for the many conversations and
meetings that contributed to the substance of this chapter.

REFERENCES
Agnew, D. J. (1997). The CCAMLR Ecosystem Monitoring Program. Antarc. Sci.,
9, 235–42.
Andrewartha, H. G. & Birch, L. C. (1984). The Ecological Web: More on the
Distribution and Abundance of Animals. Chicago, IL: University of Chicago
Press.
Beddington, J. R. & Cooke, J. G. (1983). The Potential Yield of Fish Stocks. Fisheries
Technical Paper 242. Rome, Italy: FAO.
Beddington, J. R. & de la Mare, W. K. (1985). Marine mammal–fishery interactions:
modelling and the Southern Ocean. In Marine Mammals and Fisheries, eds. J. R.
Beddington, R. J. H. Beverton & D. M. Lavigne. London: George Allen &
Unwin, pp. 94 –105.
Beddington, J. R. & May, R. M. (1982). The harvesting of interacting species in a
natural ecosystem. Sci. Am., 247, 42–9.
Beverton, R. H. J. & Holt, S. J. (1957). On the Dynamics of Exploited Fish Populations.
London: HMSO.
Boyd, I. L. (2002). Integrated environment–prey interactions off South Georgia:
implications for management of fisheries. Aquat. Conserv., 12, 119–26.
Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
Butterworth, D. S., Gluckman, G. R., Thomson, R. B. & Chalis, S. (1994). Further
computations of the consequences of setting the annual krill catch limit to a
fixed fraction of the estimate of krill biomass from a survey. CCAMLR Sci., 1,
81–106.
Chesson, P. L. & Case, T. J. (1986). Overview. Nonequilibrium community
theories: chance, variability, history, and coexistence. In Community Ecology,
eds. J. Diamond & T. J. Case. New York: Harper & Row, pp. 229–39.
Constable, A. J. (2001). The ecosystem approach to managing fisheries: achieving
conservation objectives for predators of fished species. CCAMLR Sci., 8, 37–64.
(2002). CCAMLR ecosystem monitoring and management: future work.
CCAMLR Sci., 9, 233–53.
(2004). Managing fisheries effects on marine food webs in Antarctica: trade-offs
among harvest strategies, monitoring, and assessment in achieving
conservation objectives. Bull. Mar. Sci., 74, 583–605.
Constable, A. J. & Murphy, E. (2003). Annex 4, Attachment 3: using predator
response curves to decide on the status of krill availability: updating the
definition of anomalies in predator condition – preliminary analyses. In
SC-CAMLR, Report of the Twenty-second Meeting of the Scientific Committee for
the Conservation of Antarctic Marine Living Resources. Hobart, Australia:
CCAMLR, pp. 277–82.
Constable, A. J., de la Mare, W. K., Agnew, D. J., Everson, I. & Miller, D. (2000).
Managing fisheries to conserve the Antarctic marine ecosystem: practical
 Setting management goals using information from predators 345

implementation of the Convention on the Conservation of Antarctic Marine

Living Resources (CCAMLR). ICES J. Mar. Sci., 57, 778–91.
Cooke, J. G. (1999). Improvement of fishery-management advice through
simulation testing of harvest algorithms. ICES J. Mar. Sci., 56, 797–810.
de la Mare, W. K. (1986). Simulation Studies on Management Procedures. IWC
SC/37/O 14. Cambridge, UK: International Whaling Commission.
(1996). Some recent developments in the management of marine living
resources. In Frontiers of Population Ecology, eds. R. B. Floyd, A. W. Sheppard &
P. J. De Barro. Melbourne, Australia: CSIRO Publishing, pp. 599–616.
(1998). Tidier fisheries management requires a new MOP (management-
oriented paradigm). Rev. Fish Biol. Fish., 8, 349–56.
de la Mare, W. K. & Constable, A. J. (2000). Utilising data from ecosystem
monitoring for managing fisheries: development of statistical summaries of
indices arising from the CCAMLR Ecosystem Monitoring Program. CCAMLR
Sci., 7, 101–17.
Downes, B. J., Barmuta, L. A., Fairweather, P. G. et al. (2002). Monitoring Ecological
Impacts: Concepts and Practice in Flowing Waters. Cambridge, UK: Cambridge
University Press.
Everson, I. (1984). Marine interactions. In Antarctic Ecology, ed. R. M. Laws.
London: Academic Press, pp. 491–532.
Fairweather, P. G. (1990). Is predation capable of interacting with other
community processes on rocky reefs? Aust. J. Ecol., 15, 453–64.
FAO (Food and Agriculture Organization) (1996). Precautionary Approach to
Fisheries. Part 1: Report of the Technical Consultation. FAO Fisheries Technical
Paper 350. Rome, Italy: FAO.
Holt, S. J. & Talbot, L. M. (1978). New principles for the conservation of wild living
resources. Wildl. Mon., 59, 1–33.
Larkin, P. A. (1996). Concepts and issues in marine ecosystem management. Rev.
Fish Biol. Fish., 6, 139–64.
Ludwig, D., Hilborn, R. & Walters, C. (1993). Uncertainty, resource exploitation,
and conservation: lessons from history. Science, 260, 17–36.
Mangel, M., Talbot, L. M., Meffe, G. K. et al. (1996). Principles for the conservation
of wild living resources. Ecol. Applic., 6, 338–62.
May, R. M., Beddington, J. R., Clark, C. W., Holt, S. J. & Laws, R. M. (1979).
Management of multispecies fisheries. Science, 205, 267–77.
Murphy, E. J., Morris, D. J., Watkins, J. L. & Priddle, J. (1988). Scales of interaction
between Antarctic krill and the environment. In Antarctic Ocean and Resources
Variability, ed. D. Sahrhage. Berlin: Springer-Verlag, pp. 120–30.
Paine, R. T. (1980). Food webs: linkage, interaction strength and community
infrastructure. J. Anim. Ecol., 49, 667–85.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, T. Jr (1998). Fishing
down marine food webs. Science, 279, 860–3.
Petraitis, P. S., Latham, R. E. & Niesenbaum, R. A. (1989). The maintenance of
species diversity by disturbance. Q. Rev. Biol., 64, 393–418.
Pimm, S. L. & Gilpin, M. E. (1984). Theoretical issues in conservation biology. In
Perspectives in Ecological Theory, eds. J. Roughgarden, R. M. May & S. A. Levin.
Princeton, NJ: Princeton University Press, pp. 287–305.
Quinn, T.-I. & Deriso, R. (1999). Quantitative Fish Dynamics. Oxford, UK: Oxford
University Press.
346 A. J. Constable

Sainsbury, K. J., Punt, A. E. & Smith, A. D. M. (2000). Design of operational

management strategies for achieving fishery ecosystem objectives. ICES J.
Mar. Sci., 57, 731–41.
Smith, A. D. M. (1993). Risks of over- and under-fishing new resources. In Risk
Evaluation and Biological Reference Points for Fisheries Management, eds. S. J.
Smith, J. J. Hunt & D. Rivard. Ottawa, Ontario, Canada: National Research
Council for Canada, pp. 261–7.
Steele, J. H. (1998). Regime shifts in marine ecosystems. Ecol. Applic., 8, S33–6.
Stewart-Oaten, A., Bence, J. R. & Osenberg, C. W. (1992). Assessing effects of
unreplicated perturbations: no simple solutions. Ecology, 73, 1396–404.
Thomson, R. B., Butterworth, D. S., Boyd, I. L. & Croxall, J. P. (2000). Modeling
the consequences of Antarctic krill harvesting on Antarctic fur seals. Ecol.
Applic., 10, 1806–19.
Tilman, D. (1999). The ecological consequences of changes in biodiversity: a
search for general principles. Ecology, 80, 1455–74.
Trites, A. W., Livingston, P., Vasconcellos, M. C. et al. (1999). Ecosystem Change and
the Decline of Marine Mammals in the Eastern Bering Sea: Testing the Ecosystem
Shift and Commercial Whaling Hypotheses. Fisheries Centre Research Reports 7.
Vancouver, British Columbia, Canada: Fisheries Centre.
Underwood, A. J. (1989). The analysis of stress in natural populations. Biol. J. Linn.
Soc., 37, 51–78.
(1990). Experiments in ecology and management: their logics, functions and
interpretations. Aust. J. Ecol., 15, 365–89.
Walters, C. (1986). Adaptive Management of Renewable Resources. New York:
MacMillan.
Wilson, E. O. & Peter, F. M. (1988). Biodiversity. Washington, DC: National
Academy Press.
Yodzis, P. (1994). Predator–prey theory and management of multispecies fisheries.
Ecol. Applic., 4, 51–8.
(1996). Food webs and perturbation experiments: theory and practice. In Food
Webs Integration of Patterns and Dynamics, eds. G. A. Polis & K. O. Winemiller.
New York: Chapman and Hall, pp. 192–200.
(2000). Diffuse effects in food webs. Ecology, 81, 261–6.
 23
Marine reserves and higher predators
S. K. HOOKER

Marine-ecosystem management is not simple. In order to predict the effects

of any management activities on other components of the system, complex
ecological modelling is often required. Marine reserves have been suggested
as a conservation tool that can bypass the need for complex and often con-
troversial ecological models. To date, marine predators have attracted sig-
nificant attention in ocean conservation planning, but they have primarily
been used as figureheads, largely obscuring any potential ecological role
as indicator species. Their distribution can help identify productive ocean
areas, the protection of which will encompass a high measure of biodiver-
sity within the underlying ecosystem. In this chapter, I review the evidence
supporting marine reserves over ecosystem modelling approaches, and dis-
cuss the potential to use marine megafauna in order to identify sensitive
marine habitats.
The seas have been increasingly altered by the effects of humans (Jack-
son et al. 2001) and the risk of extinction to marine species is far greater
than has often previously been thought (Roberts & Hawkins 1999). The
most pervasive of these effects is over-fishing; but other significant threats
include pollution, degradation of water quality, habitat destruction and
anthropogenic climate change. Fisheries now consume an estimated 24%
to 35% of primary production (Pauly & Christensen 1995). In many cases
this has resulted in extinctions both of target species that are directly har-
vested (e.g. Myers et al. 1997) or of incidentally caught species additional
to the target catch (e.g. Casey & Myers 1998). Even parts of the ocean pre-
viously relatively untapped, such as the deep sea, are now facing potential
increased exploitation (Roberts 2002).

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
348 S. K. Hooker

The need for an ecosystem approach to marine conservation extends to

the conservation of higher trophic levels. These species appear to play an
important role in the maintenance of ecosystem function (Bowen 1997).
In fact, elimination of top predators from some areas has led to top-down
effects, contributing to the degradation of some coastal ecosystems (Jack-
son et al. 2001). Ecosystem shifts have been observed in other systems fol-
lowing changes in the populations of top marine mammal predators (Estes
et al. 1998). Similar changes in offshore oceanic ecosystems would probably
be massive in scale and difficult to reverse (Worm et al. 2003). Considera-
tion of higher trophic levels in marine conservation initiatives has generally
focused on individual endangered species or taxa of economic importance.
However, initiatives based on an ecosystem approach can use these higher
predators as indicators to establish protection for the wider ecological com-
munity that supports them.

MODELLING APPROACHES

Traditionally, marine-ecosystem management has taken the form of

exploitation quotas and closure of regions to exploitation, either perma-
nently or periodically depending on the perceived vulnerability of the
resource being conserved. More recently, a modelling approach has been
advocated. Yet despite advances in this field, the number of parameters that
need to be estimated to implement such models is usually prohibitive. The
resulting large requirement for data is often unrealistic. Our ability to quan-
tify ecosystems and to measure and predict ecological complexity there-
fore remains poor (Box 23.1). A food-web could be represented by a simple
three-step chain: fish prey–fish–fish predators, or a more realistic approach
would incorporate the complex myriad of relationships between different
predators potentially feeding on each other at different life stages, or com-
peting for mutual prey in a spider’s web of ecological relationships (see
Yodzis 1998). Determining which of these linkages are important to mod-
elling the system is often not straightforward. These multispecies effects –
predator–prey interactions, competition and mutualistic interactions – can
cause unpredictable changes in community structure and non-target effects
of management interventions. We see historical legacies in which ecosys-
tem structure was radically altered, resulting in dramatic and potentially
irreversible shifts in population demographics (May 1979, Worm & Duffy
2003). Removal of a top predator can cause trophic cascades, precipitating
dramatic reductions in abundance of species at lower levels (Worm & Duffy
2003).
 Marine reserves and higher predators 349

Box 23.1 Difficulties inherent in ecosystem modelling

Although models have successfully been applied to lower ecosys-
tem levels, their application over broader scales is difficult (de Young
et al. 2004). Several problems become apparent when applying such
models to higher trophic levels.

(1) The requirements for estimation of a large number of

parameters.
(2) Data are often unavailable or sparse:
r the presence of complex life histories, and often variations in
diet at different life stages, require structured demographic
modelling;
r complex behaviours – mate competition, mate choice and
mating systems – which can affect population growth rate
deleteriously during periods of intense exploitation (Rowe &
Hutchings 2003);
r the need for long time series of data to capture population
dynamics, and effects of behaviour, e.g. site fidelity;
r the presence of competition, mutualism and predator–prey
dynamics, necessitating additional submodels (e.g. functional
response);
r variation in species abundance and distribution over
ecosystem scales, which are often, in marine ecosystems, very
difficult to measure and may for some species be completely
lacking.
(3) A high degree of uncertainty associated with the representation
of processes and associated parameters, the incorporation of
which should be explicit in both parameter estimation and risk
assessment (Harwood & Stokes 2003).
(4) The resulting need to generalize and simplify ecosystem
structure or predator–prey relationships.
(5) Differences in basic philosophy for management models versus
biological models, in terms of the need for simplicity and ready
parameter estimation for good management (Taylor et al.
2000).
350 S. K. Hooker

Many previous attempts to manage marine ecosystems have therefore

not been as successful as anticipated, and several authors have suggested
that reserves, i.e. spatially explicit management areas, have greater poten-
tial to improve sustainability (Botsford et al. 1997, Roberts 1997). Since
reserves make no attempt to alter ecosystem function, but allow portions
of the ecosystem to remain intact, they have great precautionary potential.
In terrestrial systems, the protection of spatially explicit areas and the con-
servation of biodiversity has been viewed as a worthwhile way of protecting
several species at once (Myers et al. 2000). In the marine environment,
this approach has recently been applied to coral reefs (Roberts et al. 2002)
and cetaceans (Hooker et al. 1999). However, less than 1% of the world’s
marine area is designated as protected, and many of these areas have only
been declared in the last two decades. In contrast, over 5% of national land
area forms protected areas, many of which have been protected for decades
and some for centuries (Boersma & Parrish 1999).

MARINE RESERVES

Fully protected marine reserves can be defined as ‘A geographically

delimited area of the ocean completely protected from all extractive
and destructive activities.’ Fully protected marine reserves therefore have
explicit prohibitions against fishing and the removal or disturbance of liv-
ing or non-living marine resources. These may also be termed ‘ecological
reserves’ or ‘no-take areas’, and are a special class of marine protected areas
(MPAs). MPAs are defined as ‘areas of the ocean designated to enhance con-
servation of marine resources’, although the actual level of protection may
vary widely. Most allow fishing but are closed to activities such as oil and gas
extraction (Lubchenco et al. 2003), an ironic situation given the high levels
of regulation of the latter compared with the former. However, the social
implications of banning fishing are often prohibitive, making this one of
the biggest challenges to the establishment of marine reserves.
It is the spatial nature of these reserve areas that confer their benefits
when compared with other management measures – focusing on the whole
ecosystem, rather than providing a solution for one ecosystem component
(Noss 1996). Such reserves can confer multiple benefits (Box 23.2).

CAN RESERVES BE BENEFICIAL FOR HIGHLY

MOBILE ANIMALS?

A protected area would ideally encompass the majority of a species’ distribu-

tional range (Reeves 2000), and this characteristic has been used to suggest
 Marine reserves and higher predators 351

Box 23.2 Ecosystem benefits of marine reserves (based on Gell

and Roberts (2003), Lubchenco et al. (2003) and Roberts (2003))
Conservation of biodiversity
Marine reserves provide benefits to all ecosystem components. In
some cases the successional effects caused by threats outside reserve
areas can be reversed, leading to alternative species dominance in
reserve areas.

Protection of habitats from destruction

Fishing gear such as trawls, and dredging or drilling structures, can
modify the physical habitat, and this can sometimes take decades
or centuries to recover. Reserves protect specific areas from such
damage in a way that simply reducing the frequency of such threats
cannot achieve.

Increased abundance, diversity and productivity

Species within reserves live longer than those in exploited areas.
Most research has been done on fish in this context, which have
been shown to produce exponentially increasing numbers of eggs
with their size; so there tend to be increases in both fish size and fish
biomass following protection. Reserves therefore help in the recov-
ery of depleted stocks of exploited species.
Export of individuals to exploited areas
Reserves often extend their benefits outside their delimited area by
leakage or ‘spillover’ of biomass from these areas into neighbour-
ing areas. Larger fish produce more offspring, so reserves can make
disproportionately large contributions to fish population replenish-
ment, leading to net export outside the reserve boundaries. Thus the
impact of a reserve may be large relative to its area.

Protection of all ecosystem components

Balancing the anthropogenic benefits of exploitation and its associ-
ated threats, particularly to highly vulnerable species, is often diffi-
cult. These vulnerable species will be deleteriously affected by even
relatively low levels of threats, e.g. low fisheries pressure. Scaling
back exploitation to protect such species would drastically reduce
benefits (e.g. catches of more resilient species), and so managers
are reluctant to advise this. Marine reserves can provide the balance
352 S. K. Hooker

between protection and exploitation, allowing exploitation outside

reserve areas and providing refuges for vulnerable species.
Increased resilience to environmental variability
Anthropogenic impacts have affected the age structure of many
exploited populations such that species reproduce much younger
and for only a few years. This leaves populations highly vulnera-
ble when years are poor for offspring survival. Many marine species
have long life spans, presumably to see them through poor years.
Reserves allow the development of natural, extended age structures
that help populations persist, and provide insurance against envi-
ronmental or management uncertainty.

Sites for scientific investigation, baseline information, education,

recreation and inspiration
By maintaining ‘natural’ or ‘undisturbed’ areas, scientists are able to
establish baseline information to compare with non-reserve areas.

that marine reserves at sizes smaller than this will not provide protection for
mobile, wide-ranging species (Boersma & Parrish 1999). However, using
a modelling approach, Apostolaki et al. (2002) demonstrated benefits from
reserves primarily for over-exploited stocks of low-mobility species, but also
(to a lesser extent) for high-mobility species and under-exploited stocks. The
European Habitats Directive endorses Special Areas of Conservation (SACs)
which cover selected critical habitats across the whole range of a species.
Protection of critical or high-use habitats can confer protection, despite the
fact that reserves do not cover an entire species’ range.
Many higher predators are relatively site-faithful over time scales vary-
ing from daily (in the case of diurnal prey movements) to annual (in the
case of migratory species), to decadal (in the case of species following El
Niño events). Examination of animal movements often show certain high-
use areas (critical habitat or ‘hotspots’) at which protection would be valu-
able. Similarly protection should be provided for certain life-history stages
or other vulnerable stages, such as during migration (Gell & Roberts 2003).
Despite the fact that a predator might only use a protected area for a por-
tion of its life span, this would reduce the frequency with which it would
be exposed to some threats (such as fishery bycatch), and diminish the
overall cumulative impact of other threats (such as exposure to drilling
effluent).
 Marine reserves and higher predators 353

MARINE PREDATORS

Marine predators attract a great deal of attention in conservation planning,

and are often used as a lever to influence environmental policy. However, in
many ways their value as campaign figureheads has obscured their use as
ecological indicator species (species whose presence characterizes a particu-
lar habitat or biological community; Zacharias & Roff 2001). The spatial pro-
tection of such species would therefore extend to the underlying ecosystem
that supports their existence. The slow regeneration times and dynamics of
their populations means that, by definition, their presence within ecosys-
tems is indicative of a sustained underpinning trophic structure. In addi-
tion to protection for their ecosystem, marine reserves can provide benefits
to upper-trophic-level predators themselves (Hooker & Gerber 2004). Many
of the threats to these species – particularly physical threats such as ship
strikes or fisheries bycatch, competition with fisheries for prey resources,
or acoustic impacts – will be mitigated by spatial protection. Other threats
such as contaminant exposure will only be minimally reduced, if at all, by
the spatial boundaries of marine reserves, although the implementation
of such reserve areas would at least be likely to raise awareness of such
issues.
Marine mammal critical habitat can be defined in terms of the ecolog-
ical units required for successful breeding and foraging (Harwood 2001,
Hooker & Gerber 2004). For baleen whales, seabirds and pinnipeds, these
areas are often separated spatially (Box 23.3), whereas for odontocetes these
may occur in the same place. Many conservation efforts for marine mam-
mals have been based on protection of breeding habitat, such as the Año
Nuevo State Park, California which protects breeding northern elephant
seal habitat, or the Hawaiian Islands National Marine Sanctuary which
protects breeding grounds of humpback whales (Reeves 2000, Hooker &
Gerber 2004). There are numerous other examples from around the world
of protected areas being created around the terrestrial habitats used by pin-
nipeds for breeding or resting, but there are few that address the habitat
needs of these animals while they are at sea. This is because the bureau-
cratic processes for declaring terrestrial sites as protected areas is well
established and because it is simple to encompass extremely high spatial
aggregations of individuals in a small protected area. However, in many
cases, these animals do not face their greatest threats during the breed-
ing season. More attention needs to be directed towards marine mam-
mal foraging areas to establish how best to protect their access to food
resources.
354 S. K. Hooker

Box 23.3 Predator diversity hotspots

There is nothing so desperately monotonous as the sea

James Russell Lowe, nineteenth century poet
This is far from the truth; in fact the open ocean shows a rich struc-
ture in terms of species diversity, which may be based on the breed-
ing and feeding grounds of higher predators. These are often related
to the oceanographic structure of the ocean, and could be used to
focus future conservation efforts.
Life stages of some marine predators (e.g. baleen whales, pin-
nipeds and seabirds) are often separated spatially into discrete feed-
ing and breeding areas with migration between these. Each of these
areas may form spatial hotspots within which reserves can be placed
(see Fig. 23.1).
Oceanic hotspots of higher predator distributions are often asso-
ciated with topographic and oceanographic features. Three types of
oceanic hotspots have been identified (Hyrenbach et al. 2002).
(1) Static systems determined by topographic features. Example 1:
The Gully, a submarine canyon off Nova Scotia, eastern Canada.
Cetaceans show elevated abundances in the vicinity of this
bathymetric feature compared with levels in surrounding
regions. Their distribution is governed primarily by bathymetric
features and so could be well defined by spatial boundaries
(Hooker et al. 1999, 2002)
Example 2: Shannon Estuary, Ireland. Bottlenose dolphins
showed preferential use of the areas of the estuary with greatest
benthic slope and depth (Ingram & Rogan 2002).
(2) Persistent hydrographic features, such as currents and frontal
systems. Example: Bird Island, South Georgia. High numbers of
Antarctic fur seal and macaroni penguin are found to the
northwest of South Georgia where there appears to be a
persistent frontal system (Barlow et al. 2002).
(3) Ephemeral habitats, shaped by wind- or current-driven
upwelling, eddies and filaments. Example: warm core ring,
North Atlantic. Sperm whales are found primarily along the
periphery of the warm core ring (Griffin 1999).
 Marine reserves and higher predators 355

Breeding

Feeding

Migration
corridor

Feeding

Low High

Density
Fig. 23.1 Predator diversity hotspots. Based on Hooker and Gerber (2004).

PREDATOR DIVERSITY HOTSPOTS

Recent advances in telemetry have provided the ability to record predator

locations at sea (Cooke et al. 2004), and initiatives such as the census of
marine life are focusing attention on mapping biological distributions of
species in conjunction with measurement of ocean physics (O’Dor 2004).
Through this and other approaches, it is becoming apparent that certain
areas of the ocean form productive hotspots at which predators and prey
are aggregated (Worm et al. 2003; Box 23.3). These diversity hotspots often
coincide with topographic and oceanographic features. The increase in bio-
diversity found at these locations has been used in the terrestrial realm to
356 S. K. Hooker

form conservation priority areas (Myers et al. 2000). In a similar manner

such areas could be used to establish boundaries for reserves in the marine
realm (Hooker et al. 1999).
The overlaying of maps of different marine predators’ foraging habits –
together with a basic knowledge of their diet (e.g. piscivory or teuthophagy),
broader ecosystems (e.g. upwelling dynamics) and habitat variability (e.g.
persistence and spatial variation over annual and decadal cycles) – allows
researchers to identify hotspot features that could be designated for pro-
tection. There are three major types of oceanic hotspots: (a) static systems;
(b) persistent hydrographic features, such as currents and frontal systems;
and (c) ephemeral habitats, shaped by wind- or current-driven upwelling,
eddies and filiaments (Hyrenbach et al. 2000; Box 23.3). The last of these
are the most difficult to protect. However, with appropriate political will,
over some time scales or over a large enough spatial scale, protection could
be provided for these features. For example, although the upwelling to the
south of Monterey Bay in California is variable in location inter-annually, it
could be protected by delimiting a large area of ocean that would encompass
these inter-annual fluctuations. Policing of large-scale, open ocean areas
will be problematic, however, and is likely to rely on a wide range of domes-
tic and international bureaucratic measures to make such marine reserves
work effectively.

SOCIO-ECONOMIC CONFLICTS

One criticism of marine reserves has been that displacement of anthro-

pogenic impacts, e.g. fishing effort, from a closed area may cause more
harm if effort (and resulting bycatch, noise, pollution, etc.) needs to be
increased in order to catch the same quota elsewhere (Baum et al. 2003,
Worm et al. 2003). Closed-area models have been used to explore whether
closure of a hotspot to fishing represents a viable conservation option
(Worm et al. 2003). These models have investigated two possible outcomes:
‘constant-quota’ outcomes in which total fisheries catch is maintained con-
stant, and ‘constant-effort’ outcomes in which effort is constant regardless
of catch. Results were not always clear cut. Baum et al. (2003) found con-
servation benefits with constant-effort models but not with constant-quota
models. However, Worm et al. (2003) found that closure of hotspots to fish-
ing resulted in conservation benefits under both scenarios. Apostolaki et al.
(2002) also found constant-quota models to be intermediate between those
in which effort previously expended in the reserve was relocated into the
fished area, and those in which effort in the fished area was kept constant.
 Marine reserves and higher predators 357

Overall, it appears that careful design of marine reserves, if conducted in

concert with reductions in fishing effort, will always result in conservation
benefits.

RESERVE NETWORKS

Establishing a network of marine reserves provides significantly greater pro-

tection for marine communities than a single reserve, buffering against
environmental variability or anthropogenic threats such as oil spills. An
effective network therefore needs to span large geographic distances and
encompass a relatively substantial area in total (Lubchenco et al. 2003).
Choosing new reserves will depend on which areas are already being
protected, such that ideally new reserves will complement those already
established in order to form an optimal network. Under this approach,
a series of reserves are identified based on uncorrelated habitat types or
assemblages to provide a network of protected areas encompassing the
highest proportion of biodiversity (Howard et al. 2000, Reyers et al. 2000).
Such methods can also incorporate socio-economic concerns, assessing
the optimal location of reserve areas in order to minimize management
conflicts. In the southern Gulf of California, multiple levels of informa-
tion on biodiversity, ecological processes and socio-economic factors were
used to establish a network of reserves that would cover a large propor-
tion of habitat while minimizing the potential for social conflict (Sala et al.
2002).

CONCLUSION

The evidence that marine megafauna can be used to establish marine

hotspots is growing. Several species – including marine mammals,
seabirds, sea turtles, sharks, and other predatory fish – appear to congregate
at productive ocean areas, the protection of which would result in significant
conservation benefits (Worm et al. 2003). If complemented by reductions in
fishing effort, decreased reliance on destructive fishing methods and clear
allocations of fishing rights and responsibilities, such reserves would go a
long way towards reversing the current degradation of ocean ecosystems
(Gell & Roberts 2003). Theoretical work suggests that protection of 20% to
40% of the world’s oceans would achieve the greatest fishery benefits (NRC
2001), and provide a foundation for restoration and sustenance of marine
biodiversity (Roberts 2003).
358 S. K. Hooker

ACKNOWLEDGEMENTS

I gratefully acknowledge the Royal Society Dorothy Hodgkin Fellowship

scheme for support, and colleagues at the Sea Mammal Research Unit for
helpful ideas and discussion.

REFERENCES
Apostolaki, P., Milner-Gulland, E. J., McAlister, M. K. & Kirkwood, G. P. (2002).
Modelling the effects of establishing a marine reserve for mobile fish species.
Can. J. Fish. Aquat. Sci., 59, 405–15.
Barlow, K. E., Boyd, I. L., Croxall, J. P. et al. (2002). Are penguins and seals in
competition for Antarctic krill at South Georgia? Mar. Biol., 140, 205–13.
Baum, J. K., Myers, R. A., Kehler, D. G. et al. (2003). Collapse and conservation of
shark populations in the Northwest Atlantic. Science, 299, 389–92.
Boersma, P. D. & Parrish, J. K. (1999). Limiting abuse: marine protected areas, a
limited solution. Ecol. Econ., 31, 287–304.
Botsford, L. W., Castilla, J. C. & Peterson, C. H. (1997). The management of
fisheries and marine ecosystems. Science., 277, 509–15.
Bowen, W. D. (1997). Role of marine mammals in aquatic ecosystems. Mar. Ecol.
Prog. Ser., 158, 267–74.
Casey, J. M. & Myers, R. A. (1998). Near extinction of a large, widely distributed
fish. Science, 281, 690–2.
Cooke, S. J., Hinch, S. G., Wikelski, M. et al. (2004). Biotelemetry: a mechanistic
approach to ecology. Trends Ecol. Evol., 19, 334–43.
de Young, B., Heath, M., Werner, F. et al. (2004). Challenges of modeling ocean
basin ecosystems. Science, 304, 1463–6.
Estes, J. A., Tinker, M. T., Williams, T. M. & Doak, D. F. (1998). Killer whale
predation on sea otters linking oceanic and nearshore ecosystems. Science, 282,
473–6.
Gell, F. R. & Roberts, C. M. (2003). Benefits beyond boundaries: the fishery effects
of marine reserves. Trends Ecol. Evol., 18, 448–55.
Griffin, R. B. (1999). Sperm whale distributions and community ecology
associated with a warm-core ring off Georges Bank. Mar. Mamm. Sci., 15,
33–51.
Harwood, J. (2001). Marine mammals and their environment in the twenty-first
century. J. Mamm., 82, 630–40.
Harwood, J. & Stokes, K. (2003). Coping with uncertainty in ecological advice:
lessons from fisheries. Trends Ecol. Evol., 18, 617–22.
Hooker, S. K. & Gerber, L. R. (2004). Marine reserves as a tool for ecosystem-based
management: the potential importance of megafauna. Bioscience, 54, 27–39.
Hooker, S. K., Whitehead, H. & Gowans, S. (1999). Marine protected area design
and the spatial and temporal distribution of cetaceans in a submarine canyon.
Conserv. Biol., 13, 592–602.
(2002). Ecosystem consideration in conservation planning: energy demand of
foraging bottlenose whales (Hyperoodon ampullatus) in a marine protected area.
Biol. Conserv., 104, 51–8.
 Marine reserves and higher predators 359

Howard, P. C., Davenport, T. R. B., Kigenyi, F. W. et al. (2000). Protected area

planning in the tropics: Uganda’s national system of forest nature reserves.
Conserv. Biol., 14, 858–75.
Hyrenbach, K. D., Forney, K. A. & Dayton, P. K. (2000). Marine protected areas
and ocean basin management. Aquat. Conserv.: Mar. Freshwater Ecosyst., 10,
437–58.
Hyrenbach, K. D., Fernandez, P. & Anderson, D. J. (2002). Oceanographic habitats
of two sympatric North Pacific albatrosses during the breeding season. Mar.
Ecol. Prog. Ser., 233, 283–301.
Ingram, S. N. & Rogan, E. (2002). Identifying critical areas and habitat preferences
of bottlenose dolphins Tursiops truncatus. Mar. Ecol. Prog. Ser., 244, 247–55.
Jackson, J. B. C., Kirby, M. X., Berger, W. H. et al. (2001). Historical overfishing
and the recent collapse of coastal ecosystems. Science, 293, 629–38.
Lubchenco, J., Palumbi, S. R., Gaines, S. D. & Andelman, S. (2003). Plugging a
hole in the ocean: the emerging science of marine reserves. Ecol. Applic., 13
Suppl., S3–7.
May, R. M. (1979). Ecological interactions in the Southern Ocean. Nature, 277,
86–9.
Myers, N., Mittermeler, R. A., Mittermeler, C. G., da Fonseca, G. A. B. & Kent, J.
(2000). Biodiversity hotspots for conservation priorities. Nature, 403, 853–8.
Myers, R. A., Hutchings, J. A. & Barrowman, N. J. (1997). Why do fish stocks
collapse? The example of cod in Atlantic Canada. Ecol. Applic., 7, 91–106.
Noss, R. F. (1996). Ecosystems as conservation targets. Trends Ecol. Evol., 11, 351.
NRC (National Research Council) (2001). Marine Protected Areas: Tools for
Sustaining Ocean Ecosystems. Washington DC: National Academy Press.
O’Dor, R. K. (2004). A census of marine life. Bioscience, 54, 92–3.
Pauly, D. & Christensen, V. (1995). Primary production required to sustain global
fisheries. Nature, 374, 255–7.
Reeves, R. R. (2000). The Value of Sanctuaries, Parks, and Reserves (Protected Areas)
as Tools for Conserving Marine Mammals. Report prepared for the Marine
Mammal Commission, USA. Contract Number T74465385. Bethesda, MD:
Marine Mammal Commission.
Reyers, B., Jaarsveld, A. S. van & Kruger, M. (2000). Complementarity as a
biodiversity indicator strategy. Proc. R. Soc. Lond. B, 267, 505–13.
Roberts, C. M. (1997). Ecological advice for the global fisheries crisis. Trends Ecol.
Evol., 12, 35–8.
(2002). Deep impact: the rising toll of fishing in the deep sea. Trends Ecol. Evol.,
17, 242–5.
(2003). Our shifting perspectives on the oceans. Oryx, 37, 166–77.
Roberts, C. M. & Hawkins, J. P. (1999). Extinction risk in the sea. Trends Ecol. Evol.,
14, 241–6.
Roberts, C. M., McClean, C. J., Veron, J. E. N. et al. (2002). Marine biodiversity
hotspots and conservation priorities for tropical reefs. Science, 295, 1280–4.
Rowe, S. R. & Hutchings, J. A. (2003). Mating systems and the conservation of
commercially exploited marine fish. Trends Ecol. Evol., 18, 567–72.
Sala, E., Aburto-Oropeza, O., Paredes, G. et al. (2002). A general model for
designing networks of marine reserves. Science, 298, 1991–3.
360 S. K. Hooker

Taylor, B. L., Wade, P. R., DeMaster, D. P. & Barlow, J. (2000). Incorporating

uncertainty into management models for marine mammals. Conserv. Biol., 14,
1243–52.
Worm, B. & Duffy, J. E. (2003). Biodiversity, productivity and stability in real food
webs. Trends Ecol. Evol., 18, 628–32.
Worm, B., Lotze, H. K. & Myers, R. A. (2003). Predator diversity hotspots in the
blue ocean. Proc. Natl Acad. Sci. U. S. A., 100, 9884–8.
Yodzis, P. (1998). Local trophodynamics and the interaction of marine mammals
and fisheries in the Benguela ecosystem. J. Anim. Ecol., 67, 635–58.
Zacharias, M. A. & Roff, J. C. (2001). Use of focal species in marine conservation
and management: a review and critique. Aquat. Conserv.: Mar. Freshwater
Ecosyst., 11, 59–76.
 24
Marine management: can objectives be set for
marine top predators?
M. L. TASKER

There has been much recent discussion over the need for an ‘ecosystem
approach’ to be taken to the management of human activities both on
land and at sea. There are a number of definitions of ‘ecosystem approach’
but a common feature is the need to set objectives for management. The
ecosystem approach to management encompasses effects on the ecosys-
tem as well as social and economic effects. Top predators are a key part
of the marine ecosystem and thus objectives ought to be set for them.
Objectives that might be set are broadly of two types: ‘what do we want
to achieve’ (targets); and ‘what do we not want to happen’ (limits). Objec-
tives may describe several aspects of the system for which they are being
set – commonly objectives are set for the state of an animal population
or for the impact on that population. If objectives are to be useful they
need to be achievable, inter-compatible and responsive to management
actions. A major initiative in European waters at present is the establish-
ment of Ecological Quality Objectives (EcoQOs) in the North Sea, but other
implicit objectives have also been set for marine top predators. In relation
to top predators, societal wishes are usually related to the state of popula-
tions which integrates across a set of measures including population size,
growth, range, health and feeding relationships. It is often not possible to
measure these aspects of state easily or sufficiently widely to understand
what is happening to a population as a whole. In these cases, it may be
easier to use a proxy measure relating to the human pressure on the top-
predator population. This is possible if there is sufficient understanding
of the link between the proxy measure and state of the population. Use of

Top Predators in Marine Ecosystems, eds. I. L. Boyd, S. Wanless and C. J. Camphuysen.

Published by Cambridge University Press.
C Cambridge University Press 2006.
362 M. L. Tasker

such proxy measures may simplify monitoring systems and avoid the risk
of managers attempting to achieve many conflicting objectives.
As with nearly all parts of the planet, human activities have become an
important factor in ecological change in European seas. As a consequence,
there are concerns that such changes will either directly or indirectly harm
future generations. Several frameworks are being established, or have been
established, to address these concerns. Many of these are based on the prin-
ciples of sustainability – that actions today should not reduce the options of
future generations. Sustainability is often viewed as being built on three
pillars: economic, social and environmental (with some adding a fourth
administrative pillar). Sustainable development involves ensuring a bal-
ance between these pillars, but noting that the environmental pillar is
probably the one with least room for compromise. We cannot be sure
that many of the adverse environmental changes caused by humans are
reversible.
Humans cannot, of course, manage marine ecosystems; we can only
manage our activities in those systems. One commonly advocated way
of addressing the issue of sustainability in marine ecosystems has been
the ecosystem approach to management (sometimes described as the
ecosystem-based approach). For any management approach or style to suc-
ceed, one of the first prerequisites is a set of objectives.
In the case of the ecosystem approach to management, these objectives
need to relate to individual parts of the ecosystem, or to processes or fea-
tures of the system as a whole. Objectives might relate also to several fea-
tures of individual parts of the ecosystem. If objectives are set, then ways of
measuring whether or not they are being achieved need also to be devised.
One classification of these parts, processes or features of the environment
is the DPSIR framework. In this framework, social and economic develop-
ments or Driving forces exert Pressures on the environment resulting in
changes to its State. This leads to Impacts on environmental quality, which
may elicit a societal or policy Response. One objective of this chapter is to
illustrate the application of the DPSIR framework to objective setting using
top predators as an example. The measurement of whether objectives are
being met requires a series of metrics. Objectives can also be aimed either
to avoid certain conditions (limits) or to achieve certain conditions (targets).
It is possible to derive limit objectives scientifically given a set of rules to
work to (such as: we want to ensure that a species does not become extinct),
but such objectivity is more difficult for target objectives. Rules may derive
from some sort of societal choice expressed in law or agreement, or from
elsewhere.
 Marine-management objectives for top predators 363

Table 24.1. Examples of possible factors for which objectives might be set
for populations of seabirds and marine mammals, related to the DPSIR
framework

Marine mammals Seabirds

Driving force Lack of oil reception facilities.

Pressure Amount of ‘dangerous’ Amount of ‘dangerous’ fishing,
fishing, chemicals in the amount of oil pollution, chemicals
environment, amount of in the environment, plastics in the
noise, food availability environment, food availability,
introduced mammalian predators
State Population size, population Population size, population
distribution, pup distribution, breeding success,
production, levels of levels of chemicals in eggs
chemicals in body tissue
Impact Proportion of population Proportion of population killed by
killed by bycatch, reduction bycatch, loss of breeding areas due
in breeding ability, amount to mammalian predators
of habitat lost to noise
Response Implementation of Implementation of management
management plan plan

OBJECTIVES FOR TOP PREDATORS

Top predators are a key part of any ecosystem, and thus if we want to manage
human effects on the ecosystem, it is logical that some objectives should be
set for these top predators. Table 24.1 is an example of the application of the
DPSIR framework to two groups of top predators.
If objectives and their associated metrics are to be established, then it
is important that the metrics are reliable. In suggesting metrics for some
features of the North Sea ecosystem (see below), ICES (2001) applied a set of
criteria that ideally ought to be met (Table 24.2). It is difficult for all criteria
to be met by all metrics and there are patterns that match within the DPSIR
framework. Thus for example, criteria (b), (c) and (e) in Table 24.2 all relate
to human activities that in general fall into the category of pressure. It is
therefore much more likely that metrics associated with a pressure will meet
these three criteria than metrics associated with state.

ECOLOGICAL QUALITY OBJECTIVES (EcoQOs)

Following a Ministerial Conference held in Bergen in 1997, the Interna-

tional Council for the Exploration of the Sea (ICES) was asked to consider
364 M. L. Tasker

Table 24.2. ICES criteria for good ecological quality (EcoQ) metrics
(ICES 2001)

(a) Relatively easy to understand by non-scientists and those who will decide on
their use.
(b) Sensitive to a manageable human activity.
(c) Relatively tightly linked in time to that activity.
(d) Easily and accurately measured, with a low error rate.
(e) Responsive primarily to a human activity, with low responsiveness to other
causes of change.
(f) Measurable over a large proportion of the area to which the EcoQ metric is to
apply.
(g) Based on an existing body or time series of data to allow a realistic setting of
objectives.

the establishment of EcoQOs for ten ecosystem features (IMM, 1997). In

carrying out this task, ICES started by looking to see what monitoring was
being carried out in the North Sea that could provide a metric against which
an objective could be set. It evaluated existing monitoring programmes
against its criteria (see above). Thus if there was an existing monitoring pro-
gramme for a feature with accepted standards and protocols, which was pro-
ducing results with a relatively low error margin, then these features were
suggested as suitable metrics for setting objectives. Few metrics actually
met all criteria, but evaluation against these criteria at least indicated where
the metric was weak, and allowed targeting of possible improvements. This
approach by ICES had the benefit of allowing existing experience to be taken
into account, but obviously could only cover parts of the environment that
were already being monitored.
At the Fifth Ministerial Conference on the Protection of the North Sea
in 2002, Ministers examined ICES advice and that of others and decided to
establish a pilot project to implement EcoQOs. Ten EcoQOs were decided
upon initially; these included five relating to the nutrient status of the North
Sea and three that included top predators. In addition, seven further eco-
logical qualities were considered suitable for development towards having
objectives set for them in the future (Table 24.3).
Most objectives or metrics for further development chosen for top preda-
tors were either pressure or state indicators. Only one of the chosen objec-
tives met all of the ICES criteria (that relating to oiled seabirds), while one
possible metric that met all of the ICES criteria (organochlorines in seabird
eggs) did not have an objective set for it.
Table 24.3. EcoQOs, areas where objectives are being developed relating to top predators in the North Sea and categorization
of the ecosystem features being considered in DPSIR framework

Ecosystem feature Objective set DPSIR Limit/target Meet ICES criteria?

Seal population trends No decline in population size or pup production of State Target 4/7
≥10% over a period of up to 10 years.
Bycatch of harbour porpoises Annual bycatch levels should be reduced to levels Impact Limit 6/7
below 1.7% of the best population estimate.
Oil pollution (measured in seabirds) Proportion of oiled common guillemots among Pressure Target 7/7
those found dead or dying on beaches should be
10% or less of the total found dead or dying, in
all areas of the North Sea.
Presence and extent of threatened and Develop further State 2/7
declining species
Utilization of seal breeding sites Develop further State 4/7
Mercury concentrations (in seabird Develop further Pressure 5/7
eggs and feathers)
Organochlorine concentrations (in Develop further Pressure 7/7
seabird eggs)
Plastic particles (in stomachs of Develop further Pressure 4/7
seabirds)
Local sandeel availability to Develop further Pressure 4/7
black-legged kittiwakes
Seabird population trends as an index Develop further State 3/7
of seabird community health
366 M. L. Tasker

FURTHER OBJECTIVES IN EUROPEAN SEAS

The EcoQOs are the only explicitly set international objectives for top preda-
tors in European seas at present. EcoQOs were also established for other
environmental features in the North Sea. However, there are some implicit
objectives for top predators within other legislation. For instance the EU’s
Habitats Directive (92/43/EEC), obliges EU Member States to:
r take all appropriate steps to avoid, in special areas of conservation, the
deterioration of natural habitats ‘and the habitats of species and the
disturbance of species for which the area was designated’.
r take measures designed to maintain or restore, at favourable
conservation status, natural habitats and species of wild fauna and
flora of Community interest.

Grey and harbour seals, harbour porpoise and bottlenose dolphin are
included in the species listed under this Directive as requiring these objec-
tives to be met. The term ‘favourable conservation status’ has caused con-
siderable debate about its meaning, but nevertheless it is evident that some
objectives are required.
Similarly in the EU’s Birds Directive (79/409/EEC), Articles require EU
Member States to:
r take the requisite measures to maintain the population of the
species . . . at a level which corresponds in particular to ecological,
scientific and cultural requirements, while taking account of economic
and recreational requirements, or to adapt the population of these
species to that level.
r ensure their survival and reproduction in their area of distribution.

This Directive covers virtually all seabirds occurring in European waters.

DISCUSSION

It is plainly possible to set goals for top predators in order to help implement
an ecosystem approach to management in European waters, but their use-
fulness and feasibility are not straightforward. There are only three ‘opera-
tional’ objectives set so far as EcoQOs, and although some implementation
has occurred (most of which has involved adopting pre-existing monitor-
ing programmes), rather little in the way of new management measures
to help meet these objectives has been undertaken. An exception to this
has been the recent EC Regulation concerning cetacean bycatches, although
 Marine-management objectives for top predators 367

implementation of aspects of this is delayed by a few years. It also remains to

be seen whether Member States will undertake all that is required under the
Regulation, especially as they have had very similar responsibilities under
the Habitats Directive since 1992 and in most cases have not met them.
The set of EcoQOs is obviously far from complete if objectives for top
predators (or any other ecosystem component) are to be fully addressed.
The Oslo and Paris Commission (OSPAR) will be considering this further
in the future, but if the set of EcoQOs is to be completed, further prob-
lems may arise. It is far from certain that it is possible to establish a reliable
metric for many of the features that are not being monitored at present.
Unreliable metrics or lack of precision in understanding whether or not an
objective is being met may lead to inappropriate management measures
being taken. The greater the number of objectives, the more likely it is that
conflicts between them might arise. This is much more likely between ‘tar-
get’ objectives than between ‘limit’ objectives. Larger numbers of objectives
would need more monitoring resources and there is some evidence that
already resources are proving to be restricting the implementation of the
current three EcoQOs.
An illustration of the difficulty of using objectives in the marine envi-
ronment occurs already in fisheries management in the North Sea. Both the
biomass and fishing mortality of most commercially important fish stocks
are estimated on an annual basis using a wide variety of relatively costly
monitoring and assessment systems. Limit objectives have been agreed
both scientifically and politically for the biomasses (Blim ) and fishing mor-
talities (Flim ) of these stocks. Stocks outside these limits are likely to have
impaired recovery potential. The inability to be precise in measuring and
assessing has led to the introduction of precautionary limit points for these
two metrics (Bpa and Fpa ) in order to avoid inadvertently exceeding the lim-
its. Despite this, a substantial proportion of stocks are outside the precau-
tionary limits, with some exceeding their biological limits. Politicians have
found it difficult to take decisions that might affect the short-term liveli-
hood of fishermen and to control fishing activity. In the North Sea, mixed
fisheries mean that some stocks in very bad condition (e.g. cod) are caught
in mixed fisheries alongside fish from stocks in apparently better condition
(e.g. haddock). In these circumstances, objectives for maximising return
from the haddock fishery conflict with controlling fishing mortality on cod.
One possibility for reducing the risk of too many objectives is to
combine individual metrics into fewer generic indicators (see Croxall
(Chapter 11 in this volume)). However, where this has happened on land, it
has usually resulted from examining trends in the individual metrics and
368 M. L. Tasker

demonstrating that their changes are correlated. Boyd and Murray (2001)
have undertaken this using information from three top predators that feed
on krill, monitored for 22 years at South Georgia. Separate analysis of vari-
ables showed that there was a significant decline in population sizes over
this period, but there was no trend in a combined index that represented for-
aging conditions during the local breeding season. Combined indices may
not always be helpful in providing management advice. There is also a risk
in combining that, unless the correlation is very close, important informa-
tion from the monitoring could be lost. De la Mare and Constable (2000)
examined this issue theoretically for Antarctic ecosystem monitoring and
found that high coefficients of correlation were also needed if there were
large numbers of missing values in the data.
There are some potential advantages in using metrics relating to pres-
sures. The information from a pressure metric is likely to be relatively well
linked to a manageable human activity. It is also often easier to measure
pressures than it is to measure state or impact. An individual pressure will
often affect the state of a number of ecosystem features. However, public
concern usually relates to the state of an ecosystem feature, so that if a pres-
sure metric is to be used for management purposes relating to that feature,
it is necessary to understand the links between the amount of pressure and
the state of the feature.
For instance, there is wide concern about the depleted state of many
fish stocks, the number of by-caught animals in fisheries and the damage
inflicted to seabeds by fishing. All of these concerns are caused by fishing
pressure. It might be easier to monitor fishing pressure rather than to mon-
itor the state of the various ecosystem features of concern. However, this
would need a good understanding of the link between these areas of con-
cern and fishing pressure, which in the case of this example still require a
substantial research effort.
It is also worth noting that trends are often easier to monitor than are
absolute values. It may be easier to set an objective that maintains a trend
and avoids reversal of the trend.
The setting of objectives and the use of the associated metrics, as noted
above, is a societal activity. It is possible to derive limit objectives scientifi-
cally given a set of rules (such as: we do not want population abundance to
be driven down by human activity, or for a species to become extinct), but
even these rules are ultimately a societal choice. At present, the mechanics
of objective setting and devising metrics for them has largely taken place
among a relatively small group of scientists and policy-making administra-
tors and it has been influenced by non-governmental organizations with
 Marine-management objectives for top predators 369

objectives in nature conservation. There is a need to ensure that any objec-

tives set are in accordance with wider societal wishes and needs.
One aspect implicit in the societal choice of objectives is the consider-
ation of management actions to meet the objective, or to be taken should
a limit be exceeded. In ideal circumstances, a set of rules might be estab-
lished in advance of the limits being broken. Examples include harvest con-
trol rules that are used in some fisheries management systems. However,
in most cases, further research and analysis will be needed before manage-
ment action is taken. This would be especially important in cases where a
change in the ecosystem could have either natural or anthropogenic causes,
or if management objectives conflict.

ACKNOWLEDGEMENTS

The work described in this chapter has benefited from the input and discus-
sions over several years with many colleagues in JNCC, ICES and OSPAR. I
thank them all, but in particular Jake Rice and members of the ICES Work-
ing Group on ecosystem effects of fishing activity. It is a personal view-
point and should not be taken as representative of the views of any of them
or of any organization. Ian Boyd and an anonymous referee considerably
improved an earlier draft of the paper.

REFERENCES
Boyd, I. L. & Murray, A. W. A. (2001). Monitoring a marine ecosystem using
responses of upper trophic level predators. J. Anim. Ecol., 70, 747–60.
de la Mare, W. K. & Constable, A. J. (2000). Utilising data from ecosystem
monitoring for managing fisheries: development of statistical summaries of
indices arising from the CCAMLR Ecosystem Monitoring Program. CCAMLR
Sci., 7, 101–17.
ICES (International Council for the Exploration of the Sea) (2001). Report of the
ICES Advisory Committee on Ecosystems, 2001. ICES Cooperative Research
Report 249. Copenhagen, Denmark: ICES.
IMM (1997). Statement of Conclusions: Intermediate Ministerial Meeting on the
Integration of Fisheries and Environmental Issues, 13–14 March 1997, Bergen,
Norway. Oslo, Norway: Ministry of Environment.
 Index

Page numbers in italic denote figures. Page numbers in bold denote tables.

aerial-pursuit 94 modelling considerations 319–20

Aethia pusilla see auklet, least predation 316–17, 318–19
Alaska, Steller sea lion population decline recruitment 315
275–89 bass, kelp, prey abundance 216
albatross, black-browed, Bird Island Bass Rock, gannet colony 240, 242, 243, 245,
Monitoring Programme 159, 160, 238–47
162, 164, 170–1, 171 Beaufort Sea, polar bears 99, 103–13
Alosa pseudoharengus see gaspereau Bering Sea
Ammodytes dubius see sand lance Eastern, food web 338, 342
Ammodytes marinus see sandeel effect of fisheries 17–18
anchovy, Peruvian 6, 13, 13, 124–5 Berwick Bank, sandeel fishery 227
Antarctic, ecosystem 17 biomass limit
Antarctic Circumpolar Current 28–9, 36 fish stock 211–13
Antarctic Circumpolar Wave 29, 36 reference point 213–15, 218–19
Anthus pratensis see pipit, meadow Bird Island
Arctic, polar bear 99 Antarctic fur seals, foraging behaviour
as indicator of ecosystem dynamics 99, 135, 136, 140
113–15 krill population sampling via predator
Arctocephalus gazella see seal, fur, Antarctic diet 250–4, 258–9
Atlantic Ocean, macaroni penguin, foraging behaviour
North 133–4, 135, 136–40
fish density 13 Bird Island Monitoring Programme, South
North Atlantic Oscillation 149–51 Georgia 37–41, 158, 171, 161–73
trophic levels 15 data 167–9
northwest, food-web shift 121, 124, 125 predator performance 162, 163, 169
southwest variables 159, 160
physical anomalies 29–31 Black Sea, effect of fisheries 18
SST variability 30, 31, 35–6, 41 bloom, spring 48, 49–50, 54–7
auk, in MSFA 89–93 as indicator 49–50
auklet, least, prey abundance 216 blubber, seal, sampling 100–1
bottom-up forcing 122–4, 143, 147, 150,
Balaenoptera acutorostrata see whale, minke 154
Bank region, Firth of Forth 48, 50–2, 55 bycatch 17, 207, 224
Barents Sea
scenario modelling 313, 310–21 Callorhinus ursinus see seal, fur, Northern
management 310–13, 317–19, 321 capelin 6, 73–4, 124–5
 Index 371

and cod performance 225 mid-water, shallow sea fronts 184,

and guillemot performance 216–17 182–4
Barents Sea scenario modelling 313–14, sandeel availability 192–3
318 diving
predation 316–17 and prey dispersal 136–40
recruitment 315 in foraging model 303–8
northwest Atlantic cold-water event 124, dolphin
125 bottlenose 366
catalysts 89, 90, 91 white-beaked 83–7
CCAMLR 8, 158, 262–3, 327, 330–2, 333–41 DPSIR framework 362, 363, 365
Ecosystem Monitoring Programme see
CEMP Ecological Quality Objectives (EcoQOs)
CEMP 160, 158–61, 171, 262–3, 335 364, 363–4, 365, 366–9
cetaceans ecosystems
in MSFA 91, 93 integrity 262–3
North Sea 93, 366–7 management 7–9, 262–3, 324–8, 343,
see also dolphin; porpoise; whale 348
Cetorhinus maximus see shark, basking direct effects on populations 330–3
Champsocephalus gunneri see icefish European seas 362–9
Circus cyaneus see harrier, hen higher trophic levels 325–6, 330
climate change indirect effects on populations
oceans 6 333–41
Eynhallow fulmar colony 150, 149–51 objectives 329, 324–43, 362: ecological
Clupea pallasi see herring, Pacific quality 363–4, 366–9; societal
cod, Atlantic 13, 17, 21, 69, 74 choice 368–9; top predators 363,
Barents Sea scenario modelling 313–14, 363, 366–9
318 stable states 342–3
predation 316–17 sustainability 362
recruitment 315 modelling difficulties 349
North Atlantic, productivity decline monitoring 158–61, 333–41, 343
225 Bird Island 161–73
North Sea, interaction with sandeel 229, see also CEMP
230, 227–30, 231, 232 El Niño 4 region 33–4, 35–6
cod, Pacific 277–8 El Niño Southern Oscillation 41, 172
composite standardized index 335–6, 337 energy budgets
conservation, marine 347–8, 356–7 North Sea birds 197, 198, 199, 193–9,
Convention on the Conservation of 200, 202, 203, 204, 205, 206, 208
Antarctic Marine Living Resources methods 194–6
see CCAMLR energy investment, grey seals 71
copepod, calanoid 179 Engraulis ringens see anchovy, Peruvian
Cystophora cristata see seal, hooded Erignathus barbatus see seal, bearded
EU Birds Directive 366
Davis Strait, polar bears 103–13 EU Habitats Directive 366–7
depth, critical, phytoplankton 47, 48, 55 Eumetopias jubatus see sea lion, Steller
Diomedea melanophris see albatross, Euphausia see krill
black-browed
dipping 90, 94 Farne Islands 84, 87–8
Dissostichus eleginoides see toothfish, Faroe Islands, fulmar colony 145, 147–8
Patagonian fat, adipose
divers 119, 121–2, 123 polar bears 100
benthic, low-tide foraging 185, 184–6 sample collection 100
372 Index

fatty acid signatures 106 flatfish 13

polar bears flounder 69, 74, 73–4
comparison with prey 102, 101–3, 104 food chain 2
quantitative analysis 100, 103–9, 110, length 18–20
111, 112, 113 food supply
quantitative analysis 106–7 and foraging trip duration 237–47
feeding, social 91 changes 5
feeding associations, multispecies see food web 15, 14–16
foraging assemblies, multispecies competition 19, 18–19
Firth of Forth effect of fisheries 12, 17–19
oceanography 50–2 fishing down 14
sandeel fishery 54, 55–7, 192, 218, 223–6, northwest Atlantic 121, 121, 124, 125
228, 233 regime shifts 122–4
seabird breeding study 47, 50–2 research 126–7
spring bloom 55 Southern Ocean 131–41
fish forage species 124–5
body size 16, 16 foraging
forage species 123, 124–5 Antarctic fur seals 132–4, 135–40
minimum stock size threshold 211–13 critical habitat protection 353
North Atlantic biomass 13, 14 grey seals 65–9, 76–9
piscivorous, impact of sandeel fisheries macaroni penguins 133–4, 136–40
224–6, 229, 230, 231, 233 model 304, 305, 294–308
spawning stock biomass 211–13 assumptions 295–7
trophic levels 15, 14–16 food particles 302, 298–302
fisheries guillemot 307, 303–8
1900s expansion 12–14 patchy environment 298, 297–8,
effect on ecosystems 11–23 299
comparison with top predators 20–2 polar bears 100
direct 12–17 seabirds 119–20
evolutionary change 16, 22 gannets, trip duration as indicator of
indirect 17–20 food supply 237–47
monitoring 343 North Sea 87–8: energy budgets
mortality of Steller sea lion 279–80 193–6, 197, 198, 199, 200; impact of
waste 148–9, 152–4, 207 oceanography 180–8
global catch 13, 12–14 foraging assemblies, multispecies, North
impact of marine reserves 356–7 Sea 89–91, 93–5, 187
management 211–13 forcing
direct effects on populations 330–3 bottom-up 122–4, 143, 147, 150,
goals 324–6, 327–8, 329, 343, 367 154
indirect effects on populations external 123–4
333–41 top-down 143, 147
scenario modelling, Barents Sea Fratercula arctica see puffin, Atlantic
310–13, 318–9, 321 fulmar, North Atlantic 6, 83–7
sandeel diet 148–9
Firth of Forth 54, 55–7, 192, 218, foraging 148–9
223–6, 228, 233 importance of fishery waste 147–9,
Shetland Islands 224 152–4
sprat, Moray Firth 224 in MSFA 90, 93–5
fishing population change 146, 144–6
down food web 14 expansion 145, 147–8
ghost 12, 17 Eynhallow colony 150, 149–51
see also fisheries hunting 147–8, 153–4
 Index 373

paleoecology 151–2 harvest see fisheries; fishing

use of proxies 151–2 herring, Atlantic
Fulmarus glacialis see fulmar, North Barents Sea scenario modelling 313–14,
Atlantic 318
predation 316–17
gadoid see fish, piscivorous recruitment 315
Gadus macrocephalus see cod, Pacific herring, Pacific 277–8
Gadus morhua see cod, Atlantic horizon of relevance 2
gannet, Northern 83–7 hotspots
chick care 241, 246 diversity 354, 355–7
dietary analysis 244, 245–6 prey 125, 126
energy budget 197, 198, 199, 193–9, 200, Hudson Bay, polar bears 99, 103–13
202, 203, 204, 205, 206, 208 human activity
foraging effect on higher trophic levels 5, 329,
behaviour: Bass Rock colony 245, 324–43, 347–8
238–47; Great Saltee colony 238–44, see also fisheries
246–47 hunting, effect on fulmar population 153–4
range 240, 238–44: and colony size Hydrobates pelagicus see storm-petrel
237–8, 242, 243, 246–7; trip duration hydrography
243, 238–44, 246–7 effect on prey 122
in MSFA 90, 89–91, 93–5 effect on seabirds 180–8
northeast Atlantic, as monitor of fish Hydrurga leptonyx see seal, leopard
stock 237–47
northwest Atlantic, food-web shift 124, ice breakup, effect on polar bear diet 111
125 icefish 17, 37
satellite telemetry 239 Iceland, fulmar colony 145, 147–8, 152, 153
travel speed 238–41 ICES, ecological quality objectives 364,
gaspereau 69 363–4
ghost fishing 12, 17 Illex illecebrosus see squid
Glyptocephalus cynoglossus see flounder inference, inverse, predator performance
Great Saltee, gannet colony 240, 242, 243, and prey availability 264, 267–8,
238–44 272
grouse, red, MSFR 269–71 initiators 89, 91
guillemot, common 47, 83–8 instrumentation, animal-borne 178
breeding success 56, 57, 58 International Whaling Commission,
capelin requirement 216–17 Revised Management Procedure
energy budget 197, 198, 199, 193–9, 200, 312, 319
202, 203, 204, 205, 206, 208 Isle of May, seabird breeding 50, 56, 58, 59,
foraging model 307, 303–8 84, 87–8, 180, 183
in MSFA 89–91, 92, 93–5
sandeel availability 54, 192–3 James Bay, polar bears 99
thermocline use 184, 181–4, 186
gull, Larus, in MSFA 83–7, 90, 89–91, 93, kittiwake
94 black-legged 8, 47, 83–7
breeding success 56, 55–8, 192–3
habitat energy budget 197, 198, 199, 193–9,
critical, protection 350–2 200, 202, 203, 204, 205, 206, 208
thermal 125 foraging at shallow sea fronts 180–1,
haddock, North Sea, interaction with 182, 183, 186
sandeel 229, 230, 227–30, 231, 232 in MSFA 90, 89–91, 92, 93–5
Halichoerus grypus see seal, grey sandeel requirement 54, 215–16, 217,
harrier, hen, MSFR 269–71 218
374 Index

kleptoparasites 89, 90, 94 primary production patterns, Firth of

krill 6, 124–5 Forth 48, 50–1, 52, 53
abundance variability QFASA, polar bear diet 103–13
sea-surface temperature 258, 257–8 scenario
South Georgia 29–31, 36, 131–41, Barents Sea 310–21: modelling
161–3, 164, 165, 166, 166, 168, 169, considerations 319–21
249–58, 259, 263, 272, 331–2, 336, whole-system 1, 4
337, 367–8 Moray Firth 84, 87–8
biomass limit 213, 331–2 Morus bassanus see gannet, Northern
demography, South Georgia 254–7 multispecies assessment 121, 121–2
harvesting 18, 158, 166 multispecies foraging assemblies see
importance in food web 333 foraging assemblies, multispecies
migration in water column 135–6
population dynamics 252–4, 255 NOAA sea-ice dataset 32–3
sampling 250–1, 252, 258–9 North Sea
predator requirements 215 foraging habitat 88–9
size composition, Antarctic fur seal diet marine mammals 93
36, 250–1, 253, 254, 254, 258 marine-management objectives
362–9
lactation, grey seals 65, 71, 73, 78 oceanography 178–9
Lagenorhynchus albirostris see dolphin, seabird breeding 46–59
white-beaked seabird foraging 87–8, 192–3
Lagopus lagopus scoticus see grouse, red seasonal oceanographical cycle 47–9
Larus spp. see gull, Larus shallow sea front 88–9, 178–9, 180–1,
182–4
mackerel 13 foraging assemblies 89–93
mackerel, Atka 277–8 stratification 88–9
Mallotus villosus see capelin top predators 84, 86, 83–7, 362–9
mammals, marine
critical habitat 353 oceanography
effect on ecosystems 20–2 influence on prey 47–9, 122, 178–9,
North Sea 93, 94 180–8
recording at sea 85 animal-borne instrumentation 178
man, effect on ecosystems see fisheries; at-sea surveys 178
human activities North Sea 47–9, 178–9
marine protected areas 350 regime shifts 123, 122–4, 125
see also reserves, marine oceans
Marr Bank climate change 6
MSFA 84, 92 physics, impact on seabirds 180–8
piscivorous fish study 226–33 Odobenus rosmarus see walrus
sandeel fishery 224 Orcinus orca see whale, killer
mass oscillations see Atlantic Ocean, North
maternal postpartum, grey seal 65, 71, 73, Atlantic Oscillation; El Niño
78 Southern Oscillation
weaning, grey seal 65, 71, 73, 78 over-fishing 22, 211–13
metrics 364, 366–9
pressure 368 Pacific Ocean, sea-surface temperature 33,
Microtus agrestis see vole, field 35–6
modelling paleoecology
ecosystem 1–2, 7, 348, 349 fulmars 151–2
indirect effects of fishing 17–18 use of proxies 151–2
partial-system 2 Paralabrax clathratus see bass, kelp
 Index 375

penguin 18 inverse inference 264, 267–8, 272

gentoo North Sea 47, 226–33
Bird Island Monitoring Programme predator response 120, 163, 168,
159, 160, 162, 163, 164, 171, 196–208, 264–73
170–1: effect of SST 36, 37–41 seabirds 119–20
macaroni South Georgia 161–73
Bird Island Monitoring Programme uncertainty 266, 267–8
159, 162, 163, 164, 166, 171, 170–1 predators
energy demands 134, 133–4 apex see predators, top
foraging behaviour 133–4, 135, diversity hotspots 354, 355–7
136–40: diving behaviour 137, generalist, multispecies functional
136–40 response 264–5, 269–71, 273
krill dependence 134, 131–41 krill dependent 29–31, 36, 131–41,
petrel see storm-petrel 249–59, 272, 367–8
Phalacrocorax aristotelis see shag, European performance variables 162, 164, 161–6,
phalarope 83–7 336, 337
Phoca groenlandica see seal, harp marine mammals 20–2, 353
Phoca hispida see seal, ringed parameter indices 335–6
Phoca vitulina see seal, harbour thermal habitat 125
Phocoena phocoena see porpoise, harbour top
physics, ocean, impact on seabirds 180–8 as indicator of ecosystem dynamics
phytoplankton, critical depth 47, 48, 55 3–9, 64–9, 74, 99, 113–15, 126–7,
pinniped see sea lion; seal 161, 353
pipit, meadow 269–71 critical habitat 350–2, 353
piscivores 119, 121–2 DPSIR framework 363, 363, 365
see also fish, piscivorous human exploitation, management
planktivores 119, 121–2 325–6, 330, 333–41
Pleurogrammus monopterygius see mackerel, management objectives 362–9
Atka North Sea 84, 86, 83–7, 95
plunge-diving 90, 94 population change 143–4, 171, 170–1
polar bear prey density requirements 215
Canadian Arctic 99 species-specific sensitivity 218
adipose fat sampling 100, 101 terrestrial breeding season 3
diet 99–115: effect of ice breakup 111; pressure metric 368
as indicator of ecosystem dynamics prey
99, 113–5; variation with latitude 113 abundance, predator requirements 215
fatty acid signature 102, 101–3, 104: availability, gannet monitoring 237–47
QFASA 100, 103–9, 110, 111, 112, availability/abundance, influence of
113 oceanography 122
Pollachius virens see pollock calorific energy 264
pollock 17–18, 69, 73–4, 124–5 density monitoring 266–72
pollock, walleye 277–8, 279, 286–9 distribution 131–41
fishery management 212 hotspots 125, 126, 355–7
population change thermal habitat 125
decline of Steller sea lion 275–89 see also predator–prey interaction
top predators 143–4 producers 90, 91
Bird Island Monitoring Programme production
171, 170–1 primary 48, 49, 50, 54, 178–9
northern fulmar 144–6, 147–8 at thermocline 182–4
porpoise, harbour 83–7, 93–5, 366 model 51
predator–prey interaction 1, 21–2 productivity, predator, operational
cod 229, 230, 231, 226–33 objectives 338, 339–41
376 Index

proxies scooping 90, 94

ecosystem integrity 262–3 Scotia Sea
in palaeoecology 151–2 krill population 257
puffin sea-ice variability 30, 32–3, 35
Atlantic 83–7, 93–5 Scotian Shelf, Canada
prey abundance 216 environmental variability 71–4, 76
pursuit-diving 90, 94, 121–2 fish size 16
Pygoscelis papua see penguin, gentoo grey seals 64–5
scroungers 89, 90, 91
radiation, solar, and spring bloom 55 sea lion, Steller 212
Raja ocellata see skate, winter diet 277–9
Raja radiata see skate, thorny management plan 288–9
razorbill, in MSFA 89–91, 93–5 population decline 276, 275–89
recruitment, Barents Sea scenario competing hypotheses 278–80, 283,
modelling 315 286, 287, 288, 286–9: fishery-related
redfish 69, 74, 73–4, 77–8 mortality 279–80; food limitation
regime shifts see oceanography, regime 278–9; ‘junk food’ 279; predation
shifts mortality 280
reserves, marine 350–7 model 280, 281
benefits 351–2 sea otter 20, 21
effect on fisheries 356–7 sea-ice
high-mobility species 350–2 Southern Ocean 29
mammals 353 effect on krill 257–8
network 357 Scotia Sea 30, 32–3, 35
response, predator 120, 163, 168 seabird–prey interaction 119–20
decreased sandeel availability 196–208 multispecies assessment 121, 121–2
food supply change 5, 168 seabirds
multispecies functional 264–5, 269–71, effect of regime shifts 124
273 fish consumption 213, 214
Rissa tridactyla see kittiwake, black-legged limit reference point 213–15, 218–19
foraging 119–20
Sable Island, Nova Scotia, grey seal energy budgets 197, 198, 199,
breeding colony 64–5, 66, 66, 76 193–9, 200, 202, 203, 204, 205, 206,
St Kilda, fulmar colony 145, 147–8, 152, 153 208
salmon 13 model 303–8
sand lance 69, 74, 73–4, 77–8 impact of ocean physics 180–8
sandeel 6, 47–55, 124–5 animal-borne instrumentation 178
and seabird breeding success 54, 56, at-sea surveys 178
57–8, 192–3, 215, 217–19 trophic links 186–7
at shallow sea fronts 180–2 North Sea
at thermocline 182–4 breeding success 46–52, 59: sandeel
biomass, North Sea 213–15 availability 54, 217–19; sandeel
decline, predator response 196–208 fishery 55–7
fishery, Firth of Forth 54, 55–7, 192, 218, foraging 87–8
223–6, 228, 233 response to sandeel decrease 196–202,
life history 53, 179 203, 204, 205, 206, 208
North Sea predators 83, 93, 192–3, northwest Atlantic, food-web shift 124
196–208, 213–14, 215 recording at sea 85
sardine 6, 124–5 seal
Sardinops sagax see sardine bearded
schooling 21 blubber sampling 100–1
 Index 377

fatty acid signature 101–13 sealing, Barents Sea 318–19

in polar bear diet 113–15 Sebastes spp. see redfish
elephant, northern, breeding ground sensitivity, species-specific, top predators
353 218
fur, Antarctic shag, European
Bird Island Monitoring Programme, energy budget 197, 198, 199, 193–9, 200,
159, 162, 168, 171, 170–71 202, 203, 204, 205, 206, 208
diving 76–7, 137, 135–40 in MSFA 94
effect of SST 36, 37–41 tidal diving behaviour 182, 184–6
energy demands 134, 133–4 shallow sea front 88–9, 178–9, 180, 186
foraging behaviour 132–4, 135, 135–40 Firth of Forth 48, 50–2, 55
krill dependence 134, 131–41 foraging assemblies 89–93, 95
krill size in diet 36, 253, 254, 250–4, mid-water divers 182–4, 186
258: sampling 251 surface-feeders 180–1, 182, 186
Northern, population decline 278 shark 13
grey basking, prey abundance 216
eastern Canada 64–79: breeding shearwater, Manx, in MSFA 83–7, 90
colony 65 skate
demography 78: diet 65, 69, 70, 74, thorny 69, 74
77–9; diving behaviour 65, 67, winter 69, 74
67–9, 76–9; energy investment 71, skua
78; foraging 65–9, 68, 76–9; as Arctic, sandeel requirement 215, 216, 217
indicator of ecosystem state 64–9, in MSFA 90, 89–93, 94
74; lactation 65, 71, 73, 78; life South Georgia
history 65, 71; maternal postpartum Atlantic physical anomalies 29–31
mass 65, 71, 73, 78; prey biomass SST variability 30, 31, 32–4, 35–6, 41:
73–4, 75; pup weaning mass 65, 71, effect on predators 36–41
73, 78 teleconnections 33–4, 41
North Sea 83–7, 366 Bird Island Monitoring Programme
harbour 20, 83–7, 366 37–41, 158, 161–73
blubber sampling 100–1 predator performance 163
fatty acid signature 103 variables 159
in polar bear diet 113–5 krill-dependent predators 131–41,
population decline 278, 287, 249–59, 272
289 krill population sampling 250–4
harp Southern Ocean
Barents Sea scenario modelling food web 131–41
310–21: predation 317; recruitment physical anomalies 28–9
315; sealing 318–19 sea-ice 30, 32–3, 35
blubber sampling 100–1 teleconnections 33–4, 41
fatty acid signature 101–13 variability of krill population 249–59
in polar bear diet 114–15 sprat 54, 124–5, 179, 182–4
hooded squid 41, 69, 74
blubber sampling 100–1 Stercorarius parasiticus see skua, Arctic
fatty acid signature 101–13 Sterna paradisaea see tern, Arctic
in polar bear diet 114–15 storm-petrel 83–7
leopard, South Georgia 258 in MSFA 90, 89–95
ringed stratification, North Sea 47, 49, 55, 88–9,
blubber sampling 100–1 180
fatty acid signature 101–13 suppressors 89, 90, 91
in polar bear diet 99, 113–15 surface-feeders 119, 121–2, 123
378 Index

surface-feeders (cont.) higher

effect of cold-water event 124 management goals 325–6, 330
sandeel availability 192–3 predators see predators, top
shallow sea fronts 180–2 trout, Quebec 20
surface pecking 90, 94 tuna 13
surveys, at-sea 178
sustainability 362 uncertainty, predator performance and prey
swamping, of predator 21 availability 266, 267–8
United Nations Convention on the Law of
teleconnections 33–4, 41 the Sea 1982 327–8
telemetry, satellite Uria aalge see guillemot, common
Antarctic fur seal 134, 135, 136 Ursus maritimus see polar bear
gannet 239
macaroni penguin 134, 135 vole, field 269–71
temperature, sea-surface, variability
South Georgia 30, 31, 32–4, 35–6, walrus 20
41 blubber sampling 100–1
effect on krill 257–8 fatty acid signature 101–13
effect on predators 36–41 Wee Bankie
tern piscivorous fish study 229, 230, 231,
Arctic, sandeel availability 192–3 226–33
in MSFA 90, 89–95 sandeel fishery 224, 225, 228
Theragra chalcogramma see pollock, walleye whale, baleen 18
thermocline, importance to mid-water whale, grey 20
divers 184, 181–4, 186 whale, humpback, breeding ground 353
tides 178–9, 180, 186 whale, killer
benthic divers 182, 185, 184–6 predator of sea otter 20
toothfish, Patagonian, spawning biomass predator of Steller sea lion 278, 280, 289
333 whale, minke 83–7, 93–5
top-down forcing 143, 147 Barents Sea scenario modelling 310–21
total allowable catch predation 316–17
Barents Sea scenario model 310–21 recruitment 315
cod 225, 318 whaling 318, 319
harp seal 318–19 whaling
minke whale 318 Barents Sea 318, 319
trawling 12, 17 Bering Sea 17–18
trophic cascades 18, 325, 348 whiting, North Sea, interaction with
trophic level 15, 14–16, 20 sandeel 229, 230, 227–30, 231, 232