Editor's Note

Many of the most elaborate examples of extreme adaptive specialization can be

found in pollination systems. The interactions among flowering plants and insects,
for example, are textbook illustrations of the coevolutionary process and prompted
Darwin to write one of his most influential books, The Various Contrivances by
Which Orchids are Fertilized by Insects. For over a century now evolutionary
Emphasizing biologists have considered extreme specialization in pollination systems the com-
mon outcome of selection, with generalization only a rarity. Major morphological
new ideas features of flowers have been selectively driven through specialization into suites
of canalized characters forming "pollination syndromes" that are adapted to spe-
to cific classes of pollen vector. Likewise, pollinators exhibit extreme specialization
stimulate research in the flowers they will attempt to exploit for critical floral resources. So goes the
story.
in ecology But not so, according to Waser and colleagues. They feel extreme specialization
may not be as ubiquitous a phenomenon as we believe, and that generalization in
pollination systems may be as important an evolutionary outcome as specialization.
In a very thorough review using broad comparisons across large floras, they find
wide evidence for generalization within pollinator-plant interactions. They indi-
cate the consequences of generalization for broader ecological interactions and
raise a general caveat against assuming specialization as the inevitable outcome
of selection among interacting species. Their article should reopen and stimulate
further discussion on the old problem of adaptive specialization vs. generalization
in ecological interaction.
Leslie A. Real

Ecology, 77(4), 1996, pp. 1043-1060

© 1996 by the Ecological Society of America

GENERALIZATION IN POLLINATION SYSTEMS, AND WHY IT MATTERS1

Nickolas M. Waser,2·3 Lars Chittka,4•5 Mary V. Price,2·3
Neal M. Williams,4 and Jeff Ollertons

Abstract
One view of pollination systems is that they tend toward specialization. This
view is implicit in many discussions of angiosperm evolution and plant-pollinator
coevolution and in the long-standing concept of "pollination syndromes." But
actual pollination systems often are more generalized and dynamic than these
traditions might suggest. To illustrate the range of specialization and generalization
in pollinators' use of plants and vice versa, we draw on studies of two floras in
the United States, and of members of several plant families and solitary bee genera.
We also summarize a recent study of one local flora which suggests that, although
the colors of flowers are aggregated in "phenotype space," there is no strong
association with pollinator types as pollination syndromes would predict. That
moderate to substantial generalization often occurs is not surprising on theoretical

1 Manuscript received 24 January 1995; revised 20 July 1995; accepted 22 September

1995.
2
Department of Biology, University of California, Riverside, California 92521 USA.
3
Rocky Mountain Biological Laboratory, P.O. Box 519, Crested Butte, Colorado 81224
USA.
4 Department of Ecology and Evolution, State University of New York, Stony Brook,

New York 11794 USA.

5 Institut fiir Neurobiologie, Freie Universitlit Berlin, Konigin-Luise-Strasse 28-30, 14195

Berlin, Germany.
6 School of Environmental Sciences, Nene College, Park Campus, Moulton Park, North-

ampton NN2 7AL, United Kingdom.

1044 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

grounds. Plant generalization is predicted by a simple Introduction

model as long as temporal and spatial variance in pol- A pervasive idea in pollination biology is that the
linator quality is appreciable, different pollinator spe- interactions between plants and animal pollinators
cies do not fluctuate in unison, and they are similar tend to be specialized. For example, specialization is
in their pollination effectiveness. Pollinator general- implied by the concept of "pollination syndromes,"
ization is predicted when floral rewards are similar which appears as an organizing principle for the field
across plant species, travel is costly, constraints of
in textbooks and in the primary literature (van der Pijl
behavior and morphology are minor, and/or pollinator
1961, Baker and Hurd 1968, Keeton and Gould 1993,
lifespan is long relative to flowering of individual
plant species. Recognizing that pollination systems of- Lunau 1993, Hodges and Arnold 1994). The polli-
ten are generalized has important implications. In eco- nation syndromes are suites of floral traits proposed
logical predictions of plant reproductive success and to reflect adaptations to one or another pollinator type
population dynamics it is useful to widen the focus roughly at the level of orders (e.g., beetles vs. but-
beyond flower visitors within the "correct" pollina- terflies vs. bees) or above (e.g., beetles vs. birds). In-
tion syndrome, and to recognize temporal and spatial deed, because floral traits often distinguish angio-
fluidity of interactions. Behavioral studies of polli- sperm species, pollinator specialization has long been
nator foraging choices and information-processing considered critical to plant speciation and evolution-
abilities will benefit from understanding the selective ary radiation (Grant 1949, Baker 1963, Grant and
advantages of generalization. In studies of floral ad- Grant 1965, Stebbins 1970, Crepet 1983). Along with
aptation, microevolution, and plant speciation one this, some workers have proposed a general evolu-
should recognize that selection and gene flow vary in tionary trend toward specialization (Stebbins 1970,
time and space and that the contribution of pollinators Crepet 1983, 1984 ), with increasingly tight coevolu-
to reproductive isolation of plant species may be over- tion of plants and pollinators (Gilbert and Raven
stated. In conservation biology, generalized pollina- 1975), whose evolutionary best interests might even
tion systems imply resilience to linked extinctions, but
converge (Howell 1979, Kevan and Baker 1983).
also the possibility for introduced generalists to dis-
These ideas are tidy, and enjoy several sources of
place natives with a net loss of diversity.
support. Dramatic specialization does occur in some
Key words: behavior; behavioral constraints of polli- pollination systems. Many biologists perceive repeat-
nators; evolution; foraging theory; generalization in polli- ed patterns in the phenotypic expression of flowers,
nation systems; interaction-web connectance; mutualism; as required by the pollination syndrome concept-sug-
plant-animal interaction; pollination syndromes; solitary
gesting that the longevity of this concept (which dates
bees; specialization; surveys of local floras.
to Delpino 1868-1875) owes something to biological
Relations between flowers and insect pollinators are reality. And specialization and tight coevolution could
archetypes of the results of coevolutionary inter- explain the rapid radiation of angiosperms and taxa of
actions ... Constant pollination might allow a max- animal pollinators during the cenozoic (Crepet 1983,
imum number of plant species ... since ... each Kiester et a!. 1984, Eriksson and Bremer 1992).
plant species might have an efficient pollinator in At the same time, there are problems. Nature pre-
spite of relatively high species density. sents us with coevolution that often is diffuse (Janzen
Crepet 1983:29-30 1980, Schemske 1983, Howe 1984 ), and with many
generalized pollination systems in both temperate and
The flowers of each species are adapted in shape, tropical habitats (Waser 1983, Feinsinger 1987, Rou-
structure, color, and odor to the particular polli- bik 1992, Renner and Feil 1993, Waser and Price
nating agents on which they depend . . . Evolving 1993). Geographic ranges of given plants and polli-
together, the plants and their pollinators become nators rarely correspond closely, indicating that their
more finely tuned to each other's peculiarities . .. interactions are not obligate; indeed, species of plants
Keeton and Gould 1993:476
(including those incapable of selfing) and of pollina-
tors that have invaded new geographic areas often are
Collectors' notes are worthless as data of anthe-
cology because the collector is looking out for par- successful even when their ancestral mutualists were
ticular kinds of insects and not for all the kinds left behind (Cox 1983, Lord 1991, Forster 1994, Her-
which occur on the flowers. Robertson 1928:4-5 rera 1995). On closer examination, even figs, yuccas,
and their visitors may not involve absolutely obligate
Die Farbe bildet ein Merkzeichen, aber keine An- interactions (Wiebes 1979, Pellmyr and Thompson
ziehung an und fur sich fur das /nsekt [Color con- 1992, Patel et a!. 1993, Dodd and Linhart 1994 ).
stitutes a cue, but no attraction in and of itself for To us the issue of how richly connected plants and
an insect]. Fore! 1910: 194 pollinators are seems fundamental, but all too little
June 1996 IS POLLINATION GENERALIZED? 1045

discussed and explored. In this paper we pose two 1908, Robertson 1928, Adams et al. 1981, Anderson
related questions. First, what does the empirical evi- et al. 1988, Cane and Payne 1993, Renner and Feil
dence actually suggest about specialization and gen- 1993). Generalization is even greater when one looks
eralization? To provide a preliminary answer, we draw at the visitors to a given plant species over time and
on several classical surveys of pollination and of plant space, whether on a local or geographic scale (Herrera
and insect taxa, and on a new study of our own. We 1988, 1996, Horvitz and Schemske 1990, Warnke et
conclude that pollination systems range from fairly al. 1993, K wak 1994, Thompson 1994, Bronstein
specialized to very generalized, with moderate gen- 1995, Fishbein and Venable 1996). In part this reflects
eralization the rule rather than the exception (we are temporal fluctuation in the local abundance, or even
forced to describe this continuum with the terms "spe- presence/absence, of given pollinator species; such
cialization" and "generalization," and return to lim- fluctuation can be dramatic (Petanidou and Ellis 1993,
itations of this dichotomy under Conclusions and im- Stubblefield et al. 1993). Pollinators in temperates and
plications). Second, what is the theoretical expecta- tropics likewise tend to visit multiple plant species in
tion? Here we draw on a simple model of plant fitness a single genus, or of diverse genera and families (e.g.,
and on foraging theory, both of which suggest that Feinsinger 1983, 1987, Schemske 1983, Sommeijer et
conditions commonly will favor generalization in pol- a!. 1983, Waser 1983, Absy et a!. 1984, Ackerman
lination systems. We also discuss evidence that this 1985, Pedro 1993, Scott et al. 1993).
conclusion is not jeopardized by constraints of be-
havior, morphology, or physiology. We end by con- Surveying an entire pollination assemblage
templating why it is important for biologists to rec- The studies just cited usually focus on one or a few
ognize the generalized nature and flexibility of many plant or animal species at a single time and place, and
pollination systems. so might represent a biased sample of all pollination
systems. To reduce the chance of bias we advocate
Generalization in Plant-Pollinator Interactions surveys of entire local assemblages of plants and pol-
Ideals to contemplate linators (a daunting task!), without any prior selection
To assess the degree of specialization or general- of species. Several of the studies cited above and oth-
ization we need to know several things. First, we ers analyzed by Jordano (1987) in fact explore portions
would like to characterize pollination from the point of such assemblages. However, the treatment in all
of view of a plant species, recording visitors and as- cases is circumscribed by some limit of taxonomic
sessing which ones pollinate and how effectively (e.g., focus (for example on "hummingbird-pollinated"
Motten et al. 1981, Tepedino 1981 a, Herrera 1987, flowers, whose insect visitors are ignored). We know
Waser and Price 1990). We also would like to under- of only a few studies that treat all or most of a local
stand the pollinator's perspective, characterizing its flora and/or its pollinating fauna.
flower choices in terms of energy and other criteria Robertson ( 1928) surveyed the entire angiosperm
(e.g., Strickler 1979, Stucky 1984, Pleasants and Was- flora around Carlinville, Illinois, USA, and distin-
er 1985, Duffield et al. 1993). Starting with one or a guished flower visitors that pollinated from those that
few species, however, should soon lead us to a web did not. The 375 native species received visits from
of interactions, unless plants and pollinators are ob- as few as one pollinator species each, and as many as
ligately related. We would like to characterize this 298 species in 84 genera, with a mean (median) of
web, and to know how it changes as we sample larger 33.5 (18) pollinator species per plant species (Fig. 1).
temporal and spatial domains, and thus ecological con- Furthermore, 91% of all plant species were visited by
texts (see also Thompson 1994). more than a single animal species and thus were at
least somewhat generalized.
Fragmentary studies Several surveys are available for faunas of solitary
Most pollination studies to date, our own included, bees, which are major pollinators in many ecosystems.
provide only a fraction of the total perspective just These surveys suggest that specialization, expressed
outlined. Despite their limitations, such studies often as "oligolecty," is rare in tropical forests, more com-
convey the impression of generalization and fluidity. mon in temperate regions, and predominant in deserts
Plants often employ a variety of pollinating agents (Table 1). However, there are major caveats. First, oli-
(e.g., Robertson 1928, Schremmer 1953, Beattie 1971, golecty is defined as specialization at the level of plant
Primack 1979, Waser 1982, Kephart 1983, Bullock et genus or family, not species. It also refers only to
al. 1989, Carpenter 1989, Pettersson 1991, Roubik plants visited for pollen, whereas solitary bees also
1992, Erhardt 1993), and mixtures of abiotic and biotic visit plants for nectar and pollinate them in the pro-
agents (e.g., wind and insects) are not taboo (Knuth cess. Even the most extreme oligoleges appear to be
1046 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

0
0.5 species were observed by Clements and Long (I923)
Cl)
·c:; to receive visits from as few as I animal species, and
Cl)
0.4 as many as 62 species in 30 genera. The frequency

-
Q.
0 distribution of visitors was right-skewed (Fig. 2A),
c

-
with means (medians) of 9.8 (4) visitor species and
ca 0.3
Q. 6.I (3) genera per plant species; 80% of all plant spe-
0
cies were visited by >I animal species. By compar-
c 0.2 ison, the 268 native animal pollinators were observed
0 to visit as few as I and as many as 37 plant species
"E in 32 genera (Fig. 2B), with means (medians) of 3.3
0 0.1
e
Q.
(1) plant species and 2.8 (1) plant genera visited; 48%
D.. of all animal species visited >I plant species. The
0
1 3 6 9 12 15 18 21 24 2730 33 36 39 42 45 48 >48
overall connectance of the interaction web (the frac-
tion of all possible plant-pollinator interactions ac-
Number of visitor species
tually realized) was 3.6%. This is lower than the av-
FIG. I. Pollinator affinities in the Carlinville, Illinois erage connectance of 29.4% reported by Jordano
(USA) flora, from records in Robertson (1928). Frequency
distribution of the proportions of 375 native plant species
receiving pollination visits from different numbers of animal 0.5
0
species. Cl) 0 Species
·c:;

-
El Genera
& 0.4
generalized to nectar sources (Westrich I989). Finally, 0

-
those few bee species that appear to visit only a single c
ca 0.3
species for pollen in a given area sometimes visit other Q.
species in other parts of their geographic range (Cross 0
and Bohart I960, Eickwort and Eickwort I969, Minck- c 0.2
0 A
ley et al. I994 ), and stray from their speciality when "E
it is in short supply (Michener and Rettenmeyer I956, 0 0.1
c.
Thorp I969, Cruden I972, Michener I979, Adams et
al. I98I, Cane and Payne I988). Thus, the surveys of
e
D..
solitary bees actually suggest again that some degree
3 6 9 12 15 18 21 24 27 30 33 36>36
of generalization and opportunism is common.
A final survey, unique as far as we know in pro- Number of visitor taxa
viding tabulations both by plants and animals, com- 0
Cl)
0.6
prised 94 native plant species in montane conifer forest ·c:;
Cl)
and grassland on Pikes Peak, Colorado, USA. These c. 0.5
0
a; 0.4
E

-
TABLE I. Relative abundances of oligolectic bees, taken
to be specialists (but see Generalization in plant-polli- "2
nator interactions: Surveying an entire pollination assem-
ca 0.3
blage for caveats), in some different biomes and areas of 0
the Americas. Data are from Heithaus (1979), Michener c 0.2
0
(1979), Moldenke (1979), Absy et al. (1984), and de Me-
nezes Pedro (1993 ). "E
0 0.1
c.
Total e
D..
no. bee No. oligo- %oligo- 0
Biome or Area species lectic lectic 3 6 9 12 15 18 21 24 27 30 33 36 >36
Desert 890 592 66.5 Number of plant taxa
Great Basin 333 165 49.5
Great Plains 500 184 36.8 FIG. 2. Plant and pollinator records from Pikes Peak,
Pacific Northwest 425 104 24.5 Colorado (USA), from records in Clements and Long (1923).
Southern forest 280 63 22.5 (A) Frequency distribution of the proportions of 94 native
Rocky Mountain 500 90 18.0 plant species receiving visits from different numbers of an-
Tundra/Muskeg 84 13 15.5
imal species and genera. (B) Frequency distribution of the
Tropical wet forest 22 I 5.0
Tropical grassland 192 proportions of 268 native animal species visiting different
3 1.5
numbers of plant species and genera.
June 1996 IS POLLINATION GENERALIZED? 1047

(1987), but the small fragments of pollination webs 0.5

0 Species
that he considered averaged only 44 species of animals
and plants combined (hereafter "S"), compared to 359 1'21 Genera
0.4
at Pikes Peak. For connectance to stay constant as S
increases requires that interactions per species rise ap- 0.3
proximately as S2 • Whereas entire food webs may be-
have this way within limits (Martinez 1992), con-
nectance does appear to decline with S in pollination 0.2
webs (Jordano 1987). At 3.6% the Pikes Peak system
actually is substantially more strongly connected than 0.1
the value of 0.1% predicted from an empirical rela-
tionship derived by Jordano (1987). This suggests that 0
fragments of pollination webs if anything underesti- 1 3 6 9 12 15 18 21 24 27 3033 36 39 42 4548>48
mate the connectance and generalization in entire as- 0.7
semblages-just as expected from the limited taxo- Ill

nomic focus inherent in such fragments. ·uCDCD 0.6

c.
Ill 0.5
Surveys by plant or animal taxon 'E
IV
Another way to reduce bias in examining pollination c. 0.4
interactions is to characterize them at the level of high- 0
c 0.3
er plant or animal taxon, without any prior selection 0
of species. Some published pollination studies already t: 0.2
8. B
focus at the level of plant or animal family or other
taxon.
...
0
Q.. 0.1
To illustrate, we begin with examples of three an-
0
giosperm families. We have extracted information on 1 3 6 9 121518 21 24 27 3033 36 3942 45 48>48
European members of the buttercup family (Ranun- 0.5
culaceae) and orchid family (Orchidaceae) from Knuth
(1908 and 1909, respectively), and on North American
0.4
members of the phlox family (Polemoniaceae) from
Grant and Grant (1965). These choices represent re-
spectively a large "primitive" family, a large "ad-
vanced" family, and a small "advanced" family (sen-
su Stebbins 1974); the latter two usually are presumed
to be specialized in pollination. For each family we
surveyed the published source in question one species
at a time, tabulating all recorded flower visitors. We
excluded any case where a plant species was examined
outside its natural habitat. The results resemble those
1 3 6 9 12 1518 21 24 2730 33 3639 42 45 48>48
from the Carlinville and Pikes Peak floras. The 39
members of the Ranunculaceae received visits from Number of visitor taxa
as few as 1 animal species, and as many as 53 species
FIG. 3. Pollinator affinities of members of different plant
in 29 genera (Fig. 3A), with means (medians) of 11.8 families, shown as frequency distributions of the proportions
(10) visitor species and 8.8 (6) genera per plant spe- of plant species receiving visits from different numbers of
cies; 95% of all plant species received visits by > 1 animal species and genera. (A) Thirty-nine European mem-
animal species. The 11 members of the Orchidaceae bers of the Ranunculaceae, from records in Knuth (1908).
(B) Eleven European members of the Orchidaceae, from rec-
received visits from as few as 2 species in two genera ords in Knuth (1909). (C) Seventy-three North American
and as many as 15 species in 13 genera, with means members of the Polemoniaceae, from records in Grant and
(medians) of 7.9 (8) visitor species and 5.0 (2) genera Grant (1965).
per plant species; none was limited to a single animal
species (Fig. 3B). The 73 members of the Polemoni-
aceae received visits from as few as 1 species and as
many as 39 species in 17 genera, with means (medians)
of 6.5 (4) visitor species and 5.4 (3) genera per plant
1048 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

0.5 ily, the Colletidae, visited on average 3.4 plant fam-

0 Species ilies for pollen or nectar; members of the most speciose
~ Genera
A

•
0.4 family, the Halictidae, visited on average 8.4 plant
Families
families, and members of the most generalized family,
the Apidae, visited on average I8.4 plant families.
0.3
These numbers suggest substantial generalization, but
at a crude taxonomic resolution. A finer resolution is
0.2 provided by Westrich's (1989) monograph on bees of
southwestern Germany. From the 26 genera containing
0.1 more than 5 species each we randomly selected An-
drena, Hylaeus, and Osmia, representing three differ-
0 ent families (respectively Andrenidae, Colletidae, and
Megachilidae). We tabulated the numbers of plant
0.3 families, genera and species visited by each bee spe-
(I) B cies. One species of Andrena and two of Osmia were
CD apparently monolectic (obligate to a single plant spe-
"(5
CD cies as pollen source) over the geographic range stud-
c.
(I)
0.2 ied by Westrich. Otherwise, generalists predominated.
CD
_g The 90 species of Andrena visited means (medians)
of 9.4 (6) plant species, 7.7 (5) genera, and 4.1 (2)
0 families (Fig. 4A). The corresponding values were 8.8
c
0 0.1 (7) species, 8.0 (7) genera, and 4.6 (4.5) families for
E the 24 species of Hylaeus (Fig. 4B); and I 0.9 (8) spe-
0
c. cies, 8.5 (6) genera, and 3.9 (2) families for the 37
~ species of Osmia (Fig. 4C). Interestingly, a recent phy-
ll.
logenetic mapping of pollen use within the tribe An-
1 2 3 4 56 7 8 910111213141516171819>19
thidiini of the Megachilidae suggests that polylecty
0.5 (generalization in plants visited for pollen) has
c evolved repeatedly from oligolecty, but not vice versa
0.4 (Miiller 1996).

0.3
Biases in surveys
We recognize that conclusions based on surveys are
imperfect. The problem that comes immediately to
0.2 mind is that surveys are likely to overestimate the
degree of generalization. Flower visitation is not a
0.1 synonym for pollination, and even if non-pollinating
visitors are excluded (as in Robertson 1928), the re-
maining visitors will differ in pollination efficiency
1 2 3 4 56 7 8 910111213141516171819>19 (see citations in Introduction). Similarly, a given in-
Number of plant taxa sect or other animal recorded at a given flower may
visit only incidentally. Ignoring these details will ex-
FIG. 4. Flower affinities of members of three different
aggerate the connectance of plant-pollinator interac-
genera of solitary bees, shown as frequency distributions of
the proportions of bee species visiting different numbers of tions. However, there also are counterbalancing biases
plant species, genera, and families, from records in Westrich towards underestimating generalization and opportun-
(1989). (A) Ninety members of the genus Andrena. (B) ism in pollination mutualisms. One of these is that
Twenty-four members of the genus Hylaeus. (C) Thirty-sev- surveys almost invariably are very restricted in tem-
en members of the genus Osmia.
poral and spatial dimensions. But pollination inter-
actions vary along these dimensions, as already noted,
species; 82% received visits from >I animal species so surveys will underestimate their complexity. Spe-
(Fig. 3C). cialization of a particular pollinator group, the solitary
Turning to pollinators, we again chose bees as our bees, is overestimated by the practice of studying pol-
focus. Heithaus (1979) used published records to es- len collection, and the paucity of records of plants
timate that members of the most specialized bee fam- visited for nectar. Finally, there may be a widespread
June 1996 IS POLLINATION GENERALIZED? 1049

unconscious bias to ignore VISitors that seem "im-

proper" under the paradigm of pollination syndromes
(see the quotation from Robertson [1928] at the be-
ginning of this paper). As a first approximation, the
presence of biases in both directions suggests that sur-
vey results may reflect large-scale patterns with rea-
sonable accuracy.
Consider two examples that illustrate bias toward
underestimating the degree of generalization. Two of
us (N. M. Waser and M. V. Price) have studied pol-
lination of the montane plant lpomopsis aggregata for
20 yr. Grant and Grant ( 1965) list eight species of
flower visitors in seven genera, to which Waser (1978,
1982) adds at least four species in four genera known
E(U
to transfer pollen (Hyles lineata, Bombus appositus,
Papilio rutulus, Halictus sp.). Clements and Long
(1923) list only two species in the genus Halictus-
thereby omitting entirely the hummingbird pollinators
of this classical "hummingbird syndrome" species.
And, even a combined listing of= 12 pollinator species
FIG. 5. The positions of 154 flowers from the nature
must be incomplete given the limited sample of sites reserve near Berlin, plotted in a hexagonal color space.
and years. Thus it appears that Grant and Grant (1965) Points in the upper part of this space are bee blue colors,
underestimated the number of species-level pollina- points in the lower left corner are bee ultraviolet (UV) colors,
tion mutualisms in which /. aggregata engages by at etc. The irregular polygon within the hexagon connects the
least 33%, and Clements and Long (1923) underesti- loci of monochromatic lights in 10-nm steps from 300 nm
to 540 nm; its curved bottom segment connects the loci of
mated by at least 83%. As an animal example, consider 300 nm and 540 nm in nine mixtures of the two colors with
the Broad-tailed Hummingbird, Selasphorus platycer- ratios of 0.9:0.1, 0.8:0.2, ... 0.1 :0.9 (see Chittka 1992). One
cus, reported by Clements and Long (1923) to polli- of the 10° sectors in which color loci were counted for Fig.
nate 7 plant species and by Waser (1983) to regularly 6 is indicated by dashed lines in the lower left (this particular
sector contains no flowers).
visit an additional 12 species in eight families (most
of which do not conform to a "hummingbird polli-
nation syndrome"). Clements' and Long's list com- er color, a central element of putative pollination syn-
prises only 37% of the total of 19 species, and this dromes. Five sites were chosen in a nature reserve
total does not come close to encompassing the floral near Berlin (Naturschutzgebiet Lange Dammwiesen),
diet of broad-tails across years and sites. and flower colors of all 154 plant species were quan-
tified by measuring their spectral reflectance from 300
A different approach: surveys of to 700 nm. A point was then calculated for each flower
floral phenotype space in a hexagonal space that describes color as the visual
The concept of pollination syndromes contributes system of honey bees and other bees will perceive it
to an impression of specialization in pollination sys- (Fig. 5). Angular position of a point in this color hexa-
tems. Do patterns in nature actually conform to those gon corresponds to hue (Wyszecki and Stiles 1982).
predicted by syndromes; i.e., do flowers form clusters To see whether flowers form clusters of similar hues
in "phenotype space" that correspond to visitation by we divided the space into sectors of 10° each and
major pollinator groups? Few studies seem to have counted the number of plant species occupying each
addressed this issue. Baumberger ( 1987) analyzed the sector (see Chittka et a!. 1994). Indeed, flowers did
corolla lengths of over 1000 "bird syndrome" flowers form distinct clusters at angular intervals of =60° (Fig.
worldwide, and found that values cluster around an 6). Furthermore, these clusters also are recognizable
intermediate length (16-26 mm) that matches bill to other color vision systems (Chittka et a!. 1994 ),
length in three major families of nectar-feeding birds including our own (Kevan 1978). However, the syn-
(mean =19 mm). Unfortunately, the definition of bird drome concept also would predict that different pol-
flowers along syndrome lines contributes a circularity. linators largely restrict their attention to different sub-
To avoid this it is important to sample plant species sets of all clusters. To see if this was so, all visitors
at random or to sample an entire flora. One of us (L. to flowers over a 1-yr period were identified at the
Chittka) adopted the latter approach in a study of flow- level of insect order. Fig. 7 shows the percentages of
 1050 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

uv UV- BLUE BLUE- GREEN UV- cializes. In any one flowering episode the fecundity
35 BLUE GREEN GREEN
0 of a generalist (hereafter its "fitness") is
Cl)
·uCl) 30
(1)
c.
...c
0 25

20
where N 1 is the abundance of pollinator species I, V1
is the flower visitation rate per individual pollinator,
ca

-...
and g 1 is the effectiveness of each visit, i.e., the degree
Q.
15 to which it contributes to individual reproductive suc-
0
cess. Terms subscripted with "2" refer to identical
Cl) 10 attributes of pollinator species 2. Now, imagine a rare
.0
E 5
mutation (for example in flower morphology and
:::J
z placement of stamens and/or pistils) that increases the
0 effectiveness of species 1 by o, and equally decreases
0 60 120 180 240 300 360 that of species 2. This mutation involves the most
complete possible trade-off in fitness value of the two
Angle in color hexagon ( 0
)
pollinators to the plant, and represents the greatest
FIG. 6. The distribution of all flowers depicted in Fig. 5 possible increase in plant specialization. Its fitness is
in the 36 10° sectors of the color hexagon. Zero degrees
represents the line separating the UV-green and UV portions Ws. 1 = N 1 V1(g 1 + o) + N 2 Vig 2 - o). (2)
of the hexagon (see Fig. 5). Comparison of Eqs. I and 2 shows that the mutation
for increased morphological specialization spreads if
Ws.J >We, i.e., if N 1 V1 > N 2 V2 • Thus, assuming that
different color categories visited by flies, beetles, but- pollinators differ in abundance, a mutation that in-
terflies, large bees (honey bees or larger), and small creases the efficiency of the most abundant visitor
bees (smaller than honey bees). There were no statis- (with the larger product of N and V) will increase in
tical differences among these animal groups in the frequency, even if in the process the plant becomes
colors of flowers visited (X 2 = 29.8, df = 20, P > more specialized on that visitor. Similarly one can
0.05), or between the colors visited by any group and imagine a mutation (perhaps in reward or floral dis-
those available in the flora (in all cases x2 ~ 7 .I, df play) that enhances the attractiveness to one pollinator
= 5, P > 0.1). To be sure, there were hints of classical species, and thus its per-capita visitation rate, at the
syndromes: flies and beetles visited slightly more bee expense of reduced visitation by the other pollinator.
blue-green flowers (mostly human white), whereas The fitness of this mutation is
large bees and butterflies visited more bee UV-blue
and bee blue flowers (human blue, purple, pink, and 0.8
D Large bee
violet). But these are only trends. For this flora, then, 0.7 IZl Small bee
we reject the idea that clusters of flower colors rep-
0.6 B Fly
resent syndromes by which plants address specific
groups of potential visitors. Clustering in color space
c B Butterfly
0 0.5
instead may mostly reflect constraints imposed by the t:0 D Beetle
chemistry of plant pigments (Chittka et a!. 1994 ). c. 0.4
e
D..
0.3
When Should Plants Generalize?
0.2
A short-term perspective
The empirical evidence suggests that pollination 0.1
mutualisms often are diversified and opportunistic,
0
and therefore constitute true food webs as opposed to
UV UV- BLUE BLUE- GREEN UV-
food chains. Can we understand why this is so? We BLUE GREEN GREEN
start by sketching out a basic model that considers the
optimal diversity of pollinating animals from a plant's
Color category
perspective. Our approach is similar to that in several FIG. 7. The colors of flowers visited by different groups
previous models of ecological processes (e.g., Real of insects. Histograms indicate the percentages of all plant
species visited by a given insect group that fell into a given
1980, Lacey et a!. 1983, Real and Ellner 1992).
color category. For example, the highest stippled histogram
The model compares a phenotype that generalizes shows that 66% of all plant species visited by beetles were
on two pollinator species with a phenotype that spe- in the bee blue-green category.
June 1996 IS POLLINATION GENERALIZED? 1051

Ws,l = NI(VI + o)gl + Nz(Vz - o)gz. (3) short-term benefit of specializing. There is no unique
solution, but the inequality is favored if Cov(X 1,X2) is
Comparison of Eqs. 1 and 3 shows that the mutation
near zero, for example because the two pollinator pop-
will spread if N 1g 1 > N28 2 • Thus, assuming that pol-
ulations fluctuate largely independently of one anoth-
linators differ in efficiency, a mutation that increases
er; or is negative, for example because the populations
the attraction of the most efficient pollinator will
fluctuate so as to compensate for one another.
spread, even at a loss of attraction to another visitor
The results of Pettersson ( 1991) are interesting in
and thus a net increase in specialization.
this context. He measured abundances of 26 moth spe-
A long-term perspective cies visiting Silene vulgaris over 4 yr. The eight most
So far, results of the model agree with the classical abundant species provided 75% of all pollination. The
view that specialization on a single best pollinator is coefficient of variation (cv) in abundance of these
favored. What happens, however, when the abun- species taken one at a time averaged 1.2, whereas the
dances, per-capita visitation rates, or (perhaps less cv for the total of eight species was 0.4. This analysis
likely) the efficiencies of different pollinators (theN, is made without specific reference to the generation
V, and g terms) vary temporally, so that in the extreme time of S. vulgaris, but it suggests that a generalist on
the "best" pollinator changes across plant generations many moth species (as S. vulgaris is) enjoys a lower
(e.g., Petanidou and Ellis 1993, Stubblefield et a!. variance discount than the average specialist would.
1993)? The long-term fitness of plants is the geometric
mean over generations (here we imply non-overlap- Conclusions
ping generations, but a similar argument applies for The model suggests that a plant may gain an ad-
overlapping generations). An approximate equiva- vantage at any given time and place by specializing
lence is on the most abundant and/or efficient pollinator. But
with temporal variation in pollinator services, spe-
W = J.L - [var(W)/(2J.L)], (4) cialization becomes less likely. This is true especially
for plant species with few reproductive episodes and
where J.L is the arithmetic mean fitness over n gener-
lacking other persistent life stages, since they face
ations (i.e., J.L = lin I W,, where i indexes generation),
local extinction if they specialize on a pollinator that
and var (W) is the temporal variance in fitness (Lacey
disappears for one or more years. Spatial variation in
et a!. 1983). Thus the arithmetic mean fitness is dis-
pollinator faunas, and in the identity of the "best"
counted by a function of the temporal variance.
pollinator, also fosters greater generalization, at least
Assume that a specialist plant genotype specializes
at the plant species level.
completely on pollinator species I, achieving short-
We might therefore expect specialization on one or
term fitness of W5, 1 = k N 1,;V1,1g 1,1• The term "i" again
a few pollinators mostly in species that are large, long-
refers to a specific generation, whereas k is a constant
lived, have many reproductive episodes, and/or have
multiplier that describes the total gain in N, V, and/or
pollinators whose populations, for whatever mysteri-
g as a consequence of specializing on one pollinator.
ous reason, fluctuate little, so that the identity of the
The fitness of the specialist over the long term, during
"best" pollinator tends to remain constant across
which there is fluctuation in the product of N, V, and
years (this is not to imply that all such species will
g (here symbolized as X), is
specialize). It is difficult to "test" these predictions
Ws,J = k E(X1) - { [k var(X1)] I [2E(X 1)]}, (5) because we lack a wealth of examples of obligate spe-
cialization across phylogenetically diverse plant taxa,
where E(X 1) is the expectation of X 1, i.e., the arithmetic
but note that figs and yuccas do seem to fit the ex-
mean, as experienced by a generalist. The long-term
pectations.
fitness of the generalist is
We= E(X 1 + X2) When Should Pollinators Generalize?
- {[var(X1) + var(X2) A short-term perspective
There is a temporal element to the benefits required
+ 2 cov(X1,X2 )] I [2E(X 1 + X2 )]}. (6)
by both partners in a pollination mutualism. As we
Under what conditions does the generalist outper- have just discussed, a plant gains reproductive success,
form the specialist (i.e., We > W5, 1)? Eqs. 5 and 6 measured over a long term. A pollinator usually gains
show that the answer depends on six parameters: arith- food, which is critical to its survival and reproduction
metic mean fitness contributions of each pollinator, over both short and long terms..
variances in fitness contributions, the covariance in A pollinator that forages efficiently is most likely
fitness contributions, and k, which encapsulates the to accumulate a surplus of energy and/or free time,
1052 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

and thus to survive day-to-day, reach maturity, and is more strictly analogous to the plant's perspective
reproduce successfully (Schoener 1971, Pyke 1984 ). (see When should plants generalize?, above). The
What are the trade-offs to specialists vs. generalists? abundances of different flower species at a given site
A foraging pollinator usually encounters a mixture of fluctuate across time periods (e.g., years; Waser and
flower types (e.g., species). If energy rewards differ Real 1979, Tepedino and Stanton 1980), and abun-
greatly among these types, costs of extracting them dances also vary through space. Thus even if polli-
are high, and/or travel costs are low, a nearby flower nators specialize within a single season, variation in
with inferior rewards can profitably be skipped in fa- plant abundance may foster generalization of lineages
vor of a superior distant flower, and a specialist that or colonies, if not of individuals.
does so achieves the highest rate of reward intake.
When rewards are similar and/or travel is costly, how- Behavioral, physiological, and
ever, the optimal strategy for a pollinator with knowl- morphological constraints
edge of upcoming options (an experienced pollinator The exploitation of a flower requires behavioral
or one that is "periscopic," i.e., can see flowers in abilities to detect signals and associate them with re-
advance of reaching them) is to generalize by includ- wards, along with morphological and physiological
ing multiple flower types in its foraging path or itin- abilities to extract and use the rewards. Are there con-
erary (Stephens and Krebs 1986, Mitchell 1989). In straints on these abilities that commonly force polli-
fact, travel is energetically costly (Lasiewski 1963, nators to specialize?
Hainsworth and Wolf 1972, Heinrich 1979, Balder- For the pollinator to add a novel flower to its diet
rama et al. 1992), and floral rewards (at least nectar it first must detect a novel floral signal. The most
rewards) often are similar across species because pol- important signal modality in pollination systems is
linator foraging tends to equalize the effective value vision, and just as flowers are brightly colored, many
of all flowers (Heinrich 1976, Martinez del Rio and pollinators are known to possess color vision. The
Eguiarte 1987, Dreisig 1995). trichromatic set of ultraviolet (UV), blue, and green
receptors found in many bees and wasps appears to
A long-term perspective be an optimal system for flower color coding (Chittka
Most flower-visiting animals face changing and and Menzel 1992), allowing its bearer in theory to
variable resources. Even if the best short-term strategy discriminate on the order of 100 distinct flower colors
is to visit only a single flower type, over the longer (Chittka et al. 1993). Equivalent color receptors are
term a pollinator often should visit many types, for found in some oligolectic solitary bees (Peitsch et al.
two reasons. 1992), suggesting that constraints on color vision can-
The first reason is seen by contemplating a neces- not explain the putative specialization of these ani-
sary condition for visiting only a single flower type: mals. The visual range of most other pollinating in-
that the pollinator's life cycle (at least the portion that sects (e.g., flies, wasps, some butterflies and beetles)
depends on floral resources) be contained within the is similarly broad, often extending from near UV
flowering period of the host plant. This condition may (around 320 nm) to near red (600-650 nm) (von Hel-
be met through a short pollinator life cycle and/or a versen 1972, Bernard and Stavenga 1979, Menzel and
long flowering period (perhaps comprised of staggered Backhaus 1991, Peitsch et al. 1992), and red flowers
flowering episodes of individuals, as in figs; Bronstein certainly are not "invisible to insects" as a pervasive
et al. 1990, Patel et al. 1993), along with synchrony truism of pollination biology claims (see also Pleas-
between the two (e.g., Galil 1977, Aker 1982, Dodd ants and Waser 1985). Pollinating birds often surpass
and Linhart 1994). More commonly, though, the con- insects in possessing a broad spectral range (Gold-
dition is not met because pollinators, although short- smith and Goldsmith 1979, Bennett and Cuthilll994).
lived, do not always achieve synchrony with their pu- Once it has detected a flower, the pollinator next must
tative host (e.g., Cruden 1972); because pollinators (as assess the quality of floral reward. Pollinators of sev-
individuals or colonies) outlive the flowering period eral species are known to possess this ability (Forel
of any one plant species (Free 1970, Heinrich 1979, 1910, Hainsworth and Wolf 1976, Heinrich et al. 1977,
Calder et al. 1983, Paton 1985, Boggs 1987); or be- Menzel 1985, Haslett 1989, Harder 1990, Greggers
cause pollinators experience successive generations and Menzell993) as well as the ability to match floral
within a single growing season that encompasses the choices to rewards (Heinrich 1976, Pleasants 1981,
flowering of many plant species (e.g., Rozen 1958, Greggers and Menzel 1993, Dreisig 1995). Further-
Gregory 1963, Fleming 1970, Kislev et al. 1972, Estes more, pollinators such as hummingbirds, bees, flies,
and Thorp 1975). and butterflies can associate color signals with rewards
The second reason to generalize over the long term (Goldsmith and Goldsmith 1979, Mazokhin-Porshnya-
June 1996 IS POLLINATION GENERALIZED? 1053

kov et a!. 1984, Waser and Price 1985, Chittka et a!. Some readers will argue that this is not a new real-
1988, 1992, Dukas and Real 1991, 1993, Menzel and ization, and, indeed, many of the classic works that
Backhaus 1991, Weiss 1991, Troje 1993, Dukas and we cite as promoting specialization do contain caveats.
Waser 1994, Fukushi 1994); can do so quickly (e.g., But these caveats seem to have been largely ignored.
Menzel 1967, 1985); and can remember the results Perhaps this reflects a very human desire to perceive
(Lindauer 1963, Menzel 1967, 1985, Dukas and Real nature as well ordered. Perhaps it derives in part from
1991). Although some insect pollinators do appear to the unfortunate implication that specialization and
have inborn preferences for certain floral signals (e.g., generalization form a dichotomy. It would help if we
Chittka and Lunau 1992, Giurfa et a!. 1995), close could replace this implied dichotomy with a contin-
inspection suggests that these are rarely fixed and usu- uum, although the tendency to dichotomize and then
ally can be overwhelmed by the insects' capacities for to adopt one option runs deep in western culture (Hil-
learning and memory (Menzel 1985, Dobson 1987). born and Stearns 1982). And, we have devised no
Differences in the morphology of pollinators, es- attractive terminology to describe the continuum of
pecially of the feeding apparatus (Beattie 1971, Wolf pollination interactions (although at a community lev-
and Hainsworth 1972, Inouye 1980, Gilbert 1981, Du- el the interaction-web connectance provides a descrip-
kas and Shmida 1989, Paton and Collins 1989, Fein- tive metric, as long as one can correct for web size).
singer 1990, Grant and Temeles 1992), but also of Other readers may feel that we have wrongly rejected
other body parts (Feinsinger and Colwelll978, Thorpe elements of the established paradigm: that there are
1979, Collins and Paton 1989) and in overall body detectable associations between plant phenotypes and
size, interact with the morphology and size of flowers pollinators (i.e., pollination syndromes); that more
to influence the efficiency of flower feeding. Con- complex flowers within a plant family and more evo-
versely, these interactions influence the efficiency of lutionarily "advanced" plant families are more spe-
pollination from the plant's viewpoint (Waser 1979, cialized, and so on. Our response is that much of the
Nilsson et a!. 1987, Armbruster 1990, Campbell et a!. acceptance of these patterns has rested on a weak em-
1991). However, the classical view that constraints of pirical foundation. Although, for example, our survey
morphological fit are strong enough to ensure spe- of small parts of three plant families suggests that
cialization (Trelease 1881, Grant 1949, 1994 )-strong members of the "primitive" Ranunculaceae do tend
enough for example to cause "mechanical" isolation to be less specialized than members of the "advanced"
of sympatric congeners-seems only rarely correct Orchidaceae or Polemoniaceae, it would be risky to
(e.g., Ramirez 1970). conclude that lineages always evolve toward special-
Other potential constraints involve digestive phys- ization (consider the many species in the "advanced"
iology. For example, hummingbirds digest different Asteraceae that appear to be generalized for pollina-
sugars with somewhat different efficiencies (Martinez tion). It is equally risky to conclude based on taxon-
del Rio 1990). Larval survival, growth rate, and size omy or morphology that a given flower is specialized,
at pupation depend markedly on species of pollen for or what "its pollinator" is. In short, there is a dis-
some solitary bees (Levin and Haydak 1957, Bohart tinction between laws and trends: only with the former
and Youssef 1976). Those pollens that are of low nu- can one argue consistently from the general to the
tritional value or even are toxic to one bee species, specific. Much of what has passed for laws in polli-
however, may support normal development of others nation biology may represent trends at best.
(Tepedino 198lb). Such effects of diet on pollinator To strengthen the empirical foundation on which we
efficiency or fitness may somewhat limit the range of base conclusions about the fundamental nature of pol-
plant species used. lination systems-who interacts with whom, where,
In summary, behavioral, morphological, and phys- when, how strongly, and under what ecological cir-
iological constraints do exist, and will influence plant- cumstances-community-wide or taxon-wide studies
pollinator associations on short and long time scales. of pollination, including the relative pollination effi-
However, these constraints rarely seem strong enough ciencies of different flower visitors, will be invaluable.
to restrict pollinators to a single plant species or only We need careful field observation and solid natural
a few species. history, with the support of experiments. And more
effort in "combing" the existing literature would be
Conclusions and Implications well repaid.
Some philosophical points A better understanding of pollination interactions
Generalization-the use of several plant species by will be very valuable for both. pure and applied in-
a pollinator and of several pollinator species by a vestigations. In what follows we support this assertion
plant-appears to be the rule rather than the exception. with some examples.
1054 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

Some implications for ecological studies populations by their mostly-localized movement of pol-
In ecological studies we seek to understand patterns len, there has been little success in quantitatively match-
of plant (or more rarely pollinator) reproductive suc- ing direct estimates of pollinator-mediated gene flow
cess across sites, seasons, and ecological contexts with spatial genetic patterns (Waser 1987, 1993, Camp-
(presence of putative competitors or other enemies; bell and Dooley 1992). In part this may be because
and of direct or indirect mutualists, e.g., Waser and some pollen-carrying visitors have been ignored. Fi-
Real 1979), and thus to understand dynamics of pop- nally, consider studies of differentiation on the scale of
ulations, distributions of species, and their relative speciation. We urge those undertaking studies of plant
abundances in communities (Feinsinger 1978, Waser hybrid zones, and other situations that provide glimpses
1978, Armbruster and Herzig 1984, Kwak et al. 1991, of the speciation process, to examine all flower visitors,
Gathmann et al. 1994 ). In order to succeed we must not only those presumed to be "correct." Acknowl-
explicitly embrace variance in interactions at different edging widespread opportunism and generalization re-
temporal scales (e.g., diurnal, seasonal, annual) and opens important questions about the contribution of pol-
spatial scales (e.g., neighborhood, landscape, geo- linators to plant speciation and their role in defining
graphic), rather than describing only average inter- plant species by influencing patterns of reproductive
actions (see also Thompson 1994, Bronstein 1995). isolation.
We must study all flower visitors, including those that
seem "improper" based on perceived pollination syn- Some implications for agriculture and conservation
dromes. Such visitors may contribute a surprising Many of the points just raised bear on applied issues
share of plant fecundity (e.g., Motten et al. 1981, Te- as well. For example, crop plants may interact with
pedino l981a, Waser 1979, Waser and Price 1990, surrounding plants in beneficial or detrimental ways
Scott et al. 1993), and of interspecific interaction. through overlap in pollinators (Free 1970). Under-
standing the behavioral abilities of pollinators is crit-
Some implications for behavioral biology ical for choosing and managing them in crop situations
In the past few decades, behavioral biology has been (Parker 1981, Tepedino 198la). And recognizing all
revitalized by fusing traditions from ethology and psy- the agents of gene flow is critical in assessing risks of
chology, by an injection of Darwinian thinking and of crop contamination, escape of engineered genes (e.g.,
optimality modelling, and by input from cognitive sci- Klinger et al. 1992), and genetic swamping of endan-
ences. Many empirical studies over this period have gered plant species (Ellstrand 1992).
involved pollination systems, for good reasons: pol- There are other implications for conservation bi-
linators often are straightforward to observe and to ology. A high connectance of interaction webs may
use in experiments, and the stimuli provided by plants influence behavior of the system following perturba-
often can be assessed. In such work it is useful to tion (Kevan 1991 ). Consider linked extinctions of
abandon preconceived ideas about who visits whom, plants and pollinators. Based on a preconception of
or about what animal group is inherently capable of specialized plant-pollinator interactions, regulatory
what, in order to better understand information-pro- agencies have sometimes anticipated linked extinc-
cessing and problem-solving abilities. Comparing the tions where they are unlikely (e.g., Nabhan and Flem-
behaviors of very different animals (e.g., birds and ing 1993). Effort could be saved if resource managers
bees) at similar flower-visiting tasks holds additional had complete information on the interactions in pol-
promise (e.g., Waser and Price 1985). lination systems they must manage. This is not meant
to imply that we should ignore extinctions or that gen-
Some implications for evolutionary studies eralization buffers systems from any ill effects. In-
Different flower visitors may each contribute to se- deed, generalized pollination relationships can penal-
lection on floral traits, and a recognition of temporal ize locally rare plant species through interspecific pol-
and spatial variation may be more fruitful than an len transfer and other forms of competition for ser-
average, equilibrium view (e.g., Thompson 1994). vices of shared pollinators (Waser 1978, Campbell
Furthermore, a focus on the most obvious pollinator at 1986, Kunin 1993, Petanidou et al. 1995). And gen-
a few sites and times may cause us to misinterpret past eralization of pollinators need not imply resistance to
and present selection (Herrera 1995). Thus floral fea- perturbation, because it does not dictate generalization
tures that seem to be an evolutionary response to one of plants (or vice versa). An example may be provided
pollinator may in fact reflect a more diverse pollination by the honey bee, Apis mellifera, which is unusual in
history. A similar point can be made about contributions being a super-generalist. Deliberate introductions of
to genetic differentiation of plants. Although pollinators this species into Northern Europe, the Americas, and
are thought to foster local differentiation within plant Australia appear in some cases to have harmed or dis-
June 1996 IS POLLINATION GENERALIZED? 1055

placed native pollinators (Roubik 1978, Sugden and Beattie, A. J. 1971. Pollination mechanisms in Viola. New
Phytologist 70:343-360.
Pyke 1991, Evertz 1993, Paton 1993, Aizen and Fein-
Bennett, A. T. D., and I. C. Cuthill. 1994. Ultraviolet vision
singer 1994). If so, the resulting pollination systems in birds: what is its function? Vision Research 34:1471-
are simplified because greater specialization and re- 1478.
liance on honey bees has been forced on the plants. Bernard, G. D., and D. G. Stavenga. 1979. Spectral sensi-
Such simplified systems may be especially vulnerable tivities of retinular cells measured in intact, living flies by
an optical method. Journal of Comparative Physiology A
to further change. 134:95-107.
Acknowledgments Boggs, C. L. 1987. Ecology of nectar and pollen feeding
The National Science Foundation (grants BSR 8313522 in Lepidoptera. Pages 369-391 in F. J. Slansky and J. G.
and BSR 8905808) and the University of California, Riv- Rodriguez, editors. Nutritional ecology of insects, mites
erside Academic Senate, provided financial support, and A. and spiders. John Wiley & Sons, New York, New York,
J. Beattie and Macquarie University provided facilities dur- USA.
ing writing. Thanks to J. Ackerman, W. S. Armbruster, A. Bohart, G. E., and N. N. Youssef. 1976. The biology and
Brody, D. Campbell, D. Cornejo, T. de Jong, A. Erhardt, P. behavior of Evyleus galpinsiae. Wasmann Journal of Bi-
Feinsinger, A. Gumbert, J. Joyner, J. Kunze, M. Kwak, R. ology 34: 185-234.
Menzel, M. Mesler, G. Nabhan, J-M. Olesen, D. Paton, A. Bronstein, J. L. 1995. The plant-pollinator landscape. Pages
Prather, L. Real, M. Westoby, and R. Whelan for ideas, dis- 257-288 in L. Hansson, L. Fahrig, and G. Merriam, edi-
cussion, and assistance, and toW. S. Armbruster for pointing tors. Mosaic landscapes and ecological processes. Chap-
out the distinction between laws and trends. man & Hall, London, England.
Bronstein, J. L., P.-H. Gouyon, C. Gliddon, F. Kjellberg, and
Literature Cited G. Michaloud. 1990. The ecological consequences of
Absy, M. L., J. M. F. Camargo, W. E. Kerr, and I. P. De flowering asynchrony in monoecious figs: a simulation
Andrade Miranda. 1984. Especies de plantas visitadas por study. Ecology 71:2145-2156.
Meliponinae (Hymenoptera: Apoidea), para coleta de Bullock, S. H., C. Martinez del Rio, and R. Ayala. 1989.
p6Ien na regiao do medio Amazonas. Revista Brasileira Bee visitation rates to trees of Prockia crucis differing in
de Biologia 44:227-273. flower number. Oecologia 78:389-393.
Ackerman, J. D. 1985. Euglossine bees and their nectar Calder, W. A., N. M. Waser, S. M. Hiebert, D. W. Inouye,
hosts. Pages 225-234 in W. G. D' Arcy and M.D. Correa and S. Miller. 1983. Site-fidelity, longevity, and popu-
A., editors. The botany and natural history of Panama. lation dynamics of broad-tailed hummingbirds: a ten-year
Missouri Botanical Garden, Saint Louis, Missouri, USA. study. Oecologia 56:359-364.
Adams, D. E., W. E. Perkins, and J. R. Estes. 1981. Pol- Campbell, D. R. 1986. Predicting plant reproductive success
lination systems in Paspalum dilatatum Poir. (Poaceae): from models of competition for pollination. Oikos 47:257-
an example of insect pollination in a temperate grass. 266.
American Journal of Botany 68:389-394. Campbell, D. R., and J. L. Dooley. 1992. The spatial scale
Aizen, M. A., and P. Feinsinger. 1994. Habitat fragmenta- of genetic differentiation in a hummingbird-pollinated
tion, native insect pollinators, and feral honey bees in plant: comparison with models of isolation-by-distance.
Argentine "Chaco Serrano." Ecological Applications 4: American Naturalist 139:735-748.
378-392. Campbell, D. R., N. M. Waser, M. V. Price, E. A. Lynch,
Aker, C. L. 1982. Spatial and temporal dispersion patterns and R. J. Mitchell. 1991. A mechanistic analysis of phe-
of pollinators and their relationship to the flowering strat- notypic selection: pollen export and flower corolla width
egy of Yucca whipplei (Agavaceae). Oecologia 54:243- in lpomopsis aggregata. Evolution 45:1458-1467.
252. Cane, J. H., and J. A. Payne. 1988. Foraging ecology of
Anderson, A. B., W. L. Overal, and A. Henderson. 1988. the bee Habropoda laboriosa (Hymenoptera: Anthopho-
Pollination ecology of a forest-dominant palm ( Orbignya ridae), an oligolege of blueberries (Ericaceae: Vaccinium)
phalerata Mart.) in northern Brazil. Biotropica 20: 192- in the southeastern United States. Annals of the Ento-
205. mological Society of America 81:419-427.
Armbruster, W. S. 1990. Estimating and testing the shapes Cane, J. H., and J. A. Payne. 1993. Regional, annual, and
of adaptive surfaces: the morphology and pollination of seasonal variation in pollinator guilds: intrinsic traits of
Dalechampia blossoms. American Naturalist 135:14-31. bees (Hymenoptera: Apoidea) underlie their patterns of
Armbruster, W. S., and A. L. Herzig. 1984. Partitioning and abundance at Vaccinium ashei (Ericaceae). Annals of the
sharing of pollinators by four sympatric species of Dale- Entomological Society of America 86:577-588.
champia (Euphorbiaceae) in Panama. Annals of the Mis- Carpenter, F. L. 1989. Pollen-transfer efficiency compen-
souri Botanical Garden 71: 1-16. sates for pollinator crashes in a specialized bird-pollinated
Baker, H. G. 1963. Evolutionary mechanisms in pollination plant. In H. Ouellet, editor. Proceedings International Or-
biology. Science 139:877-883. nithological Congress XIX:537-548.
Baker, H. G., and P. D. J. Hurd. 1968. Intrafloral ecology. Chittka, L. 1992. The colour hexagon: a chromaticity dia-
Annual Review of Entomology 13:385-414. gram based on photoreceptor excitations as a generalized
Balderrama, N. M., L. 0. Almeida de B., and J. A. Nunez. representation of colour opponency. Journal of Compar-
1992. Metabolic rate during foraging in the honeybee. ative Physiology A 170:533-543.
Journal of Comparative Physiology B 162:440-447. Chittka, L., W. Beier, H. Hertel, E. Steinmann, and R. Men-
Baumberger, R. 1987. Floral structure, coloration, and evo- zel. 1992. Opponent coding is a universal strategy to eval-
lution of bird-pollinated plants: correlation with functional uate the photoreceptor inputs in Hymenoptera. Journal of
traits in nectarivorous birds. Dissertation. Universitiit Zu- Comparative Physiology A 170:545-563.
rich, Zurich, Switzerland. Chittka, L., M. Hoffmann, and R. Menzel. 1988. Discrim-
1056 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No. 4

ination of UV-green patterns in honey bees. Page 218 in Dukas, R., and L. A. Real. 1991. Learning foraging tasks
N. Elsner and F. G. Barth, editors. Sense organs. Pro- by bees: a comparison between social and solitary species.
ceedings of the 18th Gottingen neurobiology conference. Animal Behaviour 42:269-276.
Georg Thieme, Stuttgart, Germany. Dukas, R., and L. A. Real. 1993. Learning constraints and
Chittka, L., and K. Lunau. 1992. Color coding and innate floral choice behaviour in bumble bees. Animal Behaviour
preferences for flower color patterns in bumblebees. Page 46:637-644.
298 in N. Elsner and D. W. Richter, editors. Rhythmo- Dukas, R., and A. Shmida. 1989. Correlation between the
genesis in neurons and networks. Proceedings of the 20th color, size and shape of Israeli crucifer flowers and rela-
Gottingen neurobiology conference. Georg Thieme, Stutt- tionships to pollinators. Oikos 54:281-286.
gart, Germany. Dukas, R., and N. M. Waser. 1994. Categorization of food
Chittka, L., and R. Menzel. 1992. The evolutionary adap- types enhances foraging performance of bumble bees. An-
tation of flower colours and the insect pollinator's colour imal Behaviour 48: l 00 l-1 006.
vision. Journal of Comparative Physiology A 170:171- Eickwort, G. C., and K. R. Eickwort. 1969. Aspects of the
181. biology of Costa Rican halictid bees. I. Agapostemon na-
Chittka, L., A. Shmida, N. Troje, and R. Menzel. 1994. sutus (Hymenoptera: Halictidae). Journal of the Kansas
Ultraviolet as a component of flower reflections, and the Entomological Society 42:421-452.
colour perception of hymenoptera. Vision Research 34: Ellstrand, N. C. 1992. Gene flow by pollen: implications
1489-1508. for plant conservation genetics. Oikos 63:77-86.
Chittka, L., A. Shmida, M. Vorobyev, and R. Menzel. 1993. Erhardt, A. 1993. Pollination of the edelweiss, Leontopo-
Bee colour vision-the optimal system for the discrimi- dium alpinum. Botanical Journal of the Linnean Society
nation of flower colours? Pages 211-218 inK. Wiese, S. 111:229-240.
G. Gribakin, A. V. Popov, and G. Renninger, editors. Sen- Eriksson, 0., and B. Bremer. 1992. Pollination systems,
sory systems of arthropods. Birkhliuser, Berlin, Germany. dispersal modes, life forms, and diversification rates in
Clements, R. E., and F. L. Long. 1923. Experimental pol- angiosperm families. Evolution 46:258-266.
lination. An outline of the ecology of flowers and insects. Estes, J. R., and R. W. Thorp. 1975. Pollination ecology of
Carnegie Institute of Washington, Publication number 336, Pyrrhopappus carolinianus (Compositae). American Jour-
Washington, D.C., USA. nal of Botany 62:148-159.
Collins, B. G., and D. C. Paton. 1989. Consequences of Evertz, S. 1993. Untersuchungen ziir interspezifischen
differences in body mass, wing length and leg morphology Konkurrenz zwischen Honigbienen (Apis mellifera L.) und
for nectar-feeding birds. Australian Journal of Ecology 14: solitliren Wildbienen (Hymenoptera, Apoidea). Disserta-
269-289. tion. Rheinishe-Westfalische Technische Hochschule, Aa-
Cox, P. A. 1983. Extinction of the Hawaiian avifauna re- chen, Germany.
sulted in a change of pollinators for the ieie, Freycinetia Feinsinger, P. 1978. Ecological interactions between plants
arborea. Oikos 41:195-199. and hummingbirds in a successional tropical community.
Crepet, W. L. 1983. The role of insect pollination in the Ecological Monographs 48:269-287.
evolution of the angiosperms. Pages 29-50 in L. Real, - - - . 1983. Coevolution and pollination. Pages 285-319
editor. Pollination biology. Academic Press, Orlando, in D. J. Futuyma and M. Slatkin, editors. Coevolution.
Florida, USA. Sinauer, Sunderland, Massachusetts, USA.
- - - . 1984. Advanced (constant) insect pollination mech- - - - . 1987. Approaches to nectarivore-plant interactions
anisms: pattern of evolution and implications vis-a-vis an- in the new world. Revista Chilena de Historia Natural 60:
giosperm diversity. Annals of the Missouri Botanical Gar- 285-319.
den 71:607-630. - - - . 1990. Interacciones entre plantas y colibries en sel-
Cross, E. A., and G. E. Bohart. 1960. The biology of Nomia vas tropicales. Boletin de Ia Academia Nacional de Cien-
(Epinomia) triangulifera with comparative notes on other cias (Cordoba, Argentina) 59:31-54.
species of Nomia. University of Kansas Science Bulletin Feinsinger, P., and R. K. Colwell. 1978. Community orga-
41:761-792. nization among neotropical nectar-feeding birds. Ameri-
Cruden, R. W. 1972. Pollination biology of Nemophila men- can Zoologist 18:779-795.
zeisii (Hydrophyllaceae) with comments on the evolution Fishbein, M., and D. L. Venable. 1996. Diversity and tem-
of oligolectic bees. Evolution 26:373-389. poral change in the effective pollinators of Asclepias tu-
Delpino, F. 1868-1875. Ulteriori osservazione sulla dico- berosa. Ecology 77:1061-1073.
gamia nel regno vegetale. Atti Della Societa Italiana di Fleming, R. C. 1970. Food plants of some adult sphinx
Scienze Naturali e dei Museo Civico di Storia et Natuturale moths (Lepidoptera: Sphingidae). Michigan Entomologist
di Milano 11, 12. 3:17-23.
Dobson, H. E. M. 1987. Role of flower and pollen aromas Fore!, A. 1910. Das Sinnesleben der Insekten. E. Reinhardt,
in host-plant recognition by solitary bees. Oecologia 72: Miinchen, Germany.
618-623. Forster, P. I. 1994. Diurnal insects associated with the flow-
Dodd, R. J., and Y. B. Linhart. 1994. Reproductive con- ers of Gomphocarpus physocarpus E. Mey. (Asclepiada-
sequences of interactions between Yucca glauca (Agava- ceae), an introduced weed in Australia. Biotropica 26:
ceae) and Tegeticula yuccasella (Lepidoptera) in Colora- 214-217.
do. American Journal of Botany 81:815-825. Free, J. B. 1970. Insect pollination of crops. Academic
Dreisig, H. 1995. Ideal free distributions of nectar foraging Press, London, England.
bumblebees. Oikos 72: 161-172. Fukushi, T. 1994. Colour perception of single and mixed
Duffield, G. E., R. C. Gibson, P. M. Gilhooly, A. J. Hesse, monochromatic lights in the blowfly Lucilia cuprina. Jour-
C. R. Inkley, F. S. Gilbert, and C. J. Barnard. 1993. Choice nal of Comparative Physiology A 175:15-22.
of flowers by foraging bees (Apis mellifera): possible mor- Galil, J. 1977. Fig biology. Endeavour (New Series) 1:52-
phological cues. Ecological Entomology 18: 191-197. 56.
June 1996 IS POLLINATION GENERALIZED? 1057

Gathmann, A., H. J. Greiler, and T. Tscharntke. 1994. Trap- linators: flowering plants are more than their flowers. Ecol-
nesting bees and wasps colonizing set-aside fields: suc- ogy 76:1516-1524.
cession and body size, management by cutting and sowing. - - - . 1996. Floral traits and plant adaptation to insect
Oecologia 98:8-14. pollinators: a de vii's advocate approach. InS. C. H. Barrett
Gilbert, F. S. 1981. Foraging ecology of hoverflies: mor- and D. G. Lloyd, editors. Floral biology. Chapman & Hall,
phology of the mouthparts in relation to feeding on nectar New York, New York, USA, in press.
and pollen in some common urban species. Ecological Hilborn, R., and S.C. Stearns. 1982. On inference in ecol-
Entomology 6:245-262. ogy and evolutionary biology: the problem of multiple
Gilbert, L. E., and P. H. Raven. 1975. Coevolution of an- causes. Acta Biotheoretica 31:145-164.
imals and plants. University of Texas Press, Austin, Texas, Hodges, S. A., and M. L. Arnold. 1994. Columbines: a
USA. geographically widespread species flock. Proceedings of
Giurfa, M., J. Nunez, L. Chittka, and R. Menzel. 1995. Col- the National Academy of Sciences (USA) 91:5129-5132.
our preferences of flower-naive honeybees. Journal of Horvitz, C. C., and D. W. Schemske. 1990. Spatiotemporal
Comparative Physiology A 177:247-259. variation in insect mutualists of a neotropical herb. Ecol-
Goldsmith, T. H., and K. M. Goldsmith. 1979. Discrimi- ogy 71: 1085-1097.
nation of colors by the Black-chinned Hummingbird (Ar- Howe, H. F. 1984. Constraints on the evolution of mutu-
chilochus alexandri). Journal of Comparative Physiology alism. American Naturalist 123:764-777.
A 130:209-220. Howell, D. J. 1979. Flock foraging in nectar-feeding bats:
Grant, V. 1949. Pollination systems as isolating mechanisms advantages to the bats and to the host plants. American
in angiosperms. Evolution 3:82-97. Naturalist 114:23-49.
- - - . 1994. Modes and origins of mechanical and etho- Inouye, D. W. 1980. The effect of proboscis and corolla
logical isolation in angiosperms. Proceedings of the Na- tube lengths on patterns and rates of flower visitation by
tional Academy of Sciences (USA) 91:3-10. bumblebees. Oecologia 45:197-201.
Grant, V., and K. A. Grant. 1965. Flower pollination in the Janzen, D. H. 1980. When is it coevolution? Evolution 34:
phlox family. Columbia University Press, New York, New 611-612.
York, USA. Jordano, P. 1987. Patterns of mutualistic interactions in pol-
Grant, V., and E. J. Temeles. 1992. Foraging ability of Ru- lination and seed dispersal: connectance, dependence
fous Hummingbirds on hummingbird flowers and hawk- asymmetries, and coevolution. American Naturalist 129:
moth flowers. Proceedings of the National Academy of 657-677.
Sciences (USA) 89:9400-9404. Keeton, W. T., and J. L. Gould. 1993. Biological science.
Greggers, U., and R. Menzel. 1993. Memory dynamics and Fifth edition. W. W. Norton, New York, New York, USA.
foraging strategies of honeybees. Behavioral Ecology and Kephart, S. R. 1983. The partitioning of pollinators among
Sociobiology 32:17-29. three species of Asclepias. Ecology 64:120-133.
Gregory, D. P. 1963. Hawkmoth pollination in the genus Kevan, P. G. 1978. Floral coloration, its colorimetric anal-
Oenothera. Aliso 5:357-384. ysis and significance in anthecology. Pages 51-78 in A.
Hainsworth, F. R., and L. L. Wolf. 1972. Power for hovering J. Richards, editor. The pollination of flowers by insects.
flight in relation to body size in hummingbirds. American Academic Press, London, England.
Naturalist 106:589-596. - - - . 1991. Pollination: keystone process in sustainable
Hainsworth, F. R., and L. L. Wolf. 1976. Nectar character- global productivity. Acta Horticulturae 288:103-109.
istics and food selection by hummingbirds. Oecologia 25: Kevan, P. G., and H. G. Baker. 1983. Insects as flower
101-113. visitors and pollinators. Annual Review of Entomology
Harder, L. D. 1990. Behavioral responses by bumble bees 28:407-453.
to variation in pollen availability. Oecologia 85:41-47. Kiester, A. R., R. Lande, and D. W. Schemske. 1984. Models
Haslett, J. R. 1989. Adult feeding by holometabolous in- of coevolution and speciation in plants and their polli-
sects: pollen and nectar as complementary nutrient sources nators. American Naturalist 124:220-243.
for Rhingia campestris (Diptera: Syrphidae). Oecologia Kislev, M. E., Z. Kraviz, and J. Lorch. 1972. A study of
81:361-363. hawkmoth pollination by a palynological analysis of the
Heinrich, B. 1976. The foraging specializations of individ- proboscis. Israel Journal of Botany 21:57-75.
ual bumblebees. Ecological Monographs 46:105-128. Klinger, T., P. E. Arriola, and N.C. Ellstrand. 1992. Crop-
- - - . 1979. Bumblebee economics. Harvard University weed hybridization in radish (Raphanus sativus): effects
Press, Cambridge, Massachusetts, USA. of distance and population size. American Journal of Bot-
Heinrich, B., P. R. Mudge, and P. G. Deringis. 1977. Lab- any 79:1431-1435.
oratory analysis of flower constancy in foraging bumble- Knuth, P. 1908. Handbook of flower pollination. Volume II.
bees: Bombus ternarius and B. terricola. Behavioral Ecol- Observations on flower pollination made in Europe and
ogy and Sociobiology 2:247-265. the Arctic regions on species belonging to the natural or-
Heithaus, E. R. 1979. Community structure of neotropical ders Ranunculaceae to Stylidieae. Clarendon Press, Ox-
flower-visiting bees and wasps: diversity and phenology. ford, England.
Ecology 60: 190-202. - - - . 1909. Handbook of flower pollination. Volume III.
Herrera, C. M. 1987. Components of pollinator "quality": Observations on flower pollination made in Europe and
comparative analysis of a diverse insect assemblage. Oikos the Arctic regions on species belonging to the natural or-
50:79-90. ders Goodenovieae to Cycadeae. Clarendon Press, Oxford,
- - - . 1988. Variation in mutualisms: the spatiotemporal England.
mosaic of a pollinator assemblage. Biological Journal of Kunin, W. E. 1993. Sex and the single mustard: population
the Linnean Society 35:95-125. density and pollinator behavior effects on seed-set. Ecol-
- - - . 1995. Microclimate and individual variation in pol- ogy 74:2145-2160.
1058 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No.4

Kwak, M. M. 1994. Duifkruid en haar bezoekers. Natura learctic anthidiine bees (Hymenoptera: Apoidea: Mega-
91:5-8. chilidae ). Ecological Monographs 66:235-257.
Kwak, M. M., C. van den Brand, P. Kremer, and E. Boer- Nabhan, G. P., and T. Fleming. 1993. The conservation of
richter. 1991. Visitation, flight distances and seed set in new world mutualisms. Conservation Biology 7:457-459.
populations of the rare species Phyteuma nigrum (Cam- Nilsson, L. A., L. Jonsson, L. Talison, and E. Randrianjo-
panulaceae). Acta Horticulturae 288:303-307. hany. 1987. Angraecoid orchids and hawkmoths in central
Lacey, E. P., L. Real, J. Antonovics, and D. G. Heckel. 1983. Madagascar: specialized pollination systems and gener-
Variance models in the study of life histories. American alist foragers. Biotropica 19:310-318.
Naturalist 122:114-131. Parker, F. D. 1981. How efficient are bees at pollinating
Lasiewski, R. C. 1963. Oxygen consumption of torpid, rest- sunflowers? Journal of the Kansas Entomological Society
ing, active, and flying hummingbirds. Physiological Zo- 54:61-67.
ology 36:122-140. Patel, A., M. Hossaert-McKey, and D. McKey. 1993. Ficus-
Levin, M.D., and M. H. Haydak. 1957. Comparative value pollinator research in India: past, present and future. Cur-
of different pollens in the nutrition of Osmia lignaria Say. rent Science 65:243-253.
Proceedings of the Tenth International Congress of En- Paton, D. C. 1985. Food supply, population structure, and
tomology 4: I 079-1084. behavior of New Holland Honeyeaters Phylidonyris no-
Lindauer, M. 1963. Allgemeine Sinnesphysiologie. Orien- vaehollandiae in woodland near Horsham, Victoria. Pages
tierung im Raum. Fortschritte der Zoologie 16:58-140. 219-230 in A. Keast, H. F. Recher, H. Ford and D. Saun-
Lord, J. M. 1991. Pollination and seed dispersal in Frey- ders, editors. Birds of eucalypt forests and woodlands:
cinetia baueriana, a dioecious Iiane that has lost its bat ecology, conservation, management. Surrey Beatty &
pollinator. New Zealand Journal of Botany 29:83-86. Sons, Sydney, Australia.
Lunau, K. 1993. Angeborene und erlernte Bliitenerkennung - - - . 1993. Honeybees in the Australian environment.
bei Insekten-ein entdecktes Geheimnis der Natur. Bio- BioScience 43:95-103.
logie in unserer Zeit 23:48-54. Paton, D. C., and B. G. Collins. 1989. Bills and tongues of
Martinez, N. D. 1992. Constant connectance in community nectar-feeding birds: a review of morphology, function and
food webs. American Naturalist 139:1208-1218. performance, with intercontinental comparisons. Austra-
Martinez del Rio, C. 1990. Sugar preferences of humming- lian Journal of Ecology 14:473-506.
birds: the influence of subtle chemical differences on food Pedro, S. R. M. 1993. Sobre as abelhas (Hymenoptera, Apo-
choice. Condor 92:1022-1030. idea) em urn ecossistema de Cerrado. Thesis. University
Martinez del Rio, C., and L. E. Eguiarte. 1987. Bird visi- of Sao Paulo, Riberao Preto, Brazil.
tation to Agave salmiana: comparisons among humming- Peitsch, D., A. Fietz, H. Hertel, J. de Souza, D. F. Ventura,
birds and perching birds. Condor 89:357-363. and R. Menzel. 1992. The spectral input systems of hy-
Mazokhin-Porshnyakov, G. A., 0. G. Volkov, and S. A. Se- menopteran insecta and their receptor-based colour vision.
menova. 1984. A study of color vision in flies by classical Journal of Comparative Physiology A 170:23-40.
conditioning, Protophormia terrae-novae taken as an ex- Pellmyr, 0., and J. N. Thompson. 1992. Multiple occur-
ample. Journal of General Biology (Moscow) 45:653-659. rences of mutualism in the yucca moth lineage. Proceed-
Menzel, R. 1967. Untersuchungen zum Erlernen von Spek- ings of the National Academy of Sciences (USA) 89:2927-
tralfarben durch die Honigbiene (Apis mellifica). Zeit- 2929.
schrift fiir vergleichende Physiologie 56:22-62. Petanidou, T., J. C. M. den Nijs, and J. G. B. Oostermeijer.
- - - . 1985. Learning in honey bees in an ecological and
1995. Pollination ecology and constraints on seed set of
behavioral context. Pages 55-74 in B. Holldobler and M.
the rare perennial Gentiana cruciata L. in the Netherlands.
Lindauer, editors. Experimental behavioral ecology and
Acta Botanica Neerlandica 44:55-74.
sociobiology. Sinauer, Sunderland, Massachusetts, USA.
Petanidou, T., and W. N. Ellis. 1993. Pollinating fauna of
Menzel, R., and W. Backhaus. 1991. Colour vision in in-
a phryganic ecosystem: composition and diversity. Bio-
sects. Pages 262-293 in P. Gouras, editor. Vision and vi-
diversity Letters 1:9-22.
sual dysfunction. Volume 6. The perception of colour.
MacMillan, Houndsmills, England. Pettersson, M. W. 1991. Pollination by a guild of fluctuating
Michener, C. D. 1979. Biogeography of the bees. Annals moth populations: option for unspecialization in the blad-
of the Missouri Botanical Garden 66:277-347. der campion, Silene vulgaris. Journal of Ecology 79:591-
Michener, C. D., and C. W. Rettenmeyer. 1956. The ethology 604.
of Andrena erythromii with comparative data on other spe- Pleasants, J. M. 1981. Bumblebee responses to variation in
cies (Hymenoptera: Andrenidae). University of Kansas nectar availability. Ecology 62:1648-1661.
Science Bulletin 37:147-178. Pleasants, J. M., and N. M. Waser. 1985. Bumblebee for-
Minckley, R. L., W. T. Wcislo, D. Yanega, and S. L. Buch- aging at a "hummingbird" flower: reward economics and
mann. 1994. Behavior and phenology of a specialist bee floral choice. American Midland Naturalist 114:283-291.
(Dieunomia) and sunflower (Helianthus) pollen availabil- Primack, R. B. 1979. Reproductive biology of Discaria.
ity. Ecology 75:1406-1419. New Zealand Journal of Botany 17:9-13.
Mitchell, W. A. 1989. Informational constraints on opti- Pyke, G. H. 1984. Optimal foraging theory: a critical re-
mally foraging hummingbirds. Oikos 55:145-154. view. Annual Review of Ecology and Systematics 15:523-
Moldenke, A. R. 1979. The role of host-plant selection in 575.
bee speciation processes. Phytologia 43:433-460. - - - . 1990. Apiarists versus scientists: a bittersweet case.
Motten, A. F., D. R. Campbell, D. E. Alexander, and H. L. Australian Natural History 23:386-392.
Miller. 1981. Pollination effectiveness of specialist and Ramirez, W. B. 1970. Host specificity of fig wasps (Agaon-
generalist visitors to a North Carolina population of Clay- idae). Evolution 24:680-691.
tonia virginica. Ecology 62:1278-1287. Real, L. A. 1980. On uncertainty and the law of diminishing
Miiller, A. 1996. Host-plant specialization in western pa- returns in evolution and behavior. Pages 37-64 in J. E. R.
June 1996 IS POLLINATION GENERALIZED? 1059

Staddon, editor. Limits to action. Academic Press, New - - - . 198lb. Notes on the reproductive biology of Zig-
York, New York, USA. adenus paniculatus, a toxic range plant. Great Basin Nat-
Real, L. A., and S. Ellner. 1992. Risk and reproductive uralist 41:427-430.
effort. Ecology 73:1227-1236. Tepedino, V. J., and N. L. Stanton. 1980. Spatiotemporal
Renner, S. S., and J. P. Feil. 1993. Pollinators of tropical variation in phenology and abundance of floral resources
dioecious angiosperms. American Journal of Botany 80: on shortgrass prairie. Great Basin Naturalist 40:197-215.
1100-1107. Thompson, J. N. 1994. The coevolutionary process. Uni-
Robertson, C. 1928. Flowers and insects: lists of visitors to versity of Chicago Press, Chicago, Illinois, USA.
four hundred and fifty-three flowers. C. Robertson, Car- Thorp, R. W. 1969. Systematics and ecology of bees of
linville, Illinois, USA. the subgenus Diandrena (Hymenoptera: Andrenidae).
Roubik, D. W. 1978. Competitive interactions between neo- University of California Publications in Entmology 52:
tropical pollinators and Africanized honey bees. Science 1-146.
201:1030-1032. - - - . 1979. Structural, behavioral and physiological ad-
- - - . 1992. Loose niches in tropical communities: why aptations of bees (Apoidea) for collecting pollen. Annals
are there so few bees and so many trees? Pages 327-354 of the Missouri Botanical Garden 66:788-812.
in M. D. Hunter, T. Ohgushi, and P. W. Price, editors. Trelease, W. 1881. The fertilization of Salvia splendens by
Effects of resource distribution on animal-plant interac- birds. American Naturalist 15:265-269.
tions. Academic Press, San Diego, California, USA. Troje, N. 1993. Spectral categories in the learning behaviour
Rozen, J. G. 1958. Monographic study of the genus No- of blowflies. Zeitschrift fiir Naturforschung 48C:96-104.
madopsis Ashmead. University of California Publications van der Pijl, L. 1961. Ecological aspects of flower evolution.
in Entomology 15:1-202. II. Zoophilous flower classes. Evolution 15:44-59.
Schemske, D. W. 1983. Limits to specialization and coevo- von Helversen, 0. 1972. Zur spektralen Unterschiedsemp-
lution in plant-animal mutualisms. Pages 67-110 in M. findlichkeit der Honigbiene. Journal of Comparative Phys-
H. Nitecki, editor. Coevolution. University of Chicago iology 80:439-472.
Press, Chicago, Illinois, USA. Walker, B. H. 1992. Biodiversity and ecological redundan-
Schoener, T. W. 1971. Theory of feeding strategies. Annual cy. Conservation Biology 6:18-23.
Review of Ecology and Systematics 2:369-404. Warnke, E., U. Terndrup, V. Michelsen, and A. Erhardt.
Schremmer, F. 1953. Bliitenbiologische Beobachtungen an 1993. Flower visitors to Saxifraga hirculus in Switzerland
Labiaten. Nektar- und Pollendiebstahl. Osterreichische and Denmark, a comparative study. Botanica Helvetica
Botanische Zeitschrift 100:8-24. 103:141-147.
Scott, P. E., S. L. Buchmann, and M. K. O'Rourke. 1993. Waser, N. M. 1978. Interspecific pollen transfer and com-
Evidence for mutualism between a flower-piercing car- petition between co-occuring plant species. Oecologia 36:
penter bee and ocotillo: use of pollen and nectar by nesting 223-236.
bees. Ecological Entomology 18:234-240. - - - . 1979. Pollinator availability as a determinant of
Sommeijer, M. J., G. A. DeRooy, W. Punt, and L. L. M. flowering time in ocotillo (Fouquieria splendens). Oeco-
Bruijn. 1983. A comparative study of foraging behaviour logia 39:107-121.
and pollen resources of various stingless bees (Hymenop- - - - . 1982. A comparison of distances flown by different
tera, Meliponinae) and honeybees (Hymenoptera, Apidae) visitors to flowers of the same species. Oecologia 55:251-
in Trinidad, West Indies. Apidologie 14:205-224.
257.
Stebbins, G. L. 1970. Adaptive radiation of reproductive - - - . 1983. The adaptive nature of floral traits: ideas and
characteristics in angiosperms. 1: Pollination mechanisms.
evidence. Pages 241-285 in L.A. Real, editor. Pollination
Annual Review of Ecology and Systematics 1:307-326.
biology. Academic Press, New York, New York, USA.
- - - . 1974. Flowering plants: evolution above the species
- - - . 1987. Spatial genetic heterogeneity in a population
level. Belknap Press of Harvard University Press, Cam-
of the montane perennial plant Delphinium nelsonii. He-
bridge, Massachusetts, USA.
redity 58:249-256.
Stephens, D. W., and J. R. Krebs. 1986. Foraging theory.
- - - . 1993. Population structure, optimal outbreeding,
Princeton University Press, Princeton, New Jersey, USA.
Strickler, K. 1979. Specialization and foraging efficiency of and assortative mating in angiosperms. Pages 173-199 in
solitary bees. Ecology 60:988-1009. N. W. Thornhill, editor. The natural history of inbreeding
Stubblefield, J. W., J. Seger, J. W. Wenzel, and M. M. Heisler. and outbreeding. Theoretical and empirical perspectives.
1993. Temporal, spatial, sex-ratio, and body-size hetero- University of Chicago Press, Chicago, Illinois, USA.
geneity of prey species taken by the beewolf Philanthus Waser, N. M., and M. V. Price. 1985. The effect of nectar
sanbornii (Hymenoptera: Sphecidae). Philosophical guides on pollinator preference. Experimental studies with
Transactions of the Royal Society of London, B 339:397- a montane herb. Oecologia 67:121-126.
423. Waser, N. M., and M. V. Price. 1990. Pollination efficiency
Stucky, J. M. 1984. Forager attraction by sympatric Ipomoea and effectiveness of bumble bees and hummingbirds vis-
hederacea and /. purpurea (Convolvulaceae) and corre- iting Delphinium nelsonii. Collectanea Botanica (Barce-
sponding forager behavior and energetics. American Jour- lona) 19:9-20.
nal of Botany 71:1237-1244. Waser, N. M., and M. V. Price. 1993. Specialisation and
Sugden, E. A., and G. H. Pyke. 1991. Effects of honey bees generalisation in pollination systems. Pages 165-169 in
on colonies of Exoneura asimillima, an Australian native G. K. Veeresh, R. Uma Shaanker, and K. N. Ganeshaiah,
bee. Australian Journal of Ecology 16:171-181. editors. Pollination in Tropics. International Union for the
Tepedino, V. J. 1981 a. The pollination efficiency of the Study of Social Insects, Bangalore, India.
squash bee (Peponapis pruinosa) and the honey bee (Apis Waser, N. M., and L. A. Real. 1939. Effective mutualism
mellifera) on summer squash ( Cucurbita pepo). Journal of between sequentially flowering plant species. Nature 281:
the Kansas Entomological Society 54:359-377. 670-672.
1060 NICKOLAS M. WASER ET AL. Ecology, Vol. 77, No.4

Weiss, M. R. 1991. Floral colour changes as cues for pol- Wolf, L. L., and F. R. Hainsworth. 1972. Energetics of for-
linators. Nature 354:227-229. aging: rate and efficiency of nectar extraction by hum-
Westerkamp, C. 1991. Honeybees are poor pollinators- mingbirds. Science 176:1351-1352.
why? Plant Systematics and Evolution 177:71-75. Wyszecki, G., and W. S. Stiles. 1982. Color science: con-
Westrich, P. 1989. Die Wildbienen Baden-Wiirttembergs. cepts and methods, quantitative data and formulae. John
Volumes I and II. Ulmer Verlag, Stuttgart, Germany. Wiley & Sons, New York, New York, USA.
Wiebes, J. T. 1979. Co-evolution of figs and their insect
pollinators. Annual Review of Ecology and Systematics
10:1-12.