Draft version August 9, 2024

Typeset using LATEX twocolumn style in AASTeX631

Exploring the origin of cold gas and star formation in a rare population of strongly bulge-dominated
early-type Galaxies
Fujia Li ,1, 2, ∗ Enci Wang ,1, 2, † Ming Zhu ,3, 4, 5 Ying-jie Peng,6, 7 Jing Wang ,6 Chuan-Peng Zhang ,3, 4, 5
Zesen Lin ,8 Yu Rong ,1, 2 Hongxin Zhang ,1, 2 and Xu Kong 1, 2, ‡
1 Departmentof Astronomy, University of Science and Technology of China, Hefei 230026, China
2 School
of Astronomy and Space Science, University of Science and Technology of China, Hefei 230026, China
arXiv:2408.04256v1 [astro-ph.GA] 8 Aug 2024

3 National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, People’s Republic of China
4 Guizhou Radio Astronomical Observatory, Guizhou University, Guiyang 550000, People’s Republic of China
5 CAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, People’s Republic

of China
6 Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China
7 Department of Astronomy, School of Physics, Peking University, Beijing 100871, People’s Republic of China
8 Department of Physics, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong S.A.R., China

ABSTRACT
We analyze the properties of a rare population, the strongly bulge-dominated early-type galaxies
(referred to as sBDEs) with significant Hi gas, using the databases from the FAST All Sky Hi survey
(FASHI) and the Arecibo Legacy Fast ALFA (ALFALFA) survey. We select the sBDEs from the
Sloan Digital Sky Survey (SDSS) and cross-match with the FASHI-ALFALFA combined Hi sample,
resulting in 104 Hi-rich sBDEs. These sBDEs tend to have extremely high Hi reservoirs, which is
rare in previous studies such as ATLAS3D . 70% of the selected sBDEs are classified as quiescent
galaxies, even though they have a large Hi reservoir. We study the properties of these sBDEs from five
main aspects: stellar population, gas-phase metallicity, stacked Hi spectra, environment, and spatially
resolved MaNGA data. The majority of Hi-rich sBDEs appear to show lower gas-phase metallicity
and are located in significantly lower-density environments, suggesting an external origin for their Hi
gas. We find that star-forming sBDEs exhibit statistically higher star formation efficiency and slightly
older stellar populations compared to normal star-forming galaxies, suggesting a recent star formation
on Gyr-timescale. They also show narrower and more concentrated Hi profiles compared to control
star-forming galaxies, which may explain their higher star formation efficiency.

Keywords: Early-type galaxies; Hi gas; Star formation, galaxy evolution

1. INTRODUCTION transformation (Croton et al. 2006; Park et al. 2008;

Galaxies are typically classified into two broad Tamburri et al. 2014), and the radio-mode active galac-
types: late-type galaxies (LTGs) and early-type galaxies tic nuclei (AGNs) feedback is widely used to maintain
(ETGs), based on their morphology. ETGs, including the quiescence of star formation in hydrodynamical sim-
ellipticals and lenticulars, are commonly characterized ulations (Bower et al. 2006; Croton et al. 2006; Kurinchi-
as “red and dead” galaxies, owing to a lack of active Vendhan et al. 2023).
star formation, and dominated by an old stellar popula- Most ETGs contain significant quantities of hot gas,
tion (Strateva et al. 2001; Wetzel et al. 2012; Liu et al. but considerably fewer ETGs have been detected with
2019; Lee et al. 2023). The formation and evolution of cold gas (e.g., the atomic neutral hydrogen gas Hi,
ETGs are believed to be closely related to the multiple Sanders & Tubbs 1980; Knapp et al. 1985; Sadler et al.
mergers in their lifetime along with the morphological 2002; Grossi et al. 2009; Serra et al. 2012). The environ-
ment has been found to play an important role in it, as
ETGs in low-density environments show a higher cold
∗ E-mail: [email protected] gas detection rate, and they appear to follow different
† E-mail: [email protected]
‡
evolutionary paths compared to the ETGs in clusters
E-mail: [email protected]
2

(Governato et al. 1999; di Serego Alighieri et al. 2007; by the FASHI-ALFALFA combined sample provides us
Grossi et al. 2009; Ashley et al. 2019). The ETGs in the opportunity to statistically study the sample prop-
low-density environments may have experienced a more erties of the rare population, and to further explore the
extended star formation history or rejuvenated star for- origin of cold gas and investigate the triggers for star
mation activity (Governato et al. 1999; Fang et al. 2012; formation.
Ashley et al. 2019). The paper is organized as follows. We introduce and
There are two avenues for quenched ETGs to obtain describe our sample matching and selection from the
cold gas: internal and external mechanisms. Internal FASHI, ALFALFA, and SDSS in Section 2. Section 3
mechanisms can occur, when evolved stars within the shows the basic sample properties of our sBDEs sam-
galaxy experience stellar mass loss and this hot mate- ple from the FASHI-ALFALFA combined sample. Then
rial cools down once again (Houck & Bregman 1990; we present the results of star-forming and quiescent
Höfner & Olofsson 2018). The primary sources of ex- sBDEs separately, including stellar populations, gas-
ternal gas involve mergers with gas-rich galaxies and ac- phase metallicity, stacked Hi spectra, environments, and
cretion of cold gas directly from the environment (Lagos MaNGA observations in Section 4. We discuss the ori-
et al. 2014, 2015; Davis & Bureau 2016; Griffith et al. gin of cold gas and the possible triggers of star formation
2019; Paudel et al. 2023). The difference between the for these Hi-rich sBDEs in Section 5. Finally, we briefly
two mechanisms lies in the gas-phase metallicity and summarize our main results in Section 6. Through-
the dynamics of cold gas. The gas from former mecha- out this work, we adopt a flat ΛCDM cosmology with
nisms is metal-rich and has almost the coherent dynam- Ωm = 0.3, Ωλ = 0.7 and H0 = 70 km s−1 Mpc−1 .
ics as the stars, while the gas is typically misaligned
and metal-poor in external mechanisms, especially in 2. SAMPLE AND DATA
the case of accretion (Davis & Young 2019; Bao et al. 2.1. ALFALFA and FASHI
2022; Lee et al. 2023). Once they re-acquire gas, ETGs
ALFALFA is a blind survey with the Arecibo 305 m
have the potential to trigger star formation again (Fang
radio observatory, covering ∼7000 deg2 between 0◦ and
et al. 2012; Lee et al. 2023).
+36◦ in declination (Haynes et al. 2011, 2018). It has a
Ellipticals and lenticulars exhibit significant differ-
beam size ∼ 4′ and the average rms is about σrms = 2.0
ences in numerous aspects. In lenticulars, the presence
mJy at 10 km s−1 spectral resolution.
of the disk indicates that the kinematics are rotationally
FAST, with a diameter of 500 m and a 19-beam re-
supported, similar to spiral galaxies, while ellipticals are
ceiver, offers higher sensitivity, deeper observations, and
pressure-supported systems with slower rotation (e.g.,
higher observational efficiency in detecting Hi emission
Moran et al. 2007; Cortesi et al. 2013; Cappellari 2016).
lines than ALFALFA. FASHI aims to observe an area
In this work, we select a rare population, the strongly
of ∼22000 deg2 between −14◦ and +66◦ in declination,
bulge-dominated ETGs (but more consistent with the
with the expectation of detecting more than 100000
distribution of ellipticals in three structural parameters)
Hi sources (Zhang et al. 2023a). The beam size of
with Hi detection from two large Hi sample databases,
FAST is slightly smaller than that of ALFALFA, ap-
to explore the origin of Hi gas, the basic Hi properties,
proximately 2.9′ at 1.42 GHz, and the average rms is
and the possible triggers for star formation in this rare
about σrms = 1.5 mJy with a 6.4 km s−1 spectral reso-
population.
lution for FASHI.
The Arecibo Legacy Fast ALFA (ALFALFA; Gio-
The parent Hi samples used in this paper are from
vanelli et al. 2005; Haynes et al. 2011) is a blind Hi
the 100% ALFALFA extragalactic Hi source catalog
survey covering ∼7000 deg2 of northern sky. Consid-
(Haynes et al. 2018) and an early version of the first re-
ering that the Hi detection rate of ETGs is very low,
lease of FASHI catalog, which contains 31502 and 40343
ALFALFA alone could not enable us to obtain a rea-
galaxies, respectively. It is worth noting that there are
sonably large sample. With the continuous drift scan
some differences between the catalog we’re using and
observations of Five-hundred-meter Aperture Spherical
the final first release of the FASHI catalog (41741 in the
radio Telescope (FAST; Nan et al. 2011; Jiang et al.
complete first release catalog), but it doesn’t change the
2019; Qian et al. 2020), the FAST All Sky Hi survey
final results of this work.
(FASHI; Zhang et al. 2023a) aims to obtain more than
100000 Hi sources. Therefore, FASHI could complement 2.2. SDSS
the ALFALFA survey and potentially provide a substan-
We select the parent sample from SDSS DR7 with
tial quantity of Hi data that corresponds to that of the
both photometric and spectroscopic measurements, sim-
ALFALFA and even more. The large solid angle covered
ilar to the sample constructed in Peng et al. (2010,
 3

2012, 2015). The parent sample is obtained directly We aim to study the properties of strongly bulge-
from the “PhotoObj View” and “SpecObj View” in the dominated Hi-rich galaxies and to understand the origin
SDSS CasJobs site1 . We require the galaxies to have of their Hi gas. We adopt the following method to select
clean photometry, clean spectra, and reliable spectro- our sample:
scopic redshift measurements with redshift z < 0.1, to (i) We cross-match our Hi sample and optical sam-
achieve a complete matching with the combined Hi sam- ple with that of the morphological catalog described in
ple. We refer to it as the SDSS optical sample, resulting Domı́nguez Sánchez et al. (2018, hereafter DS18) to ob-
in 266220 galaxies. tain T-type for each galaxy. In short, the DS18 mor-
We collect stellar mass, star formation rate (SFR), phological catalog is based on Convolutional Neural Net-
and flux measurements of strong emission lines (Hα, Hβ, works to classify ∼670000 galaxies of SDSS DR7 in Hub-
[Oii]λλ3726, 3729, [Nii]λ6585, [Sii]λλ6717, 6731 and ble sequence T-type and Galaxy Zoo 2 classification
[Oiii]λ5007) from the MPA-JHU catalog2 (Kauffmann scheme. Furthermore, they train an additional model
et al. 2003; Brinchmann et al. 2004). All of the emission to classify ellipticals and lenticulars.
lines have been corrected for extinction using the Fitz- (ii) We use morphological T-type ≤ 0 to select ETGs
patrick (1999) Milky Way extinction curve, assuming from the total sample and use the additional classifica-
RV =3.1 and the Balmer decrement Hα/Hβ=2.86 un- tion PS0 < 0.5 to further exclude the lenticulars that
der the case-B recombination (Baker & Menzel 1938). may be dominated by disk.
Besides, we also retrieve the equivalent width of Hα (iii) Considering the potential errors in machine learn-
emission (EW(Hα)), the Lick index of Hδ absorption ing classification and the limitations of SDSS sensitiv-
(EW(HδA )), and 4000 Å break (Dn 4000) from the MPA- ity, we apply an additional visual morphological classi-
JHU catalog, which are typically used to diagnose the fication using the deeper images from the DESI Legacy
recent star formation and star formation history on dif- Imaging Surveys3 (Dey et al. 2019). In this step, we
ferent timescales (e.g., Bruzual & Charlot 2003; Kauff- exclude 56 galaxies that are misclassified, influenced by
mann et al. 2003; Le Borgne et al. 2006; Fumagalli et al. stars, or have a small apparent axis ratio.
2012; Wang et al. 2018; Wang & Lilly 2020). It is worth noting that, although these galaxies are
We collect three structural parameters: concentration classified as ellipticals in DS18, considering their abun-
index (C-index, defined as the ratio of r50 and r90 , that dant Hi reservoir, we expect them to show some different
represents the radii enclosing 50% and 90% of the flux properties or morphology as usual ellipticals. In other
in r-band Petrosian flux), Sérsic n index, and bulge-to- words, some Hi-rich ellipticals in our sample are not en-
total mass ratio (B/T). We obtain r50 and r90 from the tirely smooth, but show some low surface brightness fea-
SDSS catalog directly, and the Sérsic index n is taken tures (may be caused by tidal interactions or mergers,
from the NASA Sloan Atlas (NSA; Blanton et al. 2011). e.g., George 2023, or some newly formed star-forming
We use the bulge-disc decomposition catalog provided structure). Nevertheless, we have still included these
by Meert et al. (2015) to obtain B/T in r-band with the galaxies in our sample. To be precise, we do not refer to
Sérsic+Exponential model. it as an elliptical sample but rather as a strongly bulge-
dominated ETGs (sBDEs) sample. Finally, this sBDEs
sample consists of 104 galaxies, and Table 1 lists their
2.3. Sample matching and selection
basic properties.
We use a similar method from Zhang et al. (2019) Figrue 1 shows the distributions of stellar mass, Hi
to cross-match the FASHI-ALFALFA combined sample masses, and distances of the ATLAS3D and FASHI-
with SDSS. We first require the most probable optical ALFALFA combined sample. Compared with the
counterpart (OC) of each Hi detection to have a spatial ATLAS3D survey, which only detected the nearby ETGs
separation less than 5′′ with SDSS coordinates and a ve- within distances less than 42 Mpc, the galaxies in our
locity separation less than 300 km/s. Then, to avoid the sample show significantly farther distances (the median
contamination caused by the large beam sizes of FAST value is 152 Mpc). Furthermore, ATLAS3D survey pro-
and ALFALFA, we exclude the sources that have mul- vides a comprehensive and deep Hi observational study
tiple SDSS galaxies within 3′ (for FAST) or 4′ (for AL- of nearby ETGs, while our sample is biased towards Hi-
FALFA), meanwhile, being within three times velocity rich galaxies with high Hi mass (the median value is
width of the Hi line profiles. log MHI = 9.75), considering that our galaxies are se-

1 http://skyserver.sdss.org/casjobs/ 3 https://www.legacysurvey.org/viewer/
2 https://wwwmpa.mpa-garching.mpg.de/SDSS/DR7/
4

ATLAS3D ATLAS3D ATLAS3D

1.0 FASHI-ALFALFA
1.0 FASHI-ALFALFA 0.020 FASHI-ALFALFA

0.8 0.8
0.015
Density

Density

Density
0.6 0.6
0.010
0.4 0.4
0.005
0.2 0.2

0.0 9 10 11 12 0.0 6 8 10 0.000 0 100 200 300

log M* (M ) log MHI (M ) Distance (kpc)
Figure 1. From left to right: histograms of the stellar masses, Hi masses, and distances for the ATLAS3D (red) and FASHI-
ALFALFA combined sample (blue). The median values are indicated with vertical dashed lines.

lected from large flux-limited surveys. In other words, In addition, we refer to the total Hi sample (after
we focus on a rare population of ETGs that are more matching with the SDSS optical sample, MPA-JUH cat-
bulge-dominated and more Hi-rich compared to those alog, and DS18 morphological catalog) as the FASHI-
in the ATLAS3D . Such a rare population can only be ALFALFA combined Hi sample (14518 galaxies, 9468 in
found within a sufficiently large volume, but the rela- ALFALFA and 5050 in FASHI).
tively shallow depth of ALFALFA and FASHI may re-
sult in completeness issues with our sample. Specifi- 2.4. Hi profile stacking
cally, the incompleteness of our sample could impact We use the stacking technique to study the Hi profile
measurements such as the Hi mass function and the dis- properties of different sub-samples following the method
tribution of Hi column densities (Serra et al. 2012) if from Fabello et al. (2011). For each galaxy, we first shift
not appropriately corrected. Given the limited size of each Hi spectrum to rest frequency according to the Hi
our sample, this study aims to mitigate the influence of spectral center. We don’t use the central velocity pre-
sample incompleteness by focusing on comparisons with sented in ALFALFA and FASHI release catalogs, which
well-defined control samples. The main results in this is typically measured as the midpoint between the veloc-
work are also not sensitive to the effect of sample in- ities corresponding to the 50% of the two peak luminosi-
completeness. Furthermore, we have cross-matched our ties. Instead, we use the flux intensity-weighted central
sample with the NSA catalog to obtain the B/A value velocity to avoid potential issues caused by the complex
of SF-sBDEs and Q-sBDEs and find no differences be- Hi spectrum of ETGs when using the above-mentioned
tween them (the median values of 0.872 and 0.877, re- method. Then we stack the Hi spectra weighted by the
spectively). This ensures that the differences in velocity −2
noise level rms (w = σrms ). The final stacked spectra are
and SFR within SF-sBDEs and Q-sBDEs are not influ- normalized by the integrated luminosity to 1. It is worth
enced by inclination. noting that using a different stacking method, such as
Therefore, the results of this work are still robust. It’s without weighting, or using a different velocity center in
worth noting that we include both code 1 (high signal- the stacking process does not change our main results.
to-noise ratio, S/N) and code 2 (low S/N) objects from
ALFALFA. This means that the S/N of our sample is not 2.5. MaNGA observations
very high, and the median value of S/N is 7.0 from AL-
We cross-match our final sBDEs sample with galaxies
FALFA and 14.6 from FASHI. This may increase some
of the Mapping Nearby Galaxies at Apache Point Obser-
uncertainty in analyzing the Hi spectra properties. How-
vatory (MaNGA) survey, which results in six galaxies.
ever, the mass uncertainty, derived following Jones et al.
The MaNGA survey used integral field spectroscopy to
(2018) by combining the uncertainty in the integrated
obtain the two-dimensional spectra for more than 100,00
Hi line flux and distance with a minimum of 10% un-
local (0.01 < z < 0.15) galaxies (Bundy et al. 2015). The
certainty as provided in both the ALFALFA and FASHI
light is received by two spectrographs of MaNGA, cov-
catalogs, is not significantly low. The median value is
ering the wavelength ranges of 3600-6000Å and 6000-
only 0.06 for both the FASHI and ALFALFA samples.
10300Å, with a resolution of R ∼ 2000 (Drory et al.
2015).
 5

Therefore, the MaNGA view of our six galaxies would range from low to high mass. Young et al. (2014) has
provide a glimpse of spatially-resolved properties for the shown that the red sequence contains lots of gas-rich
sBDEs sample, such as stellar velocity (V∗ ), Hα velocity ETGs in the ATLAS3D sample, as some of them with low
(VHα ), Hα equivalent width (EW(Hα)), HδA equivalent mass are in the green valley and even into the blue cloud.
width (EW(HδA )) and Dn 4000. We adopt the measure- Similarly, our sample also contains some SF-sBDEs but
ments of these properties from the data release of the with higher stellar masses (log M∗ > 10.7) and higher
MaNGA data analysis pipeline (DAP, Westfall et al. SFRs. The difficulty in forming stars in ETGs may be
2019). due to the gas column surface densities being lower than
the critical surface density for star formation, resulting
3. SAMPLE PROPERTIES in star formation activities occurring in smaller regions
3.1. M∗ -SFR relation and with lower star formation efficiency (Bigiel et al.
2008; Oosterloo et al. 2010; Serra et al. 2012). Different
1.5 from the ATLAS3D sample, our sample tends to be more
SF-sBDEs Hi-rich, making it more likely to trigger star formation
1.0 over a large region within galaxies, leading to relatively
Q-sBDEs high star formation efficiencies (as later shown in Figure
0.5 4).
The SFRs from the MPA-JHU catalog are based
log SFR (M yr 1 )

0.0 on the SDSS spectra which only cover the central 3′′
of galaxies (Brinchmann et al. 2004). To verify the
SFRs from the MPA-JHU catalog, we cross-match our
0.5
sample with the GALEX-SDSS-WISE Legacy Catalog
(GSWLC-A2; Salim et al. 2016, 2018), which contains
1.0
the SFRs derived from broad-band ultraviolet-to-optical
photometry with constraints from infrared data. We
1.5
find that 88 of 104 galaxies in our sample have counter-
parts in the GSWLC, and almost all SF-sBDEs are still
2.0
scattered on the main sequence (only 2 SF-sBDEs are
classified as quiescent and 1 Q-sBDE is classified as star-
8 9 10 11 12
log M* (M ) forming galaxies in GSWLC). The fact that the MPA-
JHU catalog does not systematically underestimate the
Figure 2. The relationship between SFR and stellar mass SFR in SF-sBDEs suggests that it seems necessary to
for SDSS optical sample (contours) and Hi-rich sBDEs trigger central star formation to maintain a high star
(stars). The contours show the distribution of the SDSS op- formation rate. In this work, although we use SFRs
tical sample on the diagram, which includes 5%, 20%, 35%, from the MPA-JHU catalog to classify galaxies into star-
50%, 65%, 80%, and 95% of sample galaxies from inside out forming and quenched populations, the key results do
respectively. The dashed and dotted lines show the star-
not change when adopting the GSWLC SFRs. How-
forming main sequence defined in Peng et al. (2015) and
the divide between star-forming and quiescent galaxies from
ever, we indeed find more galaxies located in Green Val-
Zhang et al. (2019). ley using SFRs from GSWLC, and these galaxies may
be experiencing low-level star formation at outer regions
but not detected by 3′′ diameter SDSS fiber. We have
Figure 2 shows the SFR-stellar mass diagram from
also checked the GALEX image and found that some Q-
the SDSS optical sample, shown as contours, with our
sBDEs show weak UV signatures, suggesting that some
Hi-rich sBDEs overlapping on it. The dashed line repre-
low-level star formation may also occur in Q-sBDEs.
sents the star-forming main sequence defined in Peng
et al. (2015) and the dotted line shows the approxi-
mate divide between star-forming and quiescent galax- 3.2. Morphological distributions
ies, taken from Zhang et al. (2019). Star-forming sB- We present the distributions of three different struc-
DEs (denoted as SF-sBDEs) and quiescent sBDEs (Q- tural parameters mentioned in Section 2.2: C-index,
sBDEs) are shown in blue and red color, respectively. Sérsic n index, and B/T, shown in Figure 3 as functions
Surprisingly, we find that 31 (30%) of Hi-rich sBDEs of stellar mass. The contours in the left main subplots
are classified as star-forming galaxies, and they almost show the galaxy number density distributions of the el-
follow the star formation main sequence, covering a wide lipticals, lenticulars, and LTGs, which are classified by
 6

1.0 lipticals, while some SF-sBDEs also extend to the low-

Density

0.5 mass end. We carry out the Kolmogorov-Smirnov (K-S)

0.0 tests on structural parameter distributions of SF-sBDEs
3.75 8.5 9.0
LTGs 9.5 10.0 10.5 11.0 11.5 3.75
Lenticulars and Q-sBDEs samples. We find that the p-values be-
3.50 Ellipticals 3.50
sBDEs tween SF-sBDEs and Q-sBDEs are 0.26, 0.32, and 0.54
3.25 SF-sBDEs 3.25
Q-sBDEs for C-index, Sérsic n index, and B/T, respectively, indi-
3.00 3.00
C index

C index
cating that the hypothesis that these two distributions
2.75 2.75
2.50 2.50
statistically come from the same population cannot be
2.25 2.25
excluded. These results validate that these Hi-rich sB-
2.00 2.00
DEs do share a more similar distribution in structural
1.75 1.75 parameters as ellipticals, although we refer to them as
6 8.5 9.0 9.5 10.0 10.5 11.0 11.5 0 2 6 strongly bulge-dominated galaxies to be precise.
log M* (M ) Density
5 5

4 4
3.3. Hi and dust properties
Sersic n

Sersic n
Figure 4 shows two correlations associated with Hi
3 3
gas. It is important to note that the contours shown in
2 2 Figure 4 represent the overall FASHI-ALFALFA com-
bined Hi sample (i.e. the Hi-detected galaxies), rather
1 1 than the SDSS optical sample shown in Figure 2. There-
1.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 0 1 1.0
Density fore, we use light blue and gray contours, respectively,
0.8 0.8 to distinguish them throughout this work.
The left panel shows the relation between MHI frac-
0.6 0.6 tion (MHI /M∗ ) and M∗ . For reference, the relations
B/T

B/T

from ALFALFA, xGASS (the extended GALEX Arecibo

0.4 0.4 SDSS Survey; Catinella et al. 2010, 2018), and JINGLE
IV (De Looze et al. 2020, a combined galaxy sample4 )
0.2 0.2
are overlaid as black, green, and yellow dashed lines,
0.0 8.5 0.0 respectively. Compared to the FASHI-ALFALFA com-
9.0 9.5 10.0 10.5 11.0 11.5 0.0 2.5
log M* (M ) Density bined sample (which consists mostly of star-forming disk
galaxies), the Hi-detected sBDEs generally exhibit less
Figure 3. The three subplots, from top to bottom, show Hi gas. xGASS is a mass-unbiased sample with some Hi
the distributions of structural parameters C-index, Sérsic n
upper limit sources and JINGLE IV combines several
index, and B/T as a function of stellar mass. Red, green,
and blue represent the ellipticals, lenticulars, and LTGs, re-
samples with a wide range of properties. In comparison
spectively, which are classified by DS18 to our SDSS optical to these deeper samples, most of the galaxies in our sB-
sample. Our sBDEs are in blue (star-forming) and red (qui- DEs sample have sufficient Hi gas, which can be referred
escent) stars. The four subplots around show the distribu- to as Hi-rich sBDEs.
tions of three samples in terms of their colors, respectively, The right panel of Figure 4 shows the relation be-
and the black line shows the distributions of the total sBDEs tween Hi star formation efficiency (SFEHi , defined as
sample. SFR/MHI ) and M∗ . The light blue dashed line shows
the best fitting of this relation for star-forming galaxies
DS18 to our SDSS optical sample, with SF-sBDEs and only. SF-sBDEs show significantly high SFEHi , with al-
Q-sBDEs overlapping on it. In the top and right three most all of them lying above the fitting line, suggesting
subplots, we display the distributions of the stellar mass that SF-sBDEs may be undergoing rapid gas consump-
and structural parameters for three samples, each shown tion and strong star formation activity, while Q-sBDEs
in their colors, respectively, and the black line shows the exhibit low SFEHi .
distributions of the total sBDEs sample.
ETGs and LTGs show bimodal distributions in the
4 Including JINGLE (the JCMT dust and gas In Nearby Galax-
three structural parameters, with ETGs typically show-
ies Legacy Exploration), HRS (the Herschel Reference Survey),
ing higher values of C-index, Sérsic n index, and B/T KINGFISH (Key Insights on Nearby Galaxies: A Far-Infrared
(e.g., Blanton & Moustakas 2009). Notably, our Hi-rich Survey with Herschel), HAPLESS (Herschel-ATLAS Phase-1
sBDEs are more consistent with the distribution of el- Limited-Extent Spatial Survey), and HiGH (Hi-selected Galaxies
in H-ATLAS)
 7

1.5
SF-sBDEs 8
SF-sBDEs
1.0 Q-sBDEs Q-sBDEs

0.5 9

log SFE (yr 1 )

log MHI /M*

0.0
10
0.5

1.0 11
ALFALFA
1.5 xGASS 12
JINGLE IV
2.0 8 9 10 11 8 9 10 11
log M* (M ) log M* (M )
Figure 4. The Correlations of Hi mass fraction (MHI /M∗ , left panel) and Hi star formation efficiency (SFR/MHI , right panel)
with stellar mass are shown for FASHI-ALFALFA combined sample (light blue contours), SF-sBDEs (blue stars), and Q-sBDEs
(red stars). The black, green, and yellow dashed lines of the left panel show the median relations from ALFALFA, xGASS, and
JINGLE IV, respectively. In the right panel, the light blue dashed line shows the best-fitting relation of star-forming galaxies
in the FASHI-ALFALFA combined Hi sample.

The Hi depletion time, denoted as tdep,Hi , is the recip- 8

rocal of SFEHi , which is useful to describe the star for- SF-sBDEs
mation activity and gas consumption (e.g., Wang et al.
7 Q-sBDEs
2020; Saintonge & Catinella 2022). The median tdep,Hi
for the SF-sBDEs sample is 3 Gyr, which is shorter than
the typical values of ∼9 Gyr for ALFALFA and ∼5 Gyr 6
for xGASS. Nevertheless, the Hi reservoir in Hi-rich SF-
sBDEs is sufficient to maintain star formation activity
H /H

for a few Gyrs, as long as it can form H2 gas. 5

Figure 5 shows the relation between Balmer decre-
ment (Hα/Hβ, as a proxy for dust content) and stellar 4
mass. Hα/Hβ can be an indicator of dust extinction or
dust content, with higher values indicating more dust
in galaxies (Kennicutt 1992). Interestingly, Barrera- 3
Ballesteros et al. (2020) proposed that optical extinc-
tion is a reliable proxy for estimating molecular gas
content in the absence of direct sub/millimeter obser- 28 9 10 11 12
vations. This is consistent with the results in Figure 5 log M* (M )
that Q-sBDEs typically have Hα/Hβ values around the
theoretical value, indicating a lack of dust and molec- Figure 5. The dust properties estimated by Balmer decre-
ular gas, while SF-sBDEs have significant dust content ment (Hα/Hβ) as a function of stellar mass, for SDSS optical
sample (contours) and Hi-rich sBDEs (stars). The horizontal
and likely molecular gas.
dotted line shows the intrinsic value of 2.86.
4. RESULTS
timescales. With respect to the derived star formation
4.1. The Stellar Population of Hi-rich sBDEs
histories from spectral fitting, these parameters are in-
As mentioned in Section 2.2, spectroscopic features sensitive to dust attenuation and model-independent.
such as EW(Hα), EW(HδA ), and Dn 4000 are used to Hα emission traces the recent intense star formation
diagnose the recent star formation histories on different
 8
2.2 10
SF-sBDEs 2.0 SF-sBDEs SF-sBDEs
Q-sBDEs Q-sBDEs 8 Q-sBDEs
2.0
6

log EW(H ) (Å)

1.8 1.5

EW(H A ) (Å)
4
Dn (4000)

1.6 1.0 2

1.4 0
0.5 2
1.2
4
0.0 0.5 1.0 1.5 2.0 0.0 5 0 5 10 1.00 1.25 1.50 1.75 2.00
log EW(H ) (Å) EW(H A ) (Å) Dn (4000)
Figure 6. The relations of the Dn 4000 with EW(Hα) (left panel), EW(Hα) with EW(HδA ) (medial panel), and EW(HδA ) with
Dn 4000 (right panel). The dashed lines in the left and right panels show the best polynomial fitting. See the caption of Figure
4 for more details on the contours and symbols.

of massive stars within most of their lifetime (∼ 10 than in the optical, and Serra et al. (2012), based on the
Myr) (e.g., Fumagalli et al. 2012; Wang & Lilly 2020). ATLAS3D , found that in more than half of the cases,
EW(HδA ) also traces the recent star formation, but it the Hi gas in ETGs is kinematically decoupled from the
does so within a longer time scale, up to 1 Gyr (e.g., stars or not disk-like. Therefore, the relation between
Bruzual & Charlot 2003; Le Borgne et al. 2006). 4000 Hi gas and stellar populations might be complex. We
Å break is sensitive to the light-weighted stellar age will discuss the resolved stellar population of the subset
over the past 2 Gyr, which shows significantly higher of sBDEs matched with the MaNGA survey in Section
values in quiescent galaxies (e.g., Balogh et al. 1999; 4.5.
Kauffmann et al. 2003; Wang et al. 2018). To better In the literature, some quiescent galaxies could expe-
understand the star formation properties and star for- rience recent secondary star formation activity, which
mation history of Hi-rich sBDEs, we focus on their stel- is referred to as “rejuvenation” (e.g., Fang et al. 2012;
lar populations using three correlations between pairs Zhang et al. 2023b; Paudel et al. 2023). They typi-
of EW(Hα), EW(HδA ), and Dn 4000, and the results are cally show low-level or medium-level star formation over
shown in Figure 6. the most recent ∼100 Myr and could transition back to
It can be seen from Figure 6 that most of the SF- the green valley or even the blue cloud from the red
sBDEs show young stellar populations and are located sequence. We speculate that our SF-sBDEs may ex-
in the normal star-forming region of the SDSS optical perience rejuvenation, which is similar to the scenario
sample. Furthermore, we can find from the left and right of H2 -rich ATLAS3D in (Young et al. 2014). How-
panels in Figure 6 that SF-sBDEs exhibit slightly larger ever, these SF-sBDEs show even longer star formation
Dn 4000, with a slight departure in distribution. This timescales, which can be regarded as strong rejuvenated
indicates that the stellar populations of SF-sBDEs are systems. Compared to typical rejuvenated galaxies, our
slightly older than those of normal star-forming galax- SF-sBDEs show relatively large EW(Hα) and EW(HδA ),
ies. This is consistent with the picture that old stellar blending in with normal star-forming galaxy popula-
populations dominate these SF-sBDEs, but additionally tion (Zhang et al. 2023b). This suggests that these SF-
have a recent star formation activity. These young sub- sBDEs may maintain a long timescale (∼1 Gyr or more)
populations of galaxies may have formed from the ac- rejuvenated star formation, which is consistent with the
creted gas, and it’s even possible to discover an inner gas gas depletion timescales of ∼3 Gyr (see the right panel
disk and the associated stellar disk component within of Figure 4).
the galaxies (Oosterloo et al. 2010; Young et al. 2014). Furthermore, most Hi-rich Q-sBDEs overlap with the
Some surveys have demonstrated that UV emission distribution of normal quiescent galaxies, indicating a
is widespread in ETGs, originating not only from old lack of recent star formation and the presence of an old
stars but also from young stars, suggesting the presence stellar population. This is consistent with the results
of recent star formation in these galaxies (e.g., Mager from ATLAS3D (e.g., Oosterloo et al. 2010; Young et al.
et al. 2018; Pandey et al. 2024). The asymmetry and 2014). These galaxies are difficult to trigger medium- or
clumpiness of ETGs are significantly larger in the UV high-level star formation over a long time-scale, whether
 9

they accrete gas from the environment or retain gas af- Sc07) and Kewley et al. (2006) (right panel, labeled
ter they transition to the ETGs. As a result, they are Ke06).
classified as quenched galaxies and only maintain old We use both the two BPT diagnostic diagrams to ex-
stellar populations. We will further discuss the possible clude AGNs based on the following criteria. We have
reasons for rejuvenated star formation in Section 5.2. divided our sample into three categories: (a) the galax-
ies above the Ke01 line in the [Nii]6583 /Hα diagnostic
4.2. The BPT Diagrams and Metallicity Properties of diagram (left panel) are AGNs shown as crosses; (b) the
Hi-rich sBDEs galaxies that lie below the Ke01 lines in both diagnos-
tic diagrams are considered as Hii-dominated galaxies
The presence of warm ionized gas in ellipticals does
shown as red and blue stars; (c) those in the compos-
not appear to be uncommon (e.g., Phillips et al. 1986;
ites region in the left panel but above the Ke01 line
Falcón-Barroso et al. 2006). In this section, we aim to
in the [Sii]6717,6731 /Hα diagnostic diagram (right panel)
study the gas-phase metallicity properties of Hi-rich sB-
are composite galaxies shown as stars with light colors.
DEs, so it’s important to select galaxies where ionized
Finally, we have 34 (31 from SF-sBDEs and 3 from
gas is dominated by Hii region.
Q-sBDEs) sBDEs belong to Hii-dominated galaxies and
BPT line diagnostic diagram is a useful tool for identi-
5 (only 1 from SF-sBDEs and 4 from Q-sBDEs) sBDEs
fying AGNs and categorizing them into Seyfert galaxies
are considered as composite galaxies. It is worth noting
and low-ionization narrow emission-line regions (LIN-
that in our sample, most Q-sBDEs have EW(Hα) < 3
ERs) (Kauffmann et al. 2003; Kewley et al. 2006; Schaw-
Å, which is not shown in Figure 7.
inski et al. 2007). However, some leaked ionizing radia-
After excluding AGNs from our sample, we derive
tion from Hii regions or old stars can ionize the diffuse
gas-phase metallicity with the O3N2 calibration from
gas and form the diffuse ionized gas (DIG), which can-
Marino et al. (2013). The metallicity of galaxies in-
not be explained by the classical model of Hii regions
creases with increasing stellar masses, known as the
(Kewley et al. 2019). The DIG can emit some strong
mass-metallicity relation (MZR), which can serve as
low ionization lines, such as [Oi], [Nii], [Sii], and [Oii],
an observational constraint for galaxy evolution models
and these emissions can enhance their line ratios with
(e.g., Lequeux et al. 1979; Andrews & Martini 2013; Gao
Hα (Zhang et al. 2017). As a result, the DIG can signif-
et al. 2018; Huang et al. 2019). The MZR is shown in
icantly impact the AGNs’ diagnostic results from BPT
the left panel of Figure 8 and our sBDEs sample is rep-
diagrams and may lead to an overestimation of metal-
resented as stars, with colors similar to those in Figure
licity when using [Nii]6583 /Hα and [Sii]6717,6731 /Hα line
7. Although the SF-sBDEs are located on a wide region
ratios (Poetrodjojo et al. 2018).
on MZR, they statistically show lower metallicity, with
In order to exclude DIG-dominated galaxies from our
most of them being below the ridge line of SDSS MZR
sample, the empirical separation of EW(Hα) > 3 Å is
as indicated by the contours. Compared to SF-sBDEs,
useful (Cid Fernandes et al. 2011; Belfiore et al. 2016).
all of the Q-sBDEs have significantly lower metallicity.
We find that all of the SF-sBDEs and only 15% of the
It’s worth noting that, even though we have made ef-
Q-sBDEs in our sample have EW(Hα) > 3 Å. Based on
forts to exclude galaxies dominated by AGNs or DIG,
this, we select Hii region dominated galaxies from our
there may still be contamination in the measurements
sample using two BPT line diagnostic diagrams with
of gas-phase metallicity of Q-sBDEs. Considering that
EW(Hα) > 3 Å and S/N > 3 in all emission lines we
DIG can overestimate metallicity with the O3N2 (Kew-
used. The BPT diagrams are shown in Figure 7, with
ley et al. 2019), correction of the DIG contamina-
our sBDEs shown as stars, overlapping with the SDSS
tion is expected to strengthen the above result. The
optical sample shown as contours.
N2O2 metallicity indicator is less contaminated by DIG
The left panel shows the [Nii]6583 /Hα and
(Zhang et al. 2017). Due to the limited number of
[Oiii]5007 /Hβ diagnostic diagram, while the right panel
galaxies meeting the criteria, we present the results in
displays the [Sii]6717,6731 /Hα and [Oiii]5007 /Hβ diagnos-
Appendix A. Using the N2O2 calibration to calculate
tic diagram. Galaxies lying above the lines of Kewley
metallicity does not change our results, and Hi-rich sB-
et al. (2001) (solid lines, labeled Ke01) are dominated
DEs also show significantly lower metallicity in MZR
by AGNs and below the line of Kauffmann et al. (2003)
(N2O2) shown in Figure 14.
(dotted line, labeled Ke01) are dominated by star-
We speculate that the cold gas in these sBDEs may
forming galaxies. The region between the two lines in
come from external mechanisms, i.e., accretion of cold
the left panel is SF-AGN composites. The division lines
gas from the environment or recent wet mergers with
between Seyferts and LINERs are shown as the dashed
gas-rich galaxies. Specifically, the lower metallicity of
line from Schawinski et al. (2007) (left panel, labeled
 10

1.0 SF-sBDEs 1.0 SF-sBDEs

Q-sBDEs Q-sBDEs

0.5 0.5
log ([OIII]/H )

log ([OIII]/H )
0.0 0.0

0.5 0.5

1.0 Ka03 1.0

Ke01 Ke01
Sc07 Ke06
1.51.5 1.0 0.5 0.0 0.5 1.5 1.00 0.75 0.50 0.25 0.00 0.25
log ([NII]/H ) log ([SII]/H )
Figure 7. The BPT line diagnostic diagrams for SDSS optical galaxies (contours) and Hi-rich sBDEs (stars and crosses)
with S/N > 3 in all five emission lines. Left panel: the [Nii]6583 /Hα and [Oiii]5007 /Hβ diagnostic diagram. Right panel: the
[Sii]6717,6731 /Hα and [Oiii]5007 /Hβ diagnostic diagram. The empirical SF/AGN boundary of Kauffmann et al. (2003) (labeled
Ka03), the theoretical relation of Kewley et al. (2001) (labeled Ke01), and the division lines between Seyfert galaxies and LINER
of Schawinski et al. (2007) (left panel, labeled Sc07) and Kewley et al. (2006) (right panel, labeled Ke06) are shown in dotted,
solid, and dashed black lines, respectively. Hi-rich sBDEs above the Ke01 line in the left panel are shown in crosses and below
the Ke01 line in the right panel are shown in red and blue stars, respectively. Marked in light red and light blue are those
sBDEs below the Ke01 line in the [Nii]6583 /Hα diagnostic diagram but above the Ke01 line in the [Sii]6717,6731 /Hα diagnostic
diagram.

8.7 8.7 8.7

SF-sBDEs SF-sBDEs SF-sBDEs
Q-sBDEs Q-sBDEs Q-sBDEs
8.6 8.6 8.6
12 + log(O/H) (O3N2)

12 + log(O/H) (O3N2)

12 + log(O/H) (O3N2)

8.5 8.5 8.5

8.4 8.4 8.4
8.3 8.3 8.3
8.2 8.2 8.2
8.1 8 9 10 11 8.18 9 10 11 8.18 9 10 11
log M* (M ) log M* 0.32 × log SFR log M* 0.35 × (log MHI 9.8)
Figure 8. Mass-metallicity relation (left panel), FMRSFR using the β = 0.32 from Mannucci et al. (2010) (middle panel), and
FMRHi using the β = 0.35 from Bothwell et al. (2013) (right panel) for the ALFALFA-FASHI combined Hi sample (contours)
and Hi-rich sBDEs (stars). See the caption of Figure 7 for more details on the color of stars.

Q-sBDEs is attributed to the recent accretion of cold sBDEs with low metallicity may have experienced a re-
metal-poor gas from the environment without undergo- cent accretion of metal-poor gas, while those with nor-
ing metal enrichment. As for SF-sBDEs sample, given mal metallicity may have already been enriched. Our
that most of our SF-sBDEs sample have relatively long findings here are consistent with the metallicity analy-
star-formation history (see Section 4.1), star formation sis of ATLAS3D (Young et al. 2014; McDermid et al.
itself can enrich the gas-phase metallicity rapidly (Davis 2015) that the most gas-rich ETGs in low-density envi-
& Young 2019). This may explain the wide distribu- ronments exhibit low stellar metallicity and signs of gas
tion of the SF-sBDEs on the MZR in Figure 8. Those accretion.
 11

We next focus on studying the fundamental metallic- and velocity resolution compared to ALFALFA, so we
ity relations (FMRs) of the sample based on SFR and display the stacked Hi spectra from FASHI separately
Hi mass, which can effectively predict the dilution of in the right two panels. However, we overlap them with
metal abundances caused by gas inflows and the associ- the same control samples from ALFALFA to facilitate
ated star formation enhancement. Many works find that straightforward comparisons. The velocity widths of the
MZR shows a secondary dependence on SFR (e.g., Elli- Hi line profile W50 are shown in each panel, which are
son et al. 2008; Mannucci et al. 2010; Huang et al. 2019) measured at the 50% level of the peak after smoothing.
and Hi mass (e.g., Bothwell et al. 2013; Jimmy et al. The uncertainties of W50 are estimated using the boot-
2015; Chen et al. 2022), which are known as the funda- strapping method. For SF-sBDEs and Q-sBDEs sub-
mental metallicity relation (FMRSFR and FMRHi ). samples, we perform 1000 iterations of resampling with
To better quantify the secondary dependence of the the sample size fixed, allowing for repetition. We also
MZR on SFR, Mannucci et al. (2010) introduces a pa- repeatedly select the same number of galaxies for the
rameter µα that combines SFR with stellar mass for the control samples but from the parent sample for spectral
FMRSFR relation: stacking. The velocity widths and uncertainties of the
stacked spectra are adopted by the median values and
µα = log(M∗ ) − α log(SFR). (1) the values of 16 and 84 percentiles in each panel.
SF-sBDEs appear to have slight differences with nar-
Additionally, Bothwell et al. (2013) defines another pa- rower, more concentrated, and single-peaked profiles
rameter µβ to describe the secondary dependence of the compared to the star-forming control sample. The
MZR on Hi mass: stacked spectra of Q-sBDEs are also slightly narrower
compared to the quiescent control sample, but they do
µβ = log(M∗ ) − β(log(MHI ) − 9.80). (2)
not exhibit the concentrated and single-peaked shape
We use α = 0.32 from Mannucci et al. (2010) and β = seen in SF-sBDEs, instead displaying a flatter profile.
0.35 from Bothwell et al. (2013) in this work to obtain The W50 of SF-sBDEs and Q-sBDEs are only marginally
FMRSFR and FMRHi , respectively. larger than control samples, invoking further analysis
The final FMRSFR and FMRHi relation of ALFALFA- with larger sample sizes. Overall, these results highlight
FASHI combined sample (contours) and sBDEs (stars) the importance of the kinematics of the Hi gas reservoir.
are shown in the middle and right panel of Figure 8. The narrower Hi profile width can be mainly attributed
Again, SF-sBDEs show an overall offset in the FMRs, to the slower rise of the rotation curve and more con-
which is more significant in FMRHi with most galaxies centrated Hi distribution (Yu et al. 2022). The wide Hi
located in the low metallicity region. This supports the profile of Q-sBDEs suggests that the gas may be dis-
above speculation that most Hi-rich sBDEs are likely tributed at a larger radius in the outer regions of the
experiencing continuous accretion of metal-poor gas or galaxy, making it difficult to effectively form stars ac-
gas-rich mergers, and the gas inflow rate is larger than cording to the star formation law (e.g. Bigiel et al. 2008;
the metal production rate (Köppen & Edmunds 1999; Shi et al. 2011). These findings are consistent with the
Grønnow et al. 2015). And this is also true for Q-sBDEs. results of previous studies that the gas distribution of
star-forming ETGs is more centrally concentrated (e.g.,
4.3. The stacked Hi spectra Tacchella et al. 2016; Lee et al. 2023). The shape of the
rotation curve may also play an important role. How-
Figure 9 shows the stacked Hi spectra of SF-sBDEs
ever, considering that the Hi distribution and kinematics
(blue) and Q-sBDEs (red), overlapping with the con-
are likely disturbed and unsettled, and that the infor-
trol samples from ALFALFA. The reader is referred to
mation obtained solely from line widths is very limited,
Section 2.5 for more detailed descriptions of Hi profile
high-resolution observations of some of these sBDEs may
stacking method. The stacked Hi profiles are still quite
be necessary in the future.
noisy (especially the SF-sBDEs from ALFALFA) due to
the low S/N of some sources. The control samples are
randomly selected from ALFALFA to have the same dis- 4.4. Environments
tribution of stellar mass with SF-sBDEs and Q-sBDEs, Many works have found that the Hi detection rate
shown as green (star-forming galaxies) and gray (quies- of ETGs and the quenching of star formation strongly
cent galaxies) colors. Due to the limited number of qui- depends on the environment density (e.g., di Serego
escent galaxies in the ALFALFA sample (less than 1000 Alighieri et al. 2007; Grossi et al. 2009; Serra et al.
galaxies), we select 500 galaxies for each control sample, 2012). In denser environments, such as the central re-
allowing for duplicates. FASHI has a different rms level gion of galaxy clusters, Hi can be easily stripped from
12

ALFALFA FASHI
Controls (star-forming) Controls (quiescent) Controls (star-forming) Controls (quiescent)
SF-sBDEs (ALFALFA) Q-sBDEs (ALFALFA) SF-sBDEs (FASHI) Q-sBDEs (FASHI)
0.006
180+34 217+79 208+26 248+13 174+36 217+79 223+29 248+13
Normalized luminosity

54 23 13 26 19 13
0.004

0.002

0.000

500 0 500 500 0 500 500 0 500 500 0 500

Velocity (km/s) Velocity (km/s)
Figure 9. Stacked Hi spectra. Each panel displays the control samples of star-forming and quiescent galaxies from ALFALFA
in green and grey colors, respectively, in the background. The left two panels show the stacked spectra for Hi-rich sBDEs from
ALFALFA, while the right two panels display those from FASHI. The stacked spectra of SF-sBDEs and Q-sBDEs are shown in
blue and red, respectively. The stacked spectra are normalized by the integrated luminosity to 1. The velocity width of the Hi
profile, W50 , along with the estimated uncertainty from bootstrapping, is shown in each panel in their respective colors.

35 1.0
Control sample (elliptical) Control sample (elliptical)
30 HI-rich sBDEs HI-rich sBDEs
0.8
Quiescent fraction (< r)

25
Average number (< r)

20 0.6

15 0.4
10
0.2
5

0 500 1000 1500 2000 0.0 500 1000 1500 2000

Distance (kpc) Distance (kpc)
Figure 10. The average galaxy numbers (left panel) and the fraction of quiescent galaxy (right panel) within the radius of
500, 1000, 1500, and 2000 kpc. The orange and black lines show the results for Hi-rich sBDEs and the control sample with
the same stellar mass and redshift distribution as the total Hi-rich sBDEs sample. The uncertainties are estimated using the
bootstrapping method.

galaxies due to factors like strong tidal stripping, galaxy ment of our sBDEs on a larger scale. For each galaxy,
harassment, and ram-pressure stripping (e.g., Giovanelli we search from the SDSS optical sample for all galax-
& Haynes 1983; Moore et al. 1996; Wang et al. 2021), ies that are located within the velocity difference of
leading to a more effective star formation quenching of 1000 km/s and some certain projected physical distances
galaxies (e.g., Balogh et al. 1998; Thomas et al. 2010; (500, 1000, 1500, 2000 kpc). We then count the total
Lee et al. 2023). number of galaxies and the number of quiescent galax-
As we have excluded galaxies with multiple objects ies using the criteria above for each individual galaxy.
within 4′ (for ALFALFA) or 3′ (for FASHI) and 3 times Additionally, we exclude galaxies within 4′ from these
the Hi velocity widths, we thus investigate the environ- counts to eliminate the bias from our sample selection.
 13

35 1.0
Control sample (star-forming) Control sample (star-forming)
30 Control sample (quiescent) Control sample (quiescent)
SF-sBDEs 0.8 SF-sBDEs
Q-sBDEs Q-sBDEs

Quiescent fraction (<r)

25
Average number (<r)

20 0.6

15 0.4
10
0.2
5

0 500 1000 1500 2000 0.0 500 1000 1500 2000

Distance (kpc) Distance (kpc)
Figure 11. Same as Figure 10, but with Hi-rich sBDEs dividing into SF-sBDEs (blue) and Q-sBDEs (red). The control samples
are constructed separately for SF-sBDEs and Q-sBDEs, which are from the SDSS sample with controlling the same stellar mass
and redshift.
After completing the counts for each galaxy, we focus on neighbors on the scale of 1 Mpc. This provides a Hi-
the statistical results of SF-sBDEs and Q-sBDEs sam- rich environment, which is the necessary condition for
ples. This provides us with the average number of to- gas accretion from nearby environments. Our finding is
tal galaxies around one galaxy and the overall quiescent also consistent with the previous studies that the detec-
fraction at a projected physical distance. tion of Hi in ETGs depends on the environment (e.g.,
First of all, to understand how the environment affects di Serego Alighieri et al. 2007; Grossi et al. 2009; Ash-
the gas properties of Hi-rich sBDEs, we build an ellip- ley et al. 2019). Furthermore, we find that there are
tical control sample that is 10 times (1040 galaxies) the only a small number of galaxies within 500 kpc in the
size of our sample. The galaxies in this control sample sBDEs sample, as shown in the the left panel of Figure
are randomly selected from SDSS with T-type ≤ 0 and 10. Consistent with this, most of the sBDEs (∼82%) are
PS0 < 0.5 to match the distribution of stellar mass and actually central galaxies. The Hi gas from sBDEs may
redshift with the total Hi-rich sBDEs sample. Following come from the direct gas accretion from its halo, or the
the same method mentioned above, we calculate the av- capture of the dwarf galaxies.
erage number of galaxies and quiescent fraction in the Then, we explore the environment of SF-sBDEs and
environment of the control sample statistically. Q-sBDEs, respectively, following the same approach
The left panel of Figure 10 shows the average number above. The control samples of star-forming and quies-
and the right panel shows the quiescent fraction for each cent galaxies are constructed separately from the SDSS
sample. The uncertainties of the average number and optical sample, matching stellar mass and redshift with
quiescent fraction are estimated using the bootstrap- SF-sBDEs and Q-sBDEs, respectively. The results are
ping method. In practice, we perform 1000 iterations shown in Figure 11. As can be seen, both the SF-sBDEs
of resampling with a sample size fixed, allowing for rep- and Q-sBDEs are more common in lower-density envi-
etition. The error bars show the range of 16 to 84 per- ronments and potentially have significantly fewer quies-
centiles in both panels. cent neighbors than the quiescent control sample, but
As shown in Figure 10, the normal ellipticals are lo- they are almost similar to the star-forming control sam-
cated in a denser environment and show a higher qui- ple. In comparison to Q-sBDEs, SF-sBDEs only show
escent fraction of their neighbors compared to Hi-rich slightly lower average galaxy numbers in the environ-
sBDEs. This result suggests that the environment plays ment. For the quiescent fraction in the right panel of
an important role for Hi-rich sBDEs to acquire cold gas. Figure 11, SF-sBDEs do not exhibit significant differ-
Hi-rich sBDEs potentially reside in lower-density regions ences compared to Q-sBDEs within 1 Mpc, only show-
and potentially have relatively much more star-forming
14

ing a slightly lower quiescent fraction at larger projected other hand, the Q-sBDE, MaNGA 1-295446, does not
physical distances. show significant recent star formation indicated by Hα
These results are consistent with the findings of emission, while EW(HδA ) map shows that this galaxy
ATLAS3D surveys that the Hi properties of ETGs has a wide area of star formation in last Gyr timescale.
strongly depend on environment density (Serra et al. Overall, these sBDEs show many special properties,
2012). Based on the FASHI-ALFALFA sample, we find indicating that the acquisition of cold gas can result
that Hi-rich sBDEs are not only more common in lower- in significant differences in galaxy properties, even for
density environments but also potentially have signifi- Q-sBDEs. These recently obtained cold gas appear to
cantly fewer quiescent neighbors than the elliptical con- form a gas disk inside the original stellar components.
trol sample and the star-forming control sample from If this is true, the growth of the disk would potentially
SDSS. On the other hand, ATLAS3D sample was also modify the galaxy morphology. This is also the reason
observed CO J=1-0 and J=2-1 emission to study the that we do not refer to these sBDEs as ellipticals, even
molecular gas properties of ETGs (Young et al. 2011, though they are classified as ellipticals from the machine
2014). They find that the environment plays only a weak learning sample. We refer to the event in ellipticals as
role in CO detection of gas-rich ETGs and the galaxies “morphology rejuvenation”, where the accreted cold gas
within the cluster have the potential to retain molecular forms stars on a disk and ultimately changes the galaxy
gas. These retained molecular gas could fuel star forma- morphology. We plan to examine whether this is the
tion in galaxies, which is why SF-sBDEs and Q-sBDEs case in the future.
samples do not show significant differences in the aver- To compare the differences between SDSS spectra and
age number and quiescent fraction in the environment the spatial stellar population properties of galaxies, we
(Figure 11). tested the distribution of this subset of MaNGA data
in stellar population relations. We get the Dn 4000,
4.5. MaNGA view EW(Hα), and EW(HδA ) at each point with S/N larger
than 5 within the MaNGA field for these galaxies. Fig-
In Figure 12 we display the spatially resolved prop-
ure 13 shows the comparison of the three SF-sBDEs
erties of 3 SF-sBDEs (top three rows) and 3 Q-sBDEs
(top panels) and three Q-sBDEs (bottom panels) galax-
(bottom three rows) that are also in the MaNGA survey.
ies measured in MaNGA and SDSS. The stars show the
From left to right, each column shows the SDSS color
measurements from SDSS spectra and the contours show
image, V∗ , VHα , EW(Hα), EW(HδA ), and Dn 4000, re-
the spatial measurements from MaNGA in correspond-
spectively. Although there are only 6 galaxies from our
ing colors. First of all, we find two SF-sBDEs (MaNGA
sample in MaNGA, they show diverse properties, allow-
1-251788 and 1-258380), compared to the center of these
ing us to better understand how these sBDEs obtain
galaxies, show slightly older age and lower EW(Hα) in
gas.
the outer part. At the same time, EW(HδA ) does not
First of all, we have identified two significantly gas-
show significant changes, indicating that they both have
versus-stars misaligned galaxies (MaNGA-ID: 1-51810
star formation with a longer time scale. And the outer
and 1-176908) and two slightly misaligned galaxies
region of MaNGA 1-51810 extends toward both larger
(MaNGA-ID: 1-258380 and 1-415777). The kinematic
EW(Hα) and EW(HδA ). This is consistent with our re-
misalignments of these galaxies suggest that they have
sults that these SF-sBDEs can be regarded as strong re-
experienced external gas accretion or gas-rich misaligned
juvenated systems with longer star formation timescales.
mergers. Such a high proportion of misaligned galax-
For Q-sBDEs, it is worth noting that we don’t obtain
ies (4/6) indicates that our sBDEs are good candidates
enough points of EW(HδA ) with S/N>5 in MaNGA 1-
for misaligned galaxies, deserving further observations
295446. MaNGA 1-415777 doesn’t exhibit significant
in the future.
changes in all three diagrams, while MaNGA 1-176908
Furthermore, we also find two SF-sBDEs show a rota-
and 1-295446 show wide distributions in the Dn 4000-
tion feature in the central region of galaxies (1-251788
EW(Hα) relation. This suggests that they may experi-
and 1-258380). Especially MaNGA 1-251788, shows a
ence patchy but low- or medium-level star formation in
small rotating structure at inner regions with respect
the outer regions of galaxies. Furthermore, MaNGA 1-
to a non-rotating stellar velocity field at large radii.
176908 exhibits a bimodal distribution in the EW(Hα)-
The inner regions show an enhanced specific SFR (high
EW(HδA ) and EW(HδA )-Dn 4000 relations. This implies
EW(Hα)) and old stellar population (high EW(HδA )
that rejuvenation events may also have been occurring
and low Dn 4000). The misaligned Q-sBDE (MaNGA
within Q-sBDEs.
1-176908) is also forming stars likely in a newly formed
disk, which can be seen in the panel of EW(Hα). On the
 15

MaNGA (SF-sBDEs)
SDSS image Stellar velocity H velocity H equivalent width H A equivalent width Dn4000
5.0 1.575
20 60 14 1.550
40 4.5
1.525
1-51810

10 12 4.0
20 1.500
0 0 10 3.5
3.0 1.475
20 8 1.450
10 2.5
40 6 1.425
20 60 2.0 1.400
4 1.5 1.375
12 5 1.80
40 300 1.75
200 10
1-251788

20 4 1.70
100 8 1.65
0 0 6 3 1.60
20 100 4 2 1.55
200 1.50
40 2 1 1.45
300
1.40
150 9 1.9
100 150 10 8
100 7 1.8
1-258380

50 50 8 6 1.7
0 0 6 5 1.6
50 50 4
100 4 3 1.5
100 150 2 1.4
2
150 1

MaNGA (Q-sBDEs)
SDSS image Stellar velocity H velocity H equivalent width H A equivalent width Dn4000
100 200 10
75 7 2.2
50 6 8
1-176908

100
25 5 2.0
0 0 4 6
1.8
25 3 4
50 100 2 1.6
75 1 2
100 200 1.4

150
100 2.5 10 2.2
100
1-415777

50 50 2.0 8 2.0
0 0 1.5 6 1.8
50 50 1.0 4 1.6
100 100 2
0.5 1.4
150

150 12 2.0
20 12
100 10
1-295446

10 10 1.8
50 8
8 1.6
0 0 6
6
10 50 4 1.4
100 4
20 2 2 1.2
150

Figure 12. The three SF-sBDEs (top three rows) and three Q-sBDEs (bottom three rows) observed in MaNGA with the
corresponding MaNGA-ID marked in the left. Each column shows, from left to right: 1. SDSS g-r-i image with the MaNGA
hexagonal Fov overlaid; 2. the stellar velocity fields; 3. Hα velocity fields; 4. Hα equivalent width; 5. HδA equivalent width; 6.
Dn 4000. These maps are obtained from the MaNGA DAP.

5. DISCUSSION pendence on both the environment and mass (Grossi

ETGs are often characterized as “red and dead” galax- et al. 2009; Ashley et al. 2019). When compared to
ies, with little or even no cold gas. However, this LTGs, ETGs do indeed tend to exhibit gas-poor prop-
paradigm is considered too simplistic, as the develop- erties, especially ellipticals. In the morphological classi-
ment of many Hi observations, such as HIPASS (Meyer fication sample of SDSS DR7 from Domı́nguez Sánchez
et al. 2004), ALFALFA (di Serego Alighieri et al. 2007; et al. (2018), with redshift less than 0.1, ellipticals and
Grossi et al. 2009), and ATLAS3D (Serra et al. 2012, lenticulars account for 8.2% and 33.1%, respectively, of
2014). At current observational depths, the Hi detec- the total number of galaxies. After matching with the
tion rate of ETGs is about 20%, with a significant de- FASHI-ALFALFA combined sample, the fractions drop
16
2.2 10
1-51810 1-51810 1-51810
2.0
2.0
1-251788 1-251788 8 1-251788
1-258380 1-258380 1-258380
6

log EW(H ) (Å)

1.8 1.5

EW(H A ) (Å)
4
Dn (4000)

1.6 2
1.0
1.4 0
0.5 2
1.2
4
1.00.0 0.5 1.0 1.5 2.0 0.0 5 0 5 10 1.00 1.25 1.50 1.75 2.00
log EW(H ) (Å) EW(H A ) (Å) Dn (4000)
2.2 10
1-176908 1-176908 1-176908
2.0
2.0
1-415777 1-415777 8 1-415777
1-295446 1-295446 1-295446
6
log EW(H ) (Å)

1.8 1.5

EW(H A ) (Å)
4
Dn (4000)

1.6 2
1.0
1.4 0
0.5 2
1.2
4
1.00.0 0.5 1.0 1.5 2.0 0.0 5 0 5 10 1.00 1.25 1.50 1.75 2.00
log EW(H ) (Å) EW(H A ) (Å) Dn (4000)
Figure 13. The spatial stellar population properties of three SF-sBDEs (top panels) and three Q-sBDEs (bottom panels)
matched to MaNGA, overlapped on the relations of the Dn 4000 with EW(Hα) (left panel), EW(Hα) with EW(HδA ) (medial
panel), and EW(HδA ) with Dn 4000 (right panel). The stars show the measurements from SDSS spectra and the contours show
the spatial measurements from MaNGA in corresponding colors.

to 1.0% and 4.5%, respectively. It is worth emphasizing lower metallicity in the FMR relations (Figure 8). The
that ellipticals constitute only a small fraction in the environment of Hi-rich sBDEs also indicates that the
Hi sample, and their ratio to lenticulars is also lower abundant cold gas cannot be attributed solely to inter-
than in the optical sample. In this study, we focus on nal origin. Instead, it is the low-density and star-forming
a rare population, the Hi-rich sBDEs only, and study environmental conditions that are the key for Hi-rich sB-
their basic properties, to uncover the origin of their Hi DEs to acquire gas (Figure 10 and 11). Furthermore, we
gas and star formation. Despite machine learning clas- do indeed find several misaligned galaxies in our sample
sifying them as ellipticals, we refer to them as sBDEs from the MaNGA survey (Figure 12).
because we allow for the presence of low surface bright- Overall, these results suggest that the primary mech-
ness features at the outskirts of galaxies, distinguishing anisms for sBDEs to obtain gas are statistically external
them from normal ellipticals. mechanisms, such as gas accretion from the surround-
ing environment or wet mergers (Lagos et al. 2014, 2015;
5.1. Possible mechanisms to obtain cold gas in Hi-rich Griffith et al. 2019; Lee et al. 2023). However, internal
sBDEs mechanisms may also play a key role in some individual
galaxies, such as the two galaxies with unusually high
The MZR and two FMRs of sBDEs can provide in-
metallicity in both MZR and FMRs (Figure 8), which
sights into the origin of the gas. If gas is of internal ori-
would require further observational data and research.
gin, as the metallicity continuously increases due to star
formation, galaxies are expected to be located above the
MZR and FMRs. However, we find that most Hi-rich 5.2. Rejuvenated star formation in Hi-rich SF-sBDEs
sBDEs do not exhibit higher metallicity in the MZR, Some physical mechanisms could efficiently consume
and more importantly, most of them show significantly cold gas, expel gas from galaxies, and prevent the ac-
 17

cretion of gas from the environment. Once galaxies de- ure 10. Further observations using integral-field spec-
plete their cold gas, the high-density environment and troscopy are needed in the future. In our sBDEs sam-
AGNs feedback can play crucial roles in the continuous ple, the absence of star-forming disk typically found in
suppression of star formation (e.g., Giovanelli & Haynes lenticulars makes it more difficult to transport the an-
1983; Grossi et al. 2009; Peng & Renzini 2020; Kurinchi- gular momentum of gas outwards (Wang & Lilly 2022),
Vendhan et al. 2023). However, the sBDEs in our sam- resulting in the presence of so many Hi-rich quiescent
ple have sufficient gas, but most of them (70%) are still sBDEs. We hypothesize that, under certain conditions
classified as quiescent galaxies. AGNs do not have a sig- such as the disk instabilities caused by mergers or in-
nificant impact on our sample, and the emission lines of teractions, the gas in these galaxies would be induced
Q-sBDEs in our sample are primarily dominated by DIG to radial inflow and have the potential to revitalize star
(see Section 4.2 and Figure 7). Furthermore, although formation.
the reservoir of gas in sBDEs can be attributed to the
6. SUMMARY
galaxies being in an Hi-rich environment favorable for
gas accretion (see Figure 10), it does not seem to be a In this work, we have selected 104 Hi-rich strongly
key factor in determining star formation efficiency (see bulge-dominated ETGs (sBDEs) from the FASHI-
Figure 11). ALFALFA combined sample and classified them into
The distribution of gas in galaxies may play a crucial star-forming and quiescent sBDEs based on their SFR
role. We find that the stacked Hi profiles of SF-sBDEs and stellar mass. We have analyzed their basic galaxy
are narrower and more concentrated, while those of Q- properties using some Hi scaling relations and stel-
sBDEs are wider and similar to the quiescent control lar population scaling relations based on EW(Hα),
sample (Figure 9). Radial transport of gas is neces- EW(HδA ), and Dn 4000. To better understand the ori-
sary to sustain star formation in galaxies (Krumholz gin of Hi gas in sBDEs and the possible reason for star
et al. 2018). However, the Hi gas in these Q-sBDEs may formation in SF-sBDEs, we have also analyzed their gas-
form a ring-like structure in the outer region of galax- phase metallicity, stacked Hi spectra, environment, and
ies. Their high angular momentum makes it difficult for spatially resolved MaNGA data.
them to flow into galaxies (Renzini et al. 2018; Peng & Our main results are summarized as follows.
Renzini 2020; Lu et al. 2022). Meanwhile, these low- 1. The SF-sBDEs and Q-sBDEs exhibit a lower gas
density Hi gas outside the galaxies may not reach the fraction compared to normal Hi-rich galaxies in
threshold for transformation to H2 gas, and it is very the FASHI-ALFALFA combined Hi sample, where
inefficient to convert gas into stars (Bigiel et al. 2008; the latter typically represent the most late-type
Shi et al. 2011). galaxies. Furthermore, SF-sBDEs exhibit signif-
Combining our results with spatially resolved studies icantly higher SFEHi and shorter gas depletion
can help us better understand the properties of Hi-rich timescale (∼ 3 Gyr) compared to normal Hi-rich
sBDEs. Lee et al. (2023) used the data from Mapping galaxies.
Nearby Galaxies at APO (MaNGA) to investigate the
2. The distribution of SF-sBDEs on the EW(Hα)-
origin of star formation in ETGs. Our results of stacked
EW(HδA ) diagram is comparable to that of nor-
Hi spectra and gas-phase metallicity indicate that the
mal star-forming galaxies, but statistically have
distributions of gas and star formation differ from those
slightly larger values of Dn 4000. This is consistent
of star-forming LTGs. This is consistent with their find-
with the idea that the SF-sBDEs are composited
ings that star-forming ETGs show more concentrated
by an old stellar population as the normal ellipti-
ionized gas and different age gradients when compared
cals, as well as a young stellar population with an
to star-forming LTGs. Besides, Lee et al. (2023) also find
age of Gyr-timescale.
that more star-forming ETGs are classified as misalign-
ment or counter-rotators compared to other types of 3. Most Hi-rich sBDEs shows lower gas-phase metal-
galaxies. A misaligned stellar-gas disk could efficiently licity in MZR, FMRSFR and FMRHi based on
deplete the angular momentum of the external gas, driv- O3N2 calibrator, indicating the dilution of metal-
ing gas inflow to the central region of galaxies and trig- licity from the external metal-poor gas. When us-
gering star formation (Bao et al. 2022). The external ing N2O2 as the metallicity calibrator, these re-
origins, such as wet mergers or gas accretion, are likely sults are more significant because it is less influ-
to form the misaligned disk. In this case, our Hi-rich enced by DIG.
sBDEs are perfect candidates for the misaligned stellar-
4. From the analysis of stacked Hi spectra, SF-sBDEs
gas galaxies, as shown by the MaNGA image in Fig-
show narrower, more concentrated, and single-
18

peaked stacked profiles compared to the star- Start-up Fund of the University of Science and Tech-
forming control sample, while the stacked profiles nology of China (No. KY2030000200). Y.R. acknowl-
of Q-sBDEs are flatter compared to those of SF- edges supports from the CAS Pioneer Hundred Tal-
sBDEs. ents Program (Category B), as well as the USTC Re-
search Funds of the Double First-Class Initiative (grant
5. We have analyzed the large-scale environment us- No. YD2030002013)
ing the statistical average galaxies number and Funding for the Sloan Digital Sky Survey IV has been
quiescent fraction within certain projected physi- provided by the Alfred P. Sloan Foundation, the U.S.
cal distances for each sub-sample. The Hi-rich sB- Department of Energy Office of Science, and the Partic-
DEs are more commonly located in significantly ipating Institutions. SDSS- IV acknowledges support
lower-density environments and potentially have and resources from the Center for High-Performance
much more star-forming neighbors compared to Computing at the University of Utah. The SDSS web
typical ellipticals. site is www.sdss.org. SDSS-IV is managed by the As-
trophysical Research Consortium for the Participating
6. Three (in six) sBDEs in the MaNGA survey show Institutions of the SDSS Collaboration including the
kinematic misaligned gas-star rotation. This sup- Brazilian Participation Group, the Carnegie Institution
ports the conclusion of an external origin for the for Science, Carnegie Mellon University, the Chilean
gas in most of these galaxies. Participation Group, the French Participation Group,
Harvard-Smithsonian Center for Astrophysics, Instituto
Our results support that external mechanisms are the de Astrofı́sica de Canarias, The Johns Hopkins Univer-
dominant sources for Hi-rich sBDEs. SF-sBDEs are sity, Kavli Institute for the Physics and Mathematics of
likely the strong rejuvenating system with abundant Hi the Universe (IPMU) / University of Tokyo, Lawrence
reservoir and high SFEHi , while the gas in Q-sBDEs Berkeley National Laboratory, Leibniz Institut für As-
may be located at the outskirts of galaxies, resulting in trophysik Potsdam (AIP), Max-Planck-Institut für As-
very low star formation efficiency. We plan to look for tronomie (MPIA Heidelberg), Max-Planck-Institut für
Hi-rich sBDEs in cosmological hydrodynamical simula- Astrophysik (MPA Garching), Max-Planck-Institut für
tions such as IllustrisTNG (Nelson et al. 2019) to study Extraterrestrische Physik (MPE), National Astronom-
their formation and evolution. ical Observatory of China, New Mexico State Univer-
sity, New York University, University of Notre Dame,
ACKNOWLEDGEMENTS Observatório Nacional / MCTI, The Ohio State Univer-
This work is supported by the National Science Foun- sity, Pennsylvania State University, Shanghai Astronom-
dation of China (NSFC, Grant No. 12233008, 12125301, ical Observatory, United Kingdom Participation Group,
12192220, 12192222, 12273037), the National Key R&D Universidad Nacional Autónoma de México, University
Program of China (2023YFA1608100), the Strategic Pri- of Arizona, University of Colorado Boulder, University
ority Research Program of Chinese Academy of Sciences of Oxford, University of Portsmouth, University of Utah,
(Grant No. XDB 41000000), the China Manned Space University of Virginia, University of Washington, Uni-
Project (No. CMS-CSST-2021-A07), the Cyrus Chun versity of Wisconsin, Vanderbilt University, and Yale
Ying Tang Foundations, and the 111 Project for “Ob- University. Funding for the NASA-Sloan Atlas has been
servational and Theoretical Research on Dark Matter provided by the NASA Astrophysics Data Analysis Pro-
and Dark Energy” (B23042). EW thanks support of the gram (08-ADP08-0072) and the NSF (AST-1211644).

APPENDIX

A. THE METALLICITY PROPERTIES BASED ON N2O2 CALIBRATION OF Hi-RICH SBDES

In order to avoid contamination from DIG, we use the N2O2 calibrator from Sanders et al. (2018) to estimate the
gas-phase metallicity (Zhang et al. 2017; Kewley et al. 2019). Figure 14 provides the mass-metallicity relation using
the N2O2 system, but we only select 19 galaxies under the criteria of EW(Hα) > 3 Å and −1.3 < log(N2O2) < 0.0.
We can find that Hi-rich sBDEs show significantly lower metallicity.
 19

9.0 SF-sBDEs 9.0 SF-sBDEs 9.0 SF-sBDEs

12 + log(O/H) (N2O2) Q-sBDEs Q-sBDEs Q-sBDEs

12 + log(O/H) (N2O2)

12 + log(O/H) (N2O2)
8.8 8.8 8.8

8.6 8.6 8.6

8.4 8.4 8.4

8.2 8.2 8.2

8.0 8.0 8.0

8 9 10 11 8 9 10 11 8 9 10 11
log M* (M ) log M* 0.32 × log SFR log M* 0.35 × (log MHI 9.8)
Figure 14. Same as Figure 8, however using the N2O2 calibration from Sanders et al. (2018).

Table 1. Galaxy Properties for sBDEs.

SDSS ObjID Samplea RA (J2000) DEC (J2000) redshift log M∗ log SFR log MHI

(deg) (deg) (M⊙ ) (M⊙ yr−1 ) (M⊙ )

587730772817739967 ALFALFA 6.495 13.929 0.04196 10.7 -1.34 9.71
587724199351812182 ALFALFA 15.995 15.247 0.04174 10.81 0.73 10.08
587724197205835940 ALFALFA 19.445 13.323 0.04793 10.22 -1.19 10.12
587728906634592521 ALFALFA 114.032 30.215 0.02331 9.58 -0.32 9.22
587739153888837728 ALFALFA 117.195 16.646 0.05027 10.09 0.15 9.8
587739114696671327 ALFALFA 117.485 16.097 0.04085 10.67 -0.76 9.55
587731680111099933 ALFALFA 118.505 29.421 0.03609 10.38 -0.78 9.72
587732157389013064 ALFALFA 118.586 25.859 0.04164 10.5 0.35 9.73
587732157391306759 ALFALFA 122.823 29.509 0.01994 10.58 -1.61 9.89
587735236346707979 ALFALFA 124.994 26.386 0.01983 10.24 -0.53 9.56
587727944563622074 ALFALFA 133.559 2.241 0.05824 11.14 -1.06 10.12
587739116316786879 ALFALFA 137.096 28.269 0.04767 11.04 -0.79 10.1
587739158186754245 ALFALFA 138.827 28.315 0.04714 10.69 -1.01 9.95
588023046404636779 ALFALFA 139.805 19.702 0.02729 9.85 -1.08 9.47
587739377237622995 ALFALFA 140.004 26.739 0.04628 10.79 -1.22 9.76
587738409786671281 ALFALFA 146.054 11.146 0.03974 10.02 -1.28 9.74
587745245237805124 ALFALFA 147.771 17.056 0.02816 9.87 -1.21 9.41
587738616481185898 ALFALFA 149.721 31.62 0.02142 10.57 -1.41 9.61
587742567856472073 ALFALFA 149.799 16.478 0.02644 10.49 -1.41 9.66
587745402005422094 ALFALFA 150.209 13.737 0.03251 10.32 -1.57 9.83
587741532242444423 ALFALFA 155.197 26.112 0.03897 10.55 -1.31 9.76
587741816247812149 ALFALFA 155.407 21.687 0.03926 10.71 -1.33 9.78
587738411401478200 ALFALFA 155.514 13.769 0.01863 10.17 -1.51 9.53
588017702383648896 ALFALFA 156.796 10.025 0.03241 11.14 -0.99 9.91
587732703939461289 ALFALFA 159.805 7.858 0.03107 10.26 -1.46 9.23
587735349102182531 ALFALFA 161.762 14.263 0.03243 10.68 -1.38 9.74
588010359078912094 ALFALFA 163.269 4.493 0.04165 10.9 -1.26 10.18
587732578313175222 ALFALFA 167.82 7.336 0.04548 10.74 -1.26 10.2
587732771586375760 ALFALFA 168.481 9.643 0.04127 10.58 -0.73 9.95

Table 1 continued
20
Table 1 (continued)

SDSS ObjID Samplea RA (J2000) DEC (J2000) redshift log M∗ log SFR log MHI
−1
(deg) (deg) (M⊙ ) (M⊙ yr ) (M⊙ )
587741709955039271 ALFALFA 170.72 27.584 0.03341 10.95 -1.15 10.0
588017567632261212 ALFALFA 173.223 13.164 0.03452 10.12 -1.18 9.7
587741727111315572 ALFALFA 174.223 25.452 0.04923 10.84 -1.32 10.42
588010880367525939 ALFALFA 175.023 5.865 0.02103 9.46 -2.0 9.11
587742773483143242 ALFALFA 175.743 16.488 0.03444 11.12 -0.67 9.79
587739646211588152 ALFALFA 176.404 31.3 0.00593 9.35 -0.41 8.78
587735347500417057 ALFALFA 182.565 14.327 0.04299 10.24 0.23 9.93
587738570850304033 ALFALFA 183.279 16.086 0.02599 10.61 -1.29 9.64
588017726012194878 ALFALFA 187.503 8.104 0.03737 -1.0 -0.51 9.46
587729157889917059 ALFALFA 188.458 3.944 0.04391 9.92 -1.1 9.79
587741726043865103 ALFALFA 190.106 24.904 0.04743 10.7 -0.68 9.92
587729158966542513 ALFALFA 195.05 4.922 0.0479 10.19 -1.15 9.99
587736541473931384 ALFALFA 197.935 9.11 0.04472 11.05 -0.99 10.05
587741721753157723 ALFALFA 202.811 26.27 0.04594 10.2 -0.74 9.79
588010879306432606 ALFALFA 204.051 4.74 0.03434 10.74 -0.41 9.93
587739504477667416 ALFALFA 205.547 29.896 0.02727 10.33 -1.66 9.77
587736542015127621 ALFALFA 208.002 8.882 0.03766 10.64 0.85 9.79
587739132416557077 ALFALFA 208.297 35.579 0.03978 10.77 0.19 10.19
587726015081480261 ALFALFA 208.736 2.096 0.01466 9.04 -0.48 9.14
588017724947759115 ALFALFA 208.87 6.596 0.02408 9.85 0.43 9.58
587739708480815182 ALFALFA 208.884 28.648 0.03462 10.67 -1.53 9.97
587729160046510114 ALFALFA 209.281 5.252 0.03968 10.86 0.84 9.75
587736542553899149 ALFALFA 212.341 8.907 0.0235 10.51 -1.15 9.78
588017991230292116 ALFALFA 212.901 8.81 0.02283 10.77 -1.38 9.11
587726101484404918 ALFALFA 215.814 4.302 0.05654 10.77 -0.77 10.05
587742629060804612 ALFALFA 220.301 14.89 0.03053 10.42 -1.01 10.0
587729158441402460 ALFALFA 221.855 3.441 0.02742 10.53 -1.38 10.06
587742578068750468 ALFALFA 223.103 16.847 0.04779 10.65 -0.63 10.04
587736543096340508 ALFALFA 225.195 8.144 0.04771 10.6 -1.06 10.07
587736478137319778 ALFALFA 238.874 7.597 0.04498 10.1 -1.23 9.98
588017990705545693 ALFALFA 240.621 5.154 0.03756 10.29 -0.82 9.85
587742611349438852 ALFALFA 242.067 10.109 0.01572 9.31 -1.08 8.77
587742627998531885 ALFALFA 246.276 8.382 0.03511 10.31 -0.13 9.71
587736920509251921 ALFALFA 247.428 25.304 0.04274 10.15 -1.1 10.04
587727223015342192 ALFALFA 339.208 14.387 0.01752 10.19 -1.3 9.17
587739845383356595 ALFALFA 217.575 21.747 0.01783 10.35 0.19 9.79
587736940378128696 ALFALFA 231.89 28.852 0.03201 10.22 -0.83 9.41
587742615635165195 ALFALFA 239.518 15.317 0.03757 10.82 0.22 10.12
587736813137035343 ALFALFA 241.208 8.481 0.01746 10.41 -0.68 9.41
587735744230916340 ALFALFA 249.366 26.205 0.05217 10.91 -1.28 10.1
587741491448250465 ALFALFA & FASHI 175.016 30.593 0.03262 10.89 -1.33 10.17
587739458835054741 ALFALFA & FASHI 219.441 28.059 0.04408 10.74 -0.99 10.07
587724242306203664 FASHI 54.905 -5.633 0.02091 10.15 -0.13 9.56
587725470662590755 FASHI 116.16 38.717 0.0631 10.92 0.35 9.75

Table 1 continued
 21
Table 1 (continued)

SDSS ObjID Samplea RA (J2000) DEC (J2000) redshift log M∗ log SFR log MHI
−1
(deg) (deg) (M⊙ ) (M⊙ yr ) (M⊙ )
587725551190802727 FASHI 113.439 36.679 0.01555 9.46 -0.36 9.61
587728932955291700 FASHI 138.301 51.281 0.06561 10.83 0.62 10.03
587729387155357807 FASHI 195.527 59.721 0.02729 10.13 -1.0 9.31
587729407545967137 FASHI 256.103 33.376 0.02964 10.67 -1.25 9.31
587732471463411876 FASHI 133.298 37.135 0.04979 10.88 0.97 9.62
587733440511410223 FASHI 237.067 42.931 0.03563 10.57 -1.42 9.7
587733605328093283 FASHI 210.019 58.37 0.02629 10.34 -1.24 8.96
587735429621743636 FASHI 212.307 46.242 0.03997 9.63 -0.01 9.78
587735490823913594 FASHI 229.161 41.03 0.03134 10.98 -1.16 8.94
587735662622670894 FASHI 148.994 37.156 0.01706 10.09 -1.57 9.44
587735665312399516 FASHI 232.087 46.187 0.03757 9.99 -1.17 9.61
587735743690244302 FASHI 241.368 31.526 0.05645 10.45 0.43 9.72
587738195043745901 FASHI 126.07 57.288 0.02625 10.64 -1.47 9.62
587738196651540694 FASHI 118.34 52.408 0.06366 10.8 -1.12 10.12
587738574068187225 FASHI 193.165 40.103 0.02624 10.76 -0.9 9.67
587738946146861087 FASHI 184.733 39.14 0.03694 10.98 -1.36 9.41
587739407320219766 FASHI 188.271 35.025 0.03219 9.98 0.26 9.5
587739407850668215 FASHI 170.462 34.628 0.0417 10.35 -1.51 9.48
587739406236188869 FASHI 160.366 31.703 0.03493 10.62 -1.1 9.5
587739721903374358 FASHI 199.985 30.119 0.02342 10.53 -0.78 9.54
588007004162293771 FASHI 127.872 48.626 0.05313 11.08 0.05 9.74
588013381666734111 FASHI 171.736 50.597 0.02321 10.59 -1.6 9.98
588017605209751562 FASHI 149.925 39.657 0.0285 10.49 -1.48 9.3
588017625628409940 FASHI 215.229 40.121 0.01754 9.99 -0.34 8.6
588017626691207193 FASHI 181.546 45.149 0.06654 10.97 1.19 10.26
588017721717555362 FASHI 181.072 42.995 0.07332 10.96 0.79 10.28
588017948277473344 FASHI 182.406 42.319 0.02256 9.74 -0.11 8.78
588017977289474168 FASHI 205.272 37.077 0.03435 10.29 -1.62 9.32
588017979438137462 FASHI 209.135 38.043 0.05013 10.29 0.45 9.74
588018254829977709 FASHI 236.425 40.212 0.02578 9.81 -0.03 9.43
588295842857287748 FASHI 170.059 50.025 0.02601 10.22 -1.49 8.64

a
Notes. The sources detected by both ALFALFA and FASHI sample used ALFALFA measurements.

REFERENCES
Andrews, B. H., & Martini, P. 2013, ApJ, 765, 140, Balogh, M. L., Morris, S. L., Yee, H. K. C., Carlberg, R. G.,
doi: 10.1088/0004-637X/765/2/140 & Ellingson, E. 1999, ApJ, 527, 54, doi: 10.1086/308056
Ashley, T., Marcum, P. M., Alpaslan, M., Fanelli, M. N., & Balogh, M. L., Schade, D., Morris, S. L., et al. 1998, ApJL,
504, L75, doi: 10.1086/311576
Frost, J. D. 2019, AJ, 157, 158,
Bao, M., Chen, Y., Zhu, P., et al. 2022, ApJL, 926, L13,
doi: 10.3847/1538-3881/ab09f3
doi: 10.3847/2041-8213/ac52ad
Baker, J. G., & Menzel, D. H. 1938, ApJ, 88, 52,
Barrera-Ballesteros, J. K., Utomo, D., Bolatto, A. D., et al.
doi: 10.1086/143959 2020, MNRAS, 492, 2651, doi: 10.1093/mnras/stz3553
22

Belfiore, F., Maiolino, R., & Bothwell, M. 2016, MNRAS, Ellison, S. L., Patton, D. R., Simard, L., & McConnachie,
455, 1218, doi: 10.1093/mnras/stv2332 A. W. 2008, ApJL, 672, L107, doi: 10.1086/527296
Bigiel, F., Leroy, A., Walter, F., et al. 2008, AJ, 136, 2846, Fabello, S., Catinella, B., Giovanelli, R., et al. 2011,
doi: 10.1088/0004-6256/136/6/2846 MNRAS, 411, 993, doi: 10.1111/j.1365-2966.2010.17742.x
Blanton, M. R., Kazin, E., Muna, D., Weaver, B. A., & Falcón-Barroso, J., Sarzi, M., Bacon, R., et al. 2006,
Price-Whelan, A. 2011, AJ, 142, 31, NewAR, 49, 515, doi: 10.1016/j.newar.2005.10.006
doi: 10.1088/0004-6256/142/1/31 Fang, J. J., Faber, S. M., Salim, S., Graves, G. J., & Rich,
Blanton, M. R., & Moustakas, J. 2009, ARA&A, 47, 159, R. M. 2012, ApJ, 761, 23,
doi: 10.1146/annurev-astro-082708-101734 doi: 10.1088/0004-637X/761/1/23
Bothwell, M. S., Maiolino, R., Kennicutt, R., et al. 2013, Fitzpatrick, E. L. 1999, PASP, 111, 63, doi: 10.1086/316293
MNRAS, 433, 1425, doi: 10.1093/mnras/stt817 Fumagalli, M., Patel, S. G., Franx, M., et al. 2012, ApJL,
Bower, R. G., Benson, A. J., Malbon, R., et al. 2006, 757, L22, doi: 10.1088/2041-8205/757/2/L22
MNRAS, 370, 645, doi: 10.1111/j.1365-2966.2006.10519.x Gao, Y., Wang, E., Kong, X., et al. 2018, ApJ, 868, 89,
Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, doi: 10.3847/1538-4357/aae9f1
MNRAS, 351, 1151, George, K. 2023, arXiv e-prints, arXiv:2308.12153,
doi: 10.1111/j.1365-2966.2004.07881.x doi: 10.48550/arXiv.2308.12153
Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000, Giovanelli, R., & Haynes, M. P. 1983, AJ, 88, 881,
doi: 10.1046/j.1365-8711.2003.06897.x doi: 10.1086/113376
Bundy, K., Bershady, M. A., Law, D. R., et al. 2015, ApJ, Giovanelli, R., Haynes, M. P., Kent, B. R., et al. 2005, 130,
798, 7, doi: 10.1088/0004-637X/798/1/7 2598, doi: 10.1086/497431
Cappellari, M. 2016, ARA&A, 54, 597, Governato, F., Gardner, J. P., Stadel, J., Quinn, T., &
doi: 10.1146/annurev-astro-082214-122432 Lake, G. 1999, AJ, 117, 1651, doi: 10.1086/300805
Catinella, B., Schiminovich, D., Kauffmann, G., et al. 2010, Griffith, E., Martini, P., & Conroy, C. 2019, MNRAS, 484,
MNRAS, 403, 683, doi: 10.1111/j.1365-2966.2009.16180.x 562, doi: 10.1093/mnras/sty3405
Catinella, B., Saintonge, A., Janowiecki, S., et al. 2018, Grønnow, A. E., Finlator, K., & Christensen, L. 2015,
MNRAS, 476, 875, doi: 10.1093/mnras/sty089 MNRAS, 451, 4005, doi: 10.1093/mnras/stv1232
Chen, X., Wang, J., & Kong, X. 2022, ApJ, 933, 39, Grossi, M., di Serego Alighieri, S., Giovanardi, C., et al.
doi: 10.3847/1538-4357/ac70d0 2009, A&A, 498, 407, doi: 10.1051/0004-6361/200810823
Cid Fernandes, R., Stasińska, G., Mateus, A., & Vale Asari, Haynes, M. P., Giovanelli, R., Martin, A. M., et al. 2011,
N. 2011, MNRAS, 413, 1687, AJ, 142, 170, doi: 10.1088/0004-6256/142/5/170
doi: 10.1111/j.1365-2966.2011.18244.x Haynes, M. P., Giovanelli, R., Kent, B. R., et al. 2018, ApJ,
Cortesi, A., Merrifield, M. R., Coccato, L., et al. 2013, 861, 49, doi: 10.3847/1538-4357/aac956
MNRAS, 432, 1010, doi: 10.1093/mnras/stt529 Höfner, S., & Olofsson, H. 2018, A&A Rv, 26, 1,
Croton, D. J., Springel, V., White, S. D. M., et al. 2006, doi: 10.1007/s00159-017-0106-5
MNRAS, 365, 11, doi: 10.1111/j.1365-2966.2005.09675.x Houck, J. C., & Bregman, J. N. 1990, ApJ, 352, 506,
Davis, T. A., & Bureau, M. 2016, MNRAS, 457, 272, doi: 10.1086/168554
doi: 10.1093/mnras/stv2998 Huang, C., Zou, H., Kong, X., et al. 2019, ApJ, 886, 31,
Davis, T. A., & Young, L. M. 2019, MNRAS, 489, L108, doi: 10.3847/1538-4357/ab4902
doi: 10.1093/mnrasl/slz138 Jiang, P., Yue, Y., Gan, H., et al. 2019, Science China
De Looze, I., Lamperti, I., Saintonge, A., et al. 2020, Physics, Mechanics, and Astronomy, 62, 959502,
MNRAS, 496, 3668, doi: 10.1093/mnras/staa1496 doi: 10.1007/s11433-018-9376-1
Dey, A., Schlegel, D. J., Lang, D., et al. 2019, AJ, 157, 168, Jimmy, Tran, K.-V., Saintonge, A., et al. 2015, ApJ, 812,
doi: 10.3847/1538-3881/ab089d 98, doi: 10.1088/0004-637X/812/2/98
di Serego Alighieri, S., Gavazzi, G., Giovanardi, C., et al. Jones, M. G., Haynes, M. P., Giovanelli, R., & Moorman,
2007, A&A, 474, 851, doi: 10.1051/0004-6361:20078205 C. 2018, MNRAS, 477, 2, doi: 10.1093/mnras/sty521
Domı́nguez Sánchez, H., Huertas-Company, M., Bernardi, Kauffmann, G., Heckman, T. M., White, S. D. M., et al.
M., Tuccillo, D., & Fischer, J. L. 2018, MNRAS, 476, 2003, MNRAS, 341, 33,
3661, doi: 10.1093/mnras/sty338 doi: 10.1046/j.1365-8711.2003.06291.x
Drory, N., MacDonald, N., Bershady, M. A., et al. 2015, Kennicutt, Robert C., J. 1992, ApJ, 388, 310,
AJ, 149, 77, doi: 10.1088/0004-6256/149/2/77 doi: 10.1086/171154
 23

Kewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, Moran, S. M., Loh, B. L., Ellis, R. S., et al. 2007, ApJ, 665,
C. A., & Trevena, J. 2001, ApJ, 556, 121, 1067, doi: 10.1086/519550
doi: 10.1086/321545 Nan, R., Li, D., Jin, C., et al. 2011, International Journal of
Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. Modern Physics D, 20, 989,
2006, MNRAS, 372, 961, doi: 10.1142/S0218271811019335
doi: 10.1111/j.1365-2966.2006.10859.x Nelson, D., Springel, V., Pillepich, A., et al. 2019,
Kewley, L. J., Nicholls, D. C., & Sutherland, R. S. 2019, Computational Astrophysics and Cosmology, 6, 2,
ARA&A, 57, 511, doi: 10.1186/s40668-019-0028-x
doi: 10.1146/annurev-astro-081817-051832 Oosterloo, T., Morganti, R., Crocker, A., et al. 2010,
Knapp, G. R., Turner, E. L., & Cunniffe, P. E. 1985, AJ, MNRAS, 409, 500, doi: 10.1111/j.1365-2966.2010.17351.x
90, 454, doi: 10.1086/113751 Pandey, D., Kaviraj, S., Saha, K., & Sharma, S. 2024,
Köppen, J., & Edmunds, M. G. 1999, MNRAS, 306, 317, arXiv e-prints, arXiv:2403.12160,
doi: 10.1046/j.1365-8711.1999.02584.x doi: 10.48550/arXiv.2403.12160
Krumholz, M. R., Burkhart, B., Forbes, J. C., & Crocker, Park, C., Gott, J. Richard, I., & Choi, Y.-Y. 2008, ApJ,
R. M. 2018, MNRAS, 477, 2716, 674, 784, doi: 10.1086/524192
doi: 10.1093/mnras/sty852 Paudel, S., Yoon, S.-J., Bait, O., et al. 2023, ApJL, 951,
Kurinchi-Vendhan, S., Farcy, M., Hirschmann, M., & L36, doi: 10.3847/2041-8213/acdef1
Valentino, F. 2023, arXiv e-prints, arXiv:2310.03083, Peng, Y., Maiolino, R., & Cochrane, R. 2015, Nature, 521,
doi: 10.48550/arXiv.2310.03083 192, doi: 10.1038/nature14439
Lagos, C. d. P., Davis, T. A., Lacey, C. G., et al. 2014, Peng, Y.-j., Lilly, S. J., Renzini, A., & Carollo, M. 2012,
MNRAS, 443, 1002, doi: 10.1093/mnras/stu1209 ApJ, 757, 4, doi: 10.1088/0004-637X/757/1/4
Lagos, C. d. P., Padilla, N. D., Davis, T. A., et al. 2015, Peng, Y.-j., & Renzini, A. 2020, MNRAS, 491, L51,
MNRAS, 448, 1271, doi: 10.1093/mnras/stu2763 doi: 10.1093/mnrasl/slz163
Le Borgne, D., Abraham, R., Daniel, K., et al. 2006, ApJ, Peng, Y.-j., Lilly, S. J., Kovač, K., et al. 2010, ApJ, 721,
642, 48, doi: 10.1086/500005 193, doi: 10.1088/0004-637X/721/1/193
Lee, Y. H., Hwang, H. S., Hwang, N., Lee, J. C., & Kim, Phillips, M. M., Jenkins, C. R., Dopita, M. A., Sadler,
K.-B. 2023, ApJ, 953, 88, doi: 10.3847/1538-4357/ace1ea E. M., & Binette, L. 1986, AJ, 91, 1062,
Lequeux, J., Peimbert, M., Rayo, J. F., Serrano, A., & doi: 10.1086/114083
Torres-Peimbert, S. 1979, A&A, 80, 155 Poetrodjojo, H., Groves, B., Kewley, L. J., et al. 2018,
Liu, C., Hao, L., Wang, H., & Yang, X. 2019, ApJ, 878, 69, MNRAS, 479, 5235, doi: 10.1093/mnras/sty1782
doi: 10.3847/1538-4357/ab1ea0 Qian, L., Yao, R., Sun, J., et al. 2020, The Innovation, 1,
Lu, S., Xu, D., Wang, S., et al. 2022, MNRAS, 509, 2707, 100053, doi: 10.1016/j.xinn.2020.100053
doi: 10.1093/mnras/stab3169 Renzini, A., Gennaro, M., Zoccali, M., et al. 2018, ApJ,
Mager, V. A., Conselice, C. J., Seibert, M., et al. 2018, 863, 16, doi: 10.3847/1538-4357/aad09b
ApJ, 864, 123, doi: 10.3847/1538-4357/aad59e Sadler, E. M., Oosterloo, T., & Morganti, R. 2002, in
Mannucci, F., Cresci, G., Maiolino, R., Marconi, A., & Astronomical Society of the Pacific Conference Series,
Gnerucci, A. 2010, MNRAS, 408, 2115, Vol. 273, The Dynamics, Structure & History of
doi: 10.1111/j.1365-2966.2010.17291.x Galaxies: A Workshop in Honour of Professor Ken
Marino, R. A., Rosales-Ortega, F. F., Sánchez, S. F., et al. Freeman, ed. G. S. Da Costa, E. M. Sadler, & H. Jerjen,
2013, A&A, 559, A114, 215, doi: 10.48550/arXiv.astro-ph/0205151
doi: 10.1051/0004-6361/201321956 Saintonge, A., & Catinella, B. 2022, ARA&A, 60, 319,
McDermid, R. M., Alatalo, K., Blitz, L., et al. 2015, doi: 10.1146/annurev-astro-021022-043545
MNRAS, 448, 3484, doi: 10.1093/mnras/stv105 Salim, S., Boquien, M., & Lee, J. C. 2018, ApJ, 859, 11,
Meert, A., Vikram, V., & Bernardi, M. 2015, MNRAS, 446, doi: 10.3847/1538-4357/aabf3c
3943, doi: 10.1093/mnras/stu2333 Salim, S., Lee, J. C., Janowiecki, S., et al. 2016, ApJS, 227,
Meyer, M. J., Zwaan, M. A., Webster, R. L., et al. 2004, 2, doi: 10.3847/0067-0049/227/1/2
MNRAS, 350, 1195, Sanders, R. H., & Tubbs, A. D. 1980, ApJ, 235, 803,
doi: 10.1111/j.1365-2966.2004.07710.x doi: 10.1086/157683
Moore, B., Katz, N., Lake, G., Dressler, A., & Oemler, A. Sanders, R. L., Shapley, A. E., Kriek, M., et al. 2018, ApJ,
1996, Nature, 379, 613, doi: 10.1038/379613a0 858, 99, doi: 10.3847/1538-4357/aabcbd
24

Schawinski, K., Thomas, D., Sarzi, M., et al. 2007, MNRAS, Wang, J., Catinella, B., Saintonge, A., et al. 2020, ApJ,
382, 1415, doi: 10.1111/j.1365-2966.2007.12487.x 890, 63, doi: 10.3847/1538-4357/ab68dd
Serra, P., Oosterloo, T., Morganti, R., et al. 2012, MNRAS, Wang, J., Staveley-Smith, L., Westmeier, T., et al. 2021,
422, 1835, doi: 10.1111/j.1365-2966.2012.20219.x ApJ, 915, 70, doi: 10.3847/1538-4357/abfc52
Serra, P., Oser, L., Krajnović, D., et al. 2014, MNRAS, 444, Westfall, K. B., Cappellari, M., Bershady, M. A., et al.
3388, doi: 10.1093/mnras/stt2496 2019, AJ, 158, 231, doi: 10.3847/1538-3881/ab44a2
Wetzel, A. R., Tinker, J. L., & Conroy, C. 2012, MNRAS,
Shi, Y., Helou, G., Yan, L., et al. 2011, ApJ, 733, 87,
424, 232, doi: 10.1111/j.1365-2966.2012.21188.x
doi: 10.1088/0004-637X/733/2/87
Young, L. M., Bureau, M., Davis, T. A., et al. 2011,
Strateva, I., Ivezić, Ž., Knapp, G. R., et al. 2001, AJ, 122,
MNRAS, 414, 940, doi: 10.1111/j.1365-2966.2011.18561.x
1861, doi: 10.1086/323301
Young, L. M., Scott, N., Serra, P., et al. 2014, MNRAS,
Tacchella, S., Dekel, A., Carollo, C. M., et al. 2016, 444, 3408, doi: 10.1093/mnras/stt2474
MNRAS, 457, 2790, doi: 10.1093/mnras/stw131 Yu, N., Ho, L. C., Wang, J., & Li, H. 2022, ApJS, 261, 21,
Tamburri, S., Saracco, P., Longhetti, M., et al. 2014, A&A, doi: 10.3847/1538-4365/ac626b
570, A102, doi: 10.1051/0004-6361/201424040 Zhang, C., Peng, Y., Ho, L. C., et al. 2019, ApJL, 884, L52,
Thomas, D., Maraston, C., Schawinski, K., Sarzi, M., & doi: 10.3847/2041-8213/ab4ae4
Silk, J. 2010, MNRAS, 404, 1775, Zhang, C.-P., Zhu, M., Jiang, P., et al. 2023a, arXiv
doi: 10.1111/j.1365-2966.2010.16427.x e-prints, arXiv:2312.06097.
Wang, E., & Lilly, S. J. 2020, ApJ, 892, 87, https://arxiv.org/abs/2312.06097
doi: 10.3847/1538-4357/ab7b7d Zhang, J., Li, Y., Leja, J., et al. 2023b, ApJ, 952, 6,
—. 2022, ApJ, 927, 217, doi: 10.3847/1538-4357/ac49ed doi: 10.3847/1538-4357/acd84a
Zhang, K., Yan, R., Bundy, K., et al. 2017, MNRAS, 466,
Wang, E., Li, C., Xiao, T., et al. 2018, ApJ, 856, 137,
3217, doi: 10.1093/mnras/stw3308
doi: 10.3847/1538-4357/aab263