The Astrophysical Journal, 961:253 (15pp), 2024 February 1 https://doi.org/10.

3847/1538-4357/ad15fd
© 2024. The Author(s). Published by the American Astronomical Society.

Galaxy Clusters from the DESI Legacy Imaging Surveys. II. Environmental Effects on
the Size–Mass Relation
Zhaobin Chen1 , Yizhou Gu2 , Hu Zou3 , and Qirong Yuan1
1
Schoolof Physics and Technology, Nanjing Normal University, No. 1, Wenyuan Road, Nanjing, 210023, People’s Republic of China; [email protected]
2
Department of Astronomy, School of Physics and Astronomy, and Shanghai Key Laboratory for Particle Physics and Cosmology, Shanghai Jiao Tong University,
Shanghai 200240, People’s Republic of China
3
Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, People’s Republic of China
Received 2023 August 18; revised 2023 December 1; accepted 2023 December 13; published 2024 January 31

Abstract
To investigate the environmental effects on the growth of galaxies, we study the size–mass relation across a broad
range of environments with a vast sample of approximately 32 million galaxies at z < 0.5 from the DESI Legacy
Imaging Surveys. This sample is divided into three subsamples representing galaxies within three different
environments: brightest cluster galaxies (BCGs), other cluster galaxies, and field galaxies. The BCGs in our large
sample are dominated by quiescent galaxies (QGs), while only a minority (∼13%) of BCGs are star-forming
galaxies (SFGs). To demonstrate the influence of the environment on size growth, we attempt to observe the
difference in size–mass relation for these three subsamples. In general, the slope differences between QGs and
SFGs within various environments are significant and tend to be greater at higher redshifts. For the mass-complete
subsamples at z < 0.5, BCGs are found to have the highest slope of size–mass relation, and no difference in size–
mass relation is found between cluster members and field galaxies. To assess whether the observed slope
differences stem from the variations in environment or mass distribution, we construct the mass-matched
subsamples for QGs and SFGs. As a result, both QGs and SFGs show negligible differences in the slope of the
size–mass relation among the galaxies within three distinct environments, indicating that stellar mass is the most
fundamental factor driving the size evolution at z < 0.5, though the mass growth mode for QGs and SFGs may
have been affected by galaxy environment.
Unified Astronomy Thesaurus concepts: Galaxy evolution (594); Galaxy environments (2029); Brightest cluster
galaxies (181)

1. Introduction et al. 2022). According to Bluck et al. (2012), mergers are

In the standard hierarchical model, weak density perturba- considered to be the predominant factor driving the observed
tions in the early Universe lead to the gradual collapse of size evolution in massive galaxies below z ∼ 3, resulting in a
dissipationless dark matter into dense dark matter halos (Mo threefold increase in mass from z = 3 to the present day solely
et al. 1998). Within these halos, primitive gas gradually cools through merging. Numerous investigations suggest that minor
to form gaseous nebulae, which, under the influence of mergers may have a noteworthy impact on the growth of
gravitational potential wells, collapse into dense, luminous quiescent galaxies (e.g., Naab et al. 2007; Greene et al. 2012;
nuclei, or galaxies (e.g., White & Rees 1978; White & Newman et al. 2012; van Dokkum et al. 2015; Whitney et al.
Frenk 1991). As galaxies evolve, collisions and mergers occur, 2019). In addition, McLure et al. (2013) report compelling
resulting in the formation of larger galaxies. Throughout this evidence supporting the notion that minor mergers play a
process, the angular momentum provided by dark matter halos crucial role in the evolution of massive galaxies at
dominates the size of galaxies (e.g., Mo et al. 1998; 1.3 < z < 1.5. van Dokkum et al. (2010) found that the mass
Kravtsov 2013; Huang et al. 2017). Thus, tracking the in the central regions of massive galaxies remains roughly
coevolution of galaxy size and mass with cosmic time provides constant with redshift, and their growth since z ∼ 2 is only due
insight into the assembly history of galaxies. to the gradual increase in their outer envelopes.
Stellar mass, as an important factor in the growth of galaxy Williams et al. (2009) suggest that the bimodal distribution
size, has been extensively studied (e.g., Shen et al. 2003; of galaxies in color space can be traced back to z ∼ 2. In
Williams et al. 2010; Shibuya et al. 2015). Stellar mass general, star-forming galaxies (SFGs) with a bluer color, whose
assembly is a crucial process in galaxy evolution, which is mass growth is dominated by star formation, become quiescent
influenced by intricate factors, such as star formation and galaxies (QGs) with a redder color after quenching (e.g.,
mergers (e.g., Ilbert et al. 2010; van Dokkum et al. 2010; Webb Schawinski et al. 2014; Bait et al. 2017; Gu et al. 2019). It is
et al. 2015). Usually, massive galaxies form earlier than lower- widely believed that the star formation activity in galaxies is
mass galaxies (e.g., Gu et al. 2018; Khullar et al. 2022), closely related to their morphologies (Calvi et al. 2018;
although some studies on individual galaxies suggest some- Nersesian et al. 2019). Morphologically speaking, SFGs with
what different histories of stellar mass assembly (e.g., Webb
disk-dominated morphologies, which are inclined to late-type
galaxies, are considered to be growing inside-out through
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further
continuous star formation (e.g., Pichon et al. 2011; Bird et al.
distribution of this work must maintain attribution to the author(s) and the title 2013). However, QGs are inclined to early-type galaxies with
of the work, journal citation and DOI. elliptical or bulge-dominated morphologies. According to the

1
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

work of Paspaliaris et al. (2023), only 8% of late-type galaxies insights, such as the study by Ascaso et al. (2011) on the
are quiescent. evolution of BCGs over the past approximately 6 Gyr. They
The correlation between galaxy morphology and environ- believe that the only mechanism that is able to explain the
mental density has long been established (e.g., Oemler 1974; evolution in BCG size and the nonevolution in the BCG Sérsic
Dressler 1980). The divergence in the environmental distribu- shape parameter during the last 6 Gyr are the feedback
tion of blue and red galaxies appears since z < 1.5 (Gu et al. processes (such as galactic winds and ejection).
2018). Field and cluster galaxies represent two distinct Taking into account the environment as a variable in our
populations residing in markedly different environments, and study, we attempt to investigate whether it plays a significant
extensive research has been conducted to explore the processes role in the variation of the size–mass relation (Re µ M a ). In
*
of star formation and mass assembly within each of these this paper, we select three subsamples for our analysis,
populations (e.g., Bundy et al. 2006; Kim et al. 2023). From a including field galaxies, member galaxies, and BCGs. Based
theoretical perspective, cluster galaxies in denser environments on the ninth publicly released data of the Dark Energy
suffer more effects of the dense and hot intracluster medium Spectroscopic Instrument (DESI) Legacy Imaging Surveys, we
than field galaxies (e.g., Gunn & Gott 1972; Bekki & can obtain more than 32 million galaxies with a sky area of
Couch 2003). Indeed, Yoon et al. (2017) find that local ∼20,000 deg2 in the redshift range of 0 < z < 0.5. The
early-type galaxies in the densest environment with stellar mass influence of the environment on size–mass relation is analyzed.
greater than 1011.2Me tend to have larger sizes than those in This paper is the second in our series of works (the first
low-density environments. Some researchers have suggested study has already been completed by Zou et al. 2021) and is
that cluster galaxies may be slightly smaller when compared to organized as follows. In Section 2, we describe the catalog of
their counterparts in the field (e.g., Poggianti et al. 2013; data from the DESI Legacy Imaging Surveys and the detailed
Cebrián & Trujillo 2014). Despite the different conclusions process to construct samples. In Section 3, we analyze the
reached by these studies, they all share the common idea that detailed influence of stellar mass and environment on the
galaxy morphology is closely related to the environment. size–mass relation. Section 4 presents our discussions about
Nevertheless, Saracco et al. (2017) demonstrate that at the dependence of the size–mass relation upon environment
approximately z = 1.3, the structures and intrinsic properties and redshift. The summary and conclusion are presented in
(namely, the equivalent median effective radius and radius Section 5. Throughout our paper, we adopt the following
normalized mass) of cluster and field elliptical galaxies are cosmological parameters: Ωm = 0.30, ΩΛ = 0.70, and H0 =
comparable. Similarly, Weinmann et al. (2009) find that no 70 km s−1 Mpc−1.
differences in radius are observed between early-type central
and satellite galaxies. The analysis of Huertas-Company et al. 2. Data and Sample Selection
(2013a) also indicates no significant environmental dependence
2.1. Data Description
for the sizes of central and satellite early-type galaxies at fixed
stellar mass at z ∼ 0. The ninth public data release (DR9) of the Legacy Imaging
Among the member galaxies of the cluster, the brightest Surveys of DESI covers about 20,000 square degrees in both
cluster galaxies (BCGs) are expected to have a more complex the South and North Galactic Caps, which provides imaging in
evolution process due to the association with the dense g/r/z bands with 5σ depth of 24.7/23.9/23.0 (e.g., Dey et al.
environment at the center of the cluster. Generally, BCGs are 2019; Zou et al. 2022). It consists of three independent optical
found close to the center of galaxy clusters with the densest surveys using three different telescopes: the Beijing-Arizona
environment, which has been determined by X-ray or Sky Survey (BASS; Zou et al. 2017), the Mayall z-band Legacy
gravitational lens observations (e.g., Jones & Forman 1984; Survey (MzLS), and the DECam Legacy Survey (DECaLS). In
Smith et al. 2005). Compared with member and field galaxies, addition, the imaging surveys provide near-infrared photo-
BCGs may have a more complex pattern of mass growth (e.g., metric data in the 3.4 and 4.6 μm Wide-field Infrared Survey
Kluge et al. 2020). Simulations and semianalytic models show Explorer (WISE) bands (i.e., WISE W1 and W2; see Wright
that BCGs assemble most of their stellar mass through multiple et al. 2010). This data release also includes the 6 yr imaging of
mergers (e.g., Dubinski 1998; De Lucia & Blaizot 2007). WISE, with 5σ depths of 20.7 and 20.0 in W1 and W2 bands,
Observational studies also suggest that merger plays an respectively.
important role in the evolution of BCGs (e.g., Lidman et al. Zou et al. (2022) expand their prior research on photometric
2013; Lavoie et al. 2016). Further investigations on the stellar redshift and cluster detection to the DR9 of DESI Legacy
mass evolution of BCGs show that their mass increases as Imaging Surveys, generating a sample of galaxy clusters over
redshift decreases (e.g., Lidman et al. 2012), while others 540,000 in the redshift range of 0 < z < 1. We start from the
maintain that the stellar mass has remained constant since at magnitude-limited sample to identify the galaxy clusters in
least z ∼ 1 (e.g., Whiley et al. 2008; Ascaso et al. 2011). The Section 2.3, using the following criteria listed in Zou et al.
work of Webb et al. (2015) demonstrates that the evolutionary (2022) on the DR9 of the legacy imaging surveys:
pattern of stellar mass varies with redshift. Specifically, when 1. r < 23 (the upper limit of r-band magnitude for excluding
redshifts exceed 1, the dominant force driving the growth of faint sources).
stellar mass may be the process of star formation, whereas at 2. type !=“PSF” (the “PSF” refers to a point source).
lower redshifts (z < 1), dry mergers are likely to play a more 3. fracmasked_[g, r, z] < 0.5 (clean photometric cuts using
significant role in this phenomenon. Some studies suggest that the profile-weighted fraction of pixels masked).
the growth rate of BCG mass may exhibit a slowdown, with 4. fracflux_[g, r, z] < 0.5 (clean photometric cuts using the
more rapid growth observed at z > 0.5 and slower growth at profile-weighted fraction of the flux).
z < 0.5 (e.g., Lin et al. 2013; Oliva-Altamirano et al. 2014). In 5. fracin_[g, r, z] > 0.3 (clean photometric cuts using the
addition, diverse perspectives in research provide unique fraction of a source’s flux within the blob).

2
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

6. it is detected in the five bands (g, r, z, W1, W2).

7. the photo-z error is set to less than 0.1×(1 + zphot).
8. the range of the r-band absolute magnitude is −25 <
Mr < − 16.
9. the stellar mass range is 106 < M*/Me < 1013.
10. the uncertainty of logarithmic stellar mass ∣D log
(M* M)∣ < 0.4 dex.
The photometric redshifts, stellar masses, colors, and star
formation rates (SFRs) are taken from the catalog of Zou et al.
(2019), which is updated to the DESI DR9 release (Zou et al.
2022).
Using a hybrid technique to estimate photometric redshift
(zphoto), a local linear regression algorithm is applied to a
dedicated spectroscopic training set, and then a spectrum
template is expertly fitted to determine k-corrections and
absolute magnitudes (Beck et al. 2016; Gao et al. 2018). The
stellar mass (M*) and SFR are derived by fitting the
photometric spectral energy distribution (SED) of galaxies
using the LePhare software4 (Arnouts et al. 2002; Ilbert et al.
2006). The default Bruzual & Charlot (2003; BC03) spectral
models are employed, in conjunction with the initial mass
function from Chabrier (2003). Three metallicities (0.004,
0.008, and 0.02) and an exponentially declining star formation
Figure 1. Illustrated in the figure is the distribution of galaxy cluster richness,
history (SFH) from 0.1 to 30 Gyr are taken. For more detailed where N1 Mpc signifies the richness of the cluster. The vertical black dashed
photometric redshift and stellar mass information, please refer line denotes the peak of the distribution at ∼18.
to Zou et al. (2019).
Throughout the paper, the size of the galaxy is determined Figure 1 displays the histogram distribution of the richness
by the half-light radius (Re) of the galaxy model with the N1 Mpc for galaxy clusters. The distribution peaks at around 18,
Tractor. The Tractor (Lang et al. 2016) performs profile-fitting which suggests that galaxy clusters with N1 Mpc > 18 can be
photometry. By assuming the same model across all the bands,
considered as rich clusters with high completeness.
model fits are determined using only the optical g/r/z data with
The catalogs of photometric redshifts and clusters used
the specific point-spread function (PSF). Four types of galaxy
in this paper are available through the ScienceDB Web site
profiles are used in the fitting procedure of DR9: round
(doi:10.11922/sciencedb.o00069.00003).
exponential galaxies with a variable radius (“REX”), de
Vaucouleurs (“DEV”) profiles (elliptical galaxies), exponential
(“EXP”) profiles (spiral galaxies), and Sérsic (“SER”) profiles.
The best-fit model is determined by convolving each model 2.3. Sample Selection
with the specific PSF for each band and minimizing the We select three distinct subsamples from the galaxy cluster
residuals. catalog, representing three different environments. These
subsamples include BCGs located in the densest environmental
region of the cluster, member galaxies inhabiting regions of
2.2. Galaxy Clusters environmental density second only to BCGs, and field galaxies
Previously, Gao et al. (2020) developed a rapid clustering isolated within low-density environments. Our approach to
algorithm called clustering by fast search and find of density constructing these subsamples is elaborated below:
peaks (CFSFDP) to identify galaxy clusters and their members
from the photometric redshift catalog. The key aspect of this 1. We start with the sample from Zou et al. (2022), selecting
algorithm lies in providing a simple and efficient approach for galaxies at z < 0.5. The upper limit of redshift 0.5 is
cluster identification, which can quickly identify galaxy cluster ultimately determined by the fact that the corresponding
centers exhibiting higher density compared to their neighboring physical size for a typical point-spread function in the
regions. This approach successfully eliminates the distorting r-band images (~0.¢¢8) is about 5 kpc at z = 0.5. If
influence of foreground and background galaxies. Using this spectroscopic redshift (zspec) is available, we use it
algorithm on their sample, Zou et al. (2021) identified 540,432 preferentially; Otherwise, photometric redshift would
galaxy clusters at z  1 from the DESI legacy imaging surveys. be used.
As the galaxy closest to the position of the density peak is not 2. To form a subsample of field galaxies, we removed all
always the BCG, they identified the r-band brightest galaxy cluster galaxies, including both BCGs and member
within 0.5 Mpc around the density peak as the BCG. galaxies.
For each galaxy cluster, the number of member galaxies 3. Taking into account the completeness of the galaxy
within a distance of 1 Mpc from its center (N1 Mpc) is used as cluster, clusters with a richness greater than 18 (i.e.,
an estimate of the cluster richness, after correcting for the N1 Mpc > 18) are selected. Once the clusters are selected,
number of galaxies expected from the background density. we extract the corresponding BCGs and member galaxies
to construct the subsamples of BCGs and member
4
https://www.cfht.hawaii.edu/arnouts/LEPHARE/lephare.html galaxies. To reduce the degree of contamination in

3
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 2. Rest-frame color-redshift diagram for galaxies with redshifts in the range of 0 < z < 0.5. The solid black line represents the BC03 template with τ = 2 Gyr.
Left panel: distribution of the sample of quiescent galaxies, which are selected using the sSFR criterion (−15 < logsSFR < − 11). Galaxies above the solid black line
are classified as QGs. Right panel: distribution of the sample of star-forming galaxies, which are selected using the sSFR criterion (−11 < logsSFR < − 7). Galaxies
below the solid black line are classified as SFGs. The color bar represents the number of galaxies.

membership, we select the galaxies within a projected of red galaxies are found to have −11 < logsSFR < − 7, and
distance of 0.5 Mpc from the BCGs as member galaxies. about 7.6% of blue galaxies have −15 < logsSFR < − 11.
Specifically, we have employed five criteria, as listed below:
To further investigate the size–mass relation in various types
of galaxies, we categorize the galaxies into two types: QGs and 1. The r-band magnitude is less than 22 to ensure the
SFGs. Numerous studies have shown that QGs and SFGs quantity of photometric redshift (see the Appendix).
represent two distinct stages of galaxy evolution, as evidenced 2. Considering the reliability of the half-light radius
by the bivariate distribution in color space (e.g., Baldry et al. measurement, the galaxies with Re sRe > 5 are selected.
2004; Schawinski et al. 2014). SFGs are a type of galaxies with 3. A more rigorous criterion than the parent catalog is applied
relatively high SFRs and active star-forming activities. On the to the photometric redshifts, szphoto < 0.05 ´ (1 + zphoto),
other hand, QGs have entered a phase of inactivity with the to exclude the galaxies with poor SED fittings or inaccurate
redshift estimates.
cessation of star formation, exhibiting lower SFRs. As reported
4. The specific star formation rate (sSFR) is utilized to
by Salim (2014) and Bait et al. (2017), there is a remarkable
distinguish between quiescent and star-forming galaxies,
correlation between morphology and specific star formation with the former having −15 < logsSFR < − 11 and the
rate (sSFR = SFR/M*). latter with −11 < logsSFR < − 7.
It is essential to provide a combined approach that can 5. The color criterion is further taken to distinguish the
distinguish between QGs and SFGs more effectively. As early SFGs and QGs. Based on the BC03 templates, the color
as the work by Weinmann et al. (2010), researchers select cut of (g − z) as a function of redshift can be established
“passive galaxies” using the criterion logsSFR < −11. Subse- by tracing the galaxy template with an exponentially
quently, an increasing number of studies adopt the criterion of declining SFH (t = 5 Gyr; Ze; τ < 2 Gyr). We select red
logsSFR = −11 to differentiate between these two galaxy and blue galaxies from the quiescent and star-forming
populations (e.g., Salim et al. 2018; Karunakaran et al. 2023). populations, respectively, and form a final sample
However, considering the uncertainties in determining the consisting of both types in three distinct environments.
stellar mass and SFR, the selection alone by the sSFR criterion Figure 2 visually illustrates the process of sample
may not be sufficient. As shown in Figure 2, many SFGs selection.
selected by −11 < logsSFR < − 7 appear to be red. Leveraging Our final total sample comprises 32,039,360 galaxies with
the advantage of the large sample size provided by the DESI r < 22 at z < 0.5. These galaxies are classified into two types:
Legacy Imaging Surveys, we can employ more stringent QGs and SFGs, and have been categorized into three different
criteria to select QGs and SFGs more reliably. The study environments: field galaxies, member galaxies, and BCGs.
conducted by Ilbert et al. (2013) reveals that, at low redshift Statistics on the number of subsamples under different
(z < 1), the independent selection of QGs using the criteria of conditions are shown in Table 1, in which field galaxies
sSFR (logsSFR < − 11) or color demonstrates a high level of occupy the majority. Additionally, we provide a count
concordance. In practice, we use both sSFR and color as distribution of BCGs, showing that star-forming galaxies
selection criteria to categorize galaxy populations. As a result, constitute a very small fraction (about 13.10%), as demon-
after applying the color criterion described below, about 22% strated in Table 2.

4
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Table 1 left panels. However, quiescent galaxies show a different

Sample Size Statistics for Different Environments pattern of deficiency compared to star-forming galaxies, with a
Type Field Member BCGs Sum more pronounced lack of the former at the faint end. After
reviewing our sample and selection criteria, we recognize that
QGs 6,844,819 461,636 29,827 7,336,282
some fainter galaxies are excluded due to our implementation
(percentage) (21.36%) (1.44%) (0.10% ) (22.90%)
of more strict criteria on both colors and sSFRs. Only the
SFGs 24,335,569 363,014 4495 24,703,078
(percentage) (75.96%) (1.13%) (0.01% ) (77,10%)
quiescent galaxies with red colors are included. In addition, the
lack of faint quiescent galaxies at lower redshift may also be
Sum 31,180,388 824,650 34,322 32,039,360 related to observational constraints and data processing
(percentage) (97.32%) (2.57%) (0.11% ) (100%) methods.
Total 32,039,360 / Due to this, we adopt the dynamic limits of apparent
magnitude to estimate the completeness of stellar mass. The
Note. The categories “Field, Member, BCGs, QGs, and SFGs” correspond to statistics of stellar mass limit and galaxy number in this
field galaxies, member galaxies, the brightest cluster galaxies, quiescent sample are shown in Table 3, in which a brighter limit of
galaxies, and star-forming galaxies, respectively. The percentages indicate the apparent magnitude is settled at lower redshift. Figure 3 also
fraction of different types of subsamples to the total sample. illustrates the adopted limit of apparent magnitude in the left
panels and the resulting mass-complete limit in the right
panels.
Table 2 Figure 4 shows the distribution of Re and M* in logarithmic
The Statistics of BCGs in Different Redshift Bins
space. The total sample is divided into quiescent and star-
Redshift 0–0.1 0.1–0.2 0.2–0.3 0.3–0.4 0.4–0.5 forming galaxies in five redshift bins (!z = 0.1). The black
NQG 601 4626 8510 7372 8718
dotted lines show the 90% mass-complete limit. In each
percentage 2.01% 15.51% 28.53% 24.72% 29.23% redshift bin, we use blue numbers to indicate the quantity of
mass-complete subsamples and red numbers to indicate the
NSFG 100 352 946 1002 2095 total number of samples within that bin.
percentage 2.22% 7.83% 21.05% 22.29% 46.61%

NALL 701 4978 9456 8374 10,813 3. Size–Mass Relation

percentage 2.05% 14.50% 27.55% 24.40% 31.50%
As an important perspective of size–mass relation, mass
Σ(NQG) 29,827 (86.90%) Σ(NSFG) 4495 (13.10%) growth has been discussed by many works. For example, both
Schawinski et al. (2014) and Webb et al. (2015) have shown
Notes.“NQG” represents the number of quiescent galaxies in the redshift bin,
that in the early phase of galaxy evolution, the dominant factor
while “NSFG” represents the number of star-forming galaxies, and “NALL”
represents the total number of galaxies. “Σ(NQG)” and “Σ(NSFG)” represent the
of mass growth is continuous star formation, and mergers
total number of quiescent and star-forming galaxies without bin distinction, dominate mass growth in the later phase. The evolution of
respectively. stellar mass directly affects size growth, which has also been
demonstrated by many observations (e.g., Huertas-Company
et al. 2013b; Lin et al. 2013 ). For instance, van Dokkum et al.
The 90% completeness limit of stellar mass (mass-complete: (2010) demonstrate that during the redshift range of
Mcomp) is estimated using a method similar to Pozzetti et al. 1.4 < z < 2, the size growth is primarily driven by star
(2010) and Gu et al. (2021). The completeness of stellar mass is formation. Massive galaxies, particularly quiescent galaxies
affected by both the redshift and mass-to-light ratio (M/L ratio) or BCGs, exhibit a higher slope α in the size–mass relation,
for each galaxy. Given the well-established fact that quiescent indicating that mergers have replaced star formation as the
galaxies usually have a higher M/L ratio compared to star- dominant factor in their mass growth (e.g., Naab et al. 2009).
forming galaxies, the completeness limits for different galaxy However, the influence of the environment on the size–mass
types can vary at different redshifts. We calculate the value of relation has not been revealed. Cooper et al. (2012) find a
Mcomp for each redshift bin by determining the stellar mass strong relationship between galaxy size and environment for
limit (Mlim ) for the 20% faintest galaxies within a given redshift the quiescent galaxies at 1 < z < 2. Some studies suggest that
bin (!z = 0.1), by applying the following formula: the environment plays an important role in massive early-type
log Mlim = log M* + 0.4 ´ (r - r lim ), where r lim is set as the galaxies at the low-redshift end (e.g., Yoon et al. 2017).
magnitude limit of our subsamples. For SFGs we use a fixed However, Kelkar et al. (2015) compare the mass-size relations
value of 22.0 as the magnitude limit for each redshift bin. For of cluster and field galaxies in the redshift range of
QGs, based on the distribution of magnitudes at z < 0.4, we 0.4 < z < 0.8 and find no significant environmental difference
pick up the r-band magnitude where 95% of galaxies are in it. So, how does environment play a role in size growth?
brighter at each redshift bin as the limiting magnitude (r lim ). As Next, the impact of stellar mass and environment on the size–
a result, the r lim for the five redshift bins are determined to be mass relation will be discussed in Sections 3.1 and 3.2,
18.4, 19.7, 20.5, 21.4, and 22.0, respectively. Subsequently, the respectively.
final Mcomp is defined as the upper envelope of the Mlim
distribution, below which lie 90% of the Mlim values at a given
3.1. The Effect of Stellar Mass
redshift bin.
Figure 3 depicts the distribution of r-band magnitude and To examine the correlation between the sizes and masses
stellar mass as a function of redshift. The imposed limit on the of quiescent and star-forming galaxies across various redshift
r-band absolute magnitude (Mr < − 16) leads to a dearth of bins, three subsamples with distinct environmental densities are
galaxies at the faint end at lower redshifts, as evident from the utilized. We employ a robust linear model with HuberRegressor

5
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 3. Left: upper and lower panels display the distribution of r-band magnitudes as a function of redshift. Different colors represent different distribution densities
of galaxies. Red numbers indicate the total number of galaxies in the subsample. The black dotted lines with black numbers represent the apparent magnitude limit
values (r lim ) in the corresponding redshift bins, while the blue dotted lines display the r-band absolute magnitude equal to -16 without considering the k-correction.
The top panel displays the distribution of QGs with r lim sequentially used as (18.4, 19.7, 20.5, 21.4, 22.0), and the bottom panel illustrates the distribution of SFGs, for
which we adopt a fixed r lim value of 22.0 in all redshift bins. The rationale for using different methods to determine the magnitude limits is detailed in Section 2.3.
Right: upper and lower panels present the distribution of stellar masses as a function of redshift. The black solid lines in the panel denote the corresponding
completeness of stellar mass (Mcomp) in different bins, while the black triangle indicates the median redshift in each bin.

Table 3 galaxies in each environment is denoted by the number of

The Complete Mass at Different Magnitude Limits in Different Redshift Bins different colors, with the total number of galaxies in the bin
highlighted in black, and the slopes are annotated in the lower
Redshift Type r lim N log (M*,comp/Me) right corner with the corresponding colors. The best-fitting
0–0.1 QGs 18.4 83,136 10.12 slopes for the QGs and SFGs in different redshift bins are
0.1–0.2 QGs 19.7 508,080 10.32 presented in Table 4.
0.2–0.3 QGs 20.5 1,194,350 10.41 The size–mass relations for QGs and SFGs within various
0.3–0.4 QGs 21.4 1,652,403 10.29 environments are Re µ M 0.31 ~ 0.52 and Re µ M 0.08 ~ 0.22 ,
0.4–0.5 QGs 22.0 2,340,092 10.28 * *
respectively. Our results confirm the consistency of the size–
0–0.1 SFGs 22.0 809,587 8.10 mass relations with previous findings in the redshift range of
0.1–0.2 SFGs 22.0 3,705,206 8.79 z < 0.5 (e.g., Shen et al. 2003; van der Wel et al. 2014;
0.2–0.3 SFGs 22.0 7,414,001 9.21 Kawinwanichakij et al. 2021). Furthermore, we compare the
0.3–0.4 SFGs 22.0 4,391,249 9.54 slopes among the galaxies in different environments and find
0.4–0.5 SFGs 22.0 3,564,723 9.85 that the slopes of field and member galaxies are similar. This
suggests that these galaxies may have undergone similar
Note. “r lim ” represents r-band magnitude; “N” represents the number of evolutionary paths, or at least, that the denser environment does
galaxies that are greater than the mass-complete limit in the corresponding
not significantly affect the size–mass relations of member
redshift interval (!z = 0.1); “log (M*,comp/Me)” represents the mass-complete
limit adopted for the corresponding redshift bin.
galaxies. However, when comparing the slopes of field and
member galaxies with those of BCGs, a noticeable steeper
slope in size–mass relation can be seen for BCGs at z > 0.1,
(Pedregosa et al. 2011) to fit the correlation, and Figure 5 regardless of star formation status, indicating a potential
displays the results. Huber loss functions possess the advantage environmental dependence. To address this complicated
of being less susceptible to the influence of outliers, while still phenomenon, we need to consider the mass-complete limit of
the sample in order to remove the potential bias caused by mass
taking into account their effects to a certain degree. If the
incompleteness.
absolute error of the sample is less than a certain threshold, the
Considering the completeness of stellar mass, we fit the size–
sample will be classified as an internal sample. We set this mass relations with the galaxies above the mass-complete limits
threshold “epsilon” to 1.35 to achieve 95% statistical efficiency. listed in Table 3. Figure 6 presents the results of our analysis
The top five panels display the distributions of QGs in five with the subsample above mass-complete limits. The best-
different redshift bins, while the bottom five panels show the fitting slopes of size–mass relations for all subsamples are
distributions of SFGs. The distributions of field galaxies, given in Table 4. Compared with Figure 5, field and member
member galaxies, and BCGs are represented by gradients of QGs have steeper size–mass relations, which leads to a slighter
blue, green, and red colors, respectively. The number of difference in slope for the QGs within various environments.

6
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 4. Size–mass distribution map (Re − M*) of the total sample. The distribution of Re and M* is shown in logarithmic space. The top five panels show the
distribution of QGs in five different redshift bins, while the bottom five panels show the distribution of SFGs. The color gradients in the figure indicate the density of
galaxies, and the blue/red number in the lower right corner indicates the number of galaxies with mass-complete/total galaxies in that particular redshift bin. The
black dotted lines represent the 90% mass-complete limits.

Figure 5. Contour plot of the size–mass distribution. The distribution of Re and M* is shown in logarithmic space. The top five panels represent quiescent galaxies in
different redshift bins. The contour lines in gradients of blue, green, and red colors depict the distributions of field galaxies, member galaxies, and BCGs, respectively.
The legend in the top left corner shows the numbers of galaxies in different subsamples, and the slopes are shown in the lower right corner. The blue, green, and red
solid lines in the figure represent the least-squares fitting results of field galaxies, member galaxies, and BCGs, respectively. The bottom five panels show the
distribution of star-forming galaxies.

The difference in slopes between star-forming BCGs and field/ slopes are lower at higher redshifts. Our study reveals
member galaxies at z > 0.1 persists noteworthy even with the different trends of redshift evolution of size–mass relation
mass-complete limit subsample. between QGs and SFGs. In general, the slope difference
The slope variations with redshift are shown in Figure 7. To between the QGs and SFGs at 0.1 < z < 0.5 is significant for
mitigate potential effects stemming from mass incomplete- various environments, and it tends to be larger at higher
ness, we fit the size–mass relation for massive galaxies redshifts.
with log (M*/Me) > 10.41. Different colored squares repre- However, it remains uncertain what exactly causes the
sent the slopes of different galaxy types, while error bars discrepancy in slope between BCGs and field/member
denote the mean absolute errors. For quiescent galaxies, we galaxies, and it is hard to definitively attribute the difference
find the highest slope in the redshift bin of 0.2 < z < 0.3, in size–mass relation to the variations in either mass
particularly pronounced for the BCGs. In the case of SFGs, distribution (namely, subsamples in different environments
steeper slopes are found at lower redshifts, whereas the are composed of galaxies of varying masses) or environmental

7
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 6. Contour plot of the size–mass distribution with a 90% completeness mass limit. The distribution of Re and M* is shown in logarithmic space. For the
purpose of easy comparison, the contour lines in different colors show the distribution of the total sample. The size–mass relations and legends display the information
of the mass-complete subsamples. The black dotted lines in the figure show the mass limit for the 90% completeness sample, and the black number next to the dotted
line indicates the corresponding mass-complete limit. The legend in the top left corner shows the numbers of galaxies above the mass limit in different subsamples.
This is the main difference from Figure 5.

Table 4
The Slopes α of Size–Mass Relation (Re µ M a ) in Different Bins
*
Sample Environment Type z < 0.1 0.1 < z < 0.2 0.2 < z < 0.3 0.3 < z < 0.4 0.4 < z < 0.5
Entire Field quiescent 0.37 ± 0.17 0.46 ± 0.16 0.52 ± 0.15 0.48 ± 0.14 0.46 ± 0.15
star-forming 0.22 ± 0.18 0.18 ± 0.17 0.13 ± 0.17 0.09 ± 0.16 0.08 ± 0.16

Member quiescent 0.31 ± 0.15 0.40 ± 0.14 0.47 ± 0.13 0.41 ± 0.13 0.40 ± 0.14
star-forming 0.21 ± 0.16 0.17 ± 0.16 0.13 ± 0.16 0.09 ± 0.16 0.10 ± 0.15

BCG quiescent 0.49 ± 0.12 0.63 ± 0.13 0.74 ± 0.12 0.68 ± 0.13 0.64 ± 0.16
star-forming 0.24 ± 0.15 0.32 ± 0.14 0.18 ± 0.15 0.15 ± 0.15 0.20 ± 0.14

Mass-complete Field quiescent 0.51 ± 0.16 0.60 ± 0.15 0.64 ± 0.13 0.61 ± 0.13 0.58 ± 0.14
star-forming 0.22 ± 0.17 0.17 ± 0.17 0.13 ± 0.16 0.10 ± 0.16 0.12 ± 0.16

Member quiescent 0.51 ± 0.15 0.57 ± 0.13 0.63 ± 0.12 0.55 ± 0.13 0.53 ± 0.14
star-forming 0.21 ± 0.15 0.16 ± 0.15 0.14 ± 0.15 0.10 ± 0.15 0.15 ± 0.15

BCG quiescent 0.50 ± 0.12 0.63 ± 0.13 0.75 ± 0.12 0.68 ± 0.13 0.64 ± 0.16
star-forming 0.24 ± 0.15 0.32 ± 0.14 0.18 ± 0.15 0.14 ± 0.15 0.21 ± 0.14

Mass-matched Field quiescent 0.51 ± 0.13 0.65 ± 0.13 0.69 ± 0.12 0.66 ± 0.12 0.64 ± 0.15
star-forming 0.27 ± 0.15 0.27 ± 0.14 0.21 ± 0.15 0.18 ± 0.16 0.24 ± 0.15

Member quiescent 0.53 ± 0.13 0.67 ± 0.12 0.72 ± 0.11 0.65 ± 0.12 0.64 ± 0.14
star-forming 0.29 ± 0.13 0.29 ± 0.13 0.23 ± 0.14 0.18 ± 0.15 0.29 ± 0.14

BCG quiescent 0.44 ± 0.12 0.58 ± 0.13 0.67 ± 0.12 0.62 ± 0.14 0.61 ± 0.16
star-forming 0.24 ± 0.15 0.32 ± 0.14 0.18 ± 0.15 0.13 ± 0.15 0.21 ± 0.14

effect. To eliminate the impact of mass distribution on size– 2013a). However, some people hold the opposite view. For
mass relation, we will apply a mass-matching method for example, Yoon et al. (2017) studied the massive early-type
unveiling the net effect of the environment in Section 3.2. galaxies with a redshift range of 0.1 „ z < 0.15, and believed
that the environment has a significant impact on the size–mass
relation when the mass is greater than 1011.2Me. In addition, at
3.2. The Effect of Environment the high redshift end, most studies show that the environment is
Whether the environment affects the size–mass relation is relevant to the size–mass relation (e.g., Lani et al. 2013; Delaye
still under debate. At the low-redshift end (z < 0.2), some et al. 2014). But all of these studies are based on small samples
studies show that the environment has no influence on the size– and different definitions of environment. DESI Legacy Imaging
mass relation (e.g., Maltby et al. 2010; Huertas-Company et al. surveys make it possible for us to construct large subsamples of

8
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 7. The slope (α) as a function of redshift (z) is depicted in the figure. The slope of the size–mass relation is fitted by the sample with stellar mass greater than
10.41. We use three panels from left to right to represent the slopes of the subsamples under different environmental densities: BCG, Mem, and Field. The slopes are
represented by colored squares, with red indicating quiescent galaxies and blue representing star-forming galaxies; the error bars denote the mean absolute error.
Dotted lines connect the slopes across different redshift bins.

Table 5
Subsample Size and the Probabilities of K-S Tests for the M*-distributions and z-distributions

Redshift Galaxy Number M*-distribution z-distribution

đRange BCGs Member Field PM−B PF−B PM−B PF−B
0–0.1 656 545 655 0.999 0.981 0.810 0.775
0.1–0.2 3874 3564 3878 0.999 0.982 0.832 0.937
0.2–0.3 6207 6681 6207 0.786 0.994 0.812 0.944
0.3–0.4 6561 6470 6555 0.998 0.996 0.998 0.977
0.4–0.5 9014 9052 9017 0.915 0.992 0.999 0.966

Note. “PM−B” represents the P-values of the K-S test between member galaxies and BCGs; “PF−B” represents the P-values between field galaxies and BCGs.

QGs and SFGs within different environments (namely, field distributions. Table 5 lists the sample size and the K-S
galaxies, member galaxies of clusters, and BCGs). probabilities (P). All of the P-values are much larger than the
In order to eliminate the possible impact of mass and redshift threshold of 0.05, which ensures the similarities in M*- and z-
distributions on size–mass relation, we build the mass- and distributions for the mass-matched samples.
redshift-matched subsamples of BCGs, member galaxies, and Figure 8 shows the size–mass relations for the QGs and
field galaxies at z < 0.5. Since the number of BCGs is minimal, SFGs in mass-matched subsamples of BCGs, member galaxies,
we try to match the field and member galaxies with similar and field galaxies. The best-fitting slopes of size–mass relations
redshift and mass distributions on the basis of the BCG for mass-matched subsamples are listed in Table 4. Compared
subsample. Following the mass-matching method described in with Figure 6, both field and member galaxies have steeper
Gu et al. (2019), for each BCG, one matching galaxy is size–mass relations, while quiescent BCGs tend to have flatter
selected from both the field galaxies and member galaxies, size–mass relations. The QGs in fields and clusters show
respectively. The selected galaxies exhibit the minimum value obvious steepening trends in size–mass relations for various
redshift bins than the SFGs. In Figure 8, we observed a clear
of d = (zi - z 0 )2 + (log M*,i - log M*,0 )2 , where (z0, M*,0) difference in the intercept of the size–mass relation between the
represents the position of the given BCG in the redshift-mass quiescent BCGs and field/member galaxies at 0.4 < z < 0.5.
space, and (zi, M*,i) represents the location of the galaxies to be However, when the uncertainties of slopes are taken into
matched. To ensure a robust redshift and mass-matching, the account, the slope differences among the mass-matched
upper limit of d is set to be 0.04. Each BCG is matched with a subsamples of field galaxies, member galaxies of clusters,
similar galaxy in both mass and redshift space. The BCGs that and BCGs are negligible for both QGs and SFGs. This suggests
fail to find counterpart galaxies in the above matching process that the significant discrepancy in slope between BCGs and
are discarded. As a result, we obtain the mass-matched field/member galaxies unveiled in Figure 6 is mainly because
subsamples of BCGs, field galaxies, and member galaxies BCGs are dominated by massive galaxies covering the high
with similar redshift- and mass distributions. Finally, three end of mass distribution. After eliminating the effect of mass
mass-matched subsamples, including 26,312 BCGs, 26,312 distribution, the environment seems to have little effect on
massive member galaxies, and 26,312 field galaxies are size–mass relations of QGs and SFGs. In other words, mass
constructed. The Kolmogorov–Smirnov (K-S) tests are per- distribution has been affected by the environment, but the
formed to check the similarities in mass and redshift ultimate galaxy size is mainly determined by its stellar mass.

9
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 8. Contour plot of the size–mass distribution with the mass-matched subsamples. The distribution of Re and M* is shown in logarithmic space.

Figure 9. The slope (α) as a function of redshift (z) is depicted in the figure. The slope of the size–mass relation is fitted by the mass-matched subsamples. We use
three panels from left to right to represent the slopes of the subsamples under different environmental densities: BCG, Mem, and Field. The slopes are represented by
colored squares, with red indicating quiescent galaxies and blue representing star-forming galaxies; the error bars denote the mean absolute error. Dotted lines connect
the slopes across different redshift bins.

The slope variations with redshift for the mass-matched matched subsamples of quiescent BCGs, member galaxies, and
subsamples are shown in Figure 9. Compared with Figure 7, it field galaxies, we achieve different results. It is found that the
is confirmed that the mass-matched subsamples of QGs have massive QGs with M* > 1011.2Me indeed exhibit a higher
the highest slope at 0.2 < z < 0.3, but the trend of SFGs having slope in their mass-size relation with respect to the entire mass-
a higher slope at z < 0.2 seems to be alleviated, showing a more matched QG subsample. However, there is no significant slope
consistent slope across different redshift bins. It should be difference among the massive BCGs, members, and field
noted that the trend of slope change with redshift for the QGs galaxies with M* > 1011.2Me. Our finding supports that the
and SFGs within three environments is more consistent for the steeper slopes in size–mass relation observed in massive
mass-matched subsamples, which further confirms that the galaxies across different environments can be attributed to the
size–mass relation is independent of the environment. difference in the distribution of stellar mass. The environment
Yoon et al. (2017) found that the early-type galaxies with does not directly influence the size–mass relation but rather
M* > 1011.2Me in denser environments tend to have larger changes the distribution of stellar mass. Stellar mass seems to
effective radii on the basis of the random sampling method, be the predominant factor for driving the size growth of
which was attributed to environmental factors. Since the galaxies.
majority of early-type galaxies in high-density regions are To sum up, in the star-forming epoch of galaxy evolution
quiescent, it is interesting for us to check this tendency via our (i.e., SFGs), continuous star formation leads to stellar mass
subsamples of quiescent galaxies. Based on our large mass- growth. With the increase in stellar mass and gas consumption,

10
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

star formation activity will halt when its gas supply and size–mass relations is found between the low- and high-mass
reservoir are destroyed. For the quiescent epoch of galaxy ends of late-type galaxies. It shows a steeper relation R µ M 0.4
*
evolution, minor mergers become the dominant mode of mass at M* > M0 = 1010.6Me, and a shallower relation R µ M 0.15
growth in QGs (Naab et al. 2007; Bird et al. 2013; van Dokkum *
below M0. They also find that the relation is significantly
et al. 2015). This leads to a higher prevalence of massive steeper for early-type galaxies (R µ M 0.55). van der Wel et al.
*
galaxies for QGs, which is why QGs exhibit a steeper size– (2014) confirm that the size–mass relation obeys a single power
mass relation than SFGs (see Figure 6). In general, massive law for the early- and late-type galaxies over a large redshift
galaxies are more likely to be attracted toward the centers of range of 0 < z < 3. A growing body of research suggests that
high-density dark matter halos due to gravitational interactions, the relationship between size and mass varies across different
which provides more opportunities for galaxy mergers. This mass ranges, deviating from a single power law. Mowla et al.
makes the high-density environment occupied by more massive (2019) fit the size–mass relation of all types of galaxies (i.e.,
galaxies, which is the reason why most of the BCGs are QGs and SFGs) with a broken power law up to z ∼ 3.
massive QGs. Our work suggests that, for QGs and SFGs, Analogously, Kawinwanichakij et al. (2021) use a broken
stellar mass is a more fundamental factor for size growth, and power law to fit the size–mass relation for QGs and SFGs at
environments only affect the distribution of stellar mass. z < 1, respectively. They demonstrate that the size–mass
relations of both populations with a clear change of slopes at
4. Discussion a pivot stellar mass Mp (Mp ∼ 1010.2–10.6Me for QGs,
Mp ∼ 1010.7Me for SFGs). It may imply that stellar mass is a
4.1. Uncertainties of Stellar Mass and SFR dominant factor in size growth.
To enhance the robustness of QG and SFG classification, we In this paper, we mainly focus on comparing the difference
simultaneously employ two criteria: color (g − z) and sSFR. in size–mass relation for the QGs and SFGs at z < 0.5 within
The stellar mass and SFR used in this work are derived with the various environments, and try to observe the fundamental
LePhare software by Zou et al. (2019). As described in Carnall factor driving the change of the slope. The majority of BCGs
et al. (2019), inferring the SFH from the galaxy SED is quite commonly have higher stellar masses, resulting in the BCG
challenging due to the strong evolution of the mass-to-light subsample covering a narrower mass range. It is hard to
ratios of stellar populations with age. Generically, the SED- perform a broken power law to fit the size–mass relations for
fitting method can recover the stellar mass at an uncertainty of mass-matched subsamples. Therefore, we adopt a single power
about 0.2–0.3 dex, and even better than 0.2 dex if near-infrared law in our fittings.
bands are involved (Muzzin et al. 2009; Zou et al. 2019). In our Using single power-law fitting, the recent work by Afanasiev
study, the robust photometric redshifts and stellar masses for et al. (2023) studied the size–mass relation for the galaxies in
galaxies are derived from both optical and near-infrared 15 clusters at 1.4 < z < 2.8 from the CARLA survey. It is
photometry. found that the size–mass relation does not evolve much from
There is a high level of consistency in stellar mass found in z ∼ 2 to the present, and their sizes evolve in a similar way for
Zou et al. (2019) when comparing their results with cluster and field galaxies. Moreover, both BCGs and satellite
MAGPHYS fitting. In addition, Kaushal et al. (2024) also galaxies are found to lie in the same size–mass relation.
indicate that, despite the various SFH patterns observed in Similarly, Strazzullo et al. (2023) compare the sizes of field
individual galaxies, the median sSFR is consistent across all galaxies at 1.4 < z < 1.7 with the data from massive clusters in
mass bins. Nevertheless, it is important to note that sSFR is an the South Pole Telescope Sunyaev–Zel’dovich effect survey,
exceptionally sensitive indicator in sampling. According to the and no significant differences in size are revealed between
findings of Leja et al. (2017), the adoption of nonparametric these two different environments. It is interesting that the
SFHs in Prospector-α, as opposed to the use of parametric massive galaxies with M* ∼ 1011Me observed by the James
SFHs in MAGPHYS, results in an average increase of 0.25 dex Webb Space Telescope (JWST) in the near-infrared band
in stellar mass and a 0.1 dex decrease in SFR. Therefore, in appear to be more compact than their rest-frame optical light
order to evaluate the possible effect due to sSFR inconsistency profiles at cosmic noon on average (Suess et al. 2022). It is
produced by various SFH treatments, some blue galaxies worth noting that their studies primarily focus on the galaxies at
within −11 < logsSFR < − 10.65 might have been misclassi- high redshifts, whereas our current work makes use of the
fied as SFGs. Number of these blue galaxies is 400,816, galaxies at z < 0.5. The environmental effect on the star
occupying only 1.62% of our star-forming subsample. Conse- formation quenching tends to be more significant at z < 0.5
quently, we have strong grounds to dismiss the influence of (Peng et al. 2010). The large sample of the z < 0.5 galaxies
different fitting models on the outcomes of our research. from the DESI Legacy Imaging Surveys has great advantages
in studying the environmental effects on size growth. Our
4.2. Size–Mass Relation Fitting and Environment Effect results suggest that stellar mass distribution dominates the size–
mass relation, and environmental factors (such as tidal
In general, changes in the mass of a galaxy will be stripping, ram pressure stripping, and harassment) only change
accompanied by alterations in its size. Numerous factors their star formation histories. The observed sizes tend to be
impact the transformations in the mass and size of a galaxy, more closely relative to the accumulated stellar masses,
including accretion, activities of star formation, mergers, and whether they are in fields or clusters, and whether they are
environment. We employ the size–mass relation (Re µ M a ), central or satellite galaxies.
*
which is a fundamental relation, to explore the size growth of
galaxies in relation to stellar mass and environment, and further
4.3. Redshift Evolution of Size–Mass Relation
examine the effects of the environment on size growth. Shen
et al. (2003) apply a power law to fit the size–mass relations for In Sections 3.1 and 3.2, we demonstrate the evolution of the
early- and late-type galaxies, respectively. A difference in the slope for QGs and SFGs in different environments with both

11
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 10. Left: the distribution of halo mass and redshift for all the galaxies selected in our work. The colored points denote the galaxies in our growth-tracking
subsample; the solid black lines in this panel represent the MAHs for the upper (12.5) and lower (11.6) halo mass limits at z = 0.5. Middle: the distribution of stellar
mass and redshift for all the galaxies selected in our work. The color bar represents the number of galaxies; the black line represents the mass-complete limit, which is
fitted by interpolation. Right: the slope distribution of the size–mass relation for the subsample of growth-tracking galaxies at 0.1 < z < 0.5. The slopes are represented
by colored squares, with red indicating QGs and blue representing SFGs; the error bars denote the mean absolute error.

mass-complete and mass-matched subsamples, respectively. As depicted in Figure 10, the left panel illustrates the
When considering uncertainties and mass distribution, a similar distribution of halo mass across different redshifts. Colored
slope is found for the SFGs in various environments, and a points represent our growth-tracking subsample, while the solid
steeper slope is found for the QGs, particularly at z > 0.1. black lines in this panel represent the MAHs for the upper and
Generally, as galaxies transition from SFGs to QGs, the mass lower-mass limits. In the middle panel, one can observe the
growth pattern also undergoes changes, which, in turn, results distribution of stellar mass for our subsample, with the color
in alternations in the slope. However, in the context of galaxy bar indicating the number of galaxies. The right panel displays
evolution, the galaxies with lower masses at higher redshifts the distribution of slopes for the growth-tracking subsample.
should be the progenitors of those at present (z = 0). It is The BCGs have been excluded from this subsample due to their
interesting to discuss the redshift evolution of the size–mass complex evolutionary processes. Due to the limited sample size
relation on the basis of the galaxy samples at various redshifts and larger slope uncertainty at z < 0.1, we primarily focus on
with some possible evolutionary link. the slope relation for 0.1 < z < 0.5. As shown in the right
In order to minimize the influence of progenitor bias, we panel, the SFGs exhibit an increase in their slope at lower
adopt the model of mass accretion histories (MAH), given by redshifts, suggesting a change in their size growth pattern. In
McBride et al. (2009) on the basis of dark matter halos from the contrast, QGs demonstrate a relatively stable slope within the
Millennium simulation, which can be formulated with range of 0.1 < z < 0.4, indicating a stable size growth pattern.
M (z ) = M0 (1 + z )b e-gz , where M0 represents the halo mass We would like to emphasize that the evolutionary link
at z = 0, and β and γ are fitting parameters. By using the MAH between the galaxies at different redshifts is strongly dependent
model mentioned above, McDonald et al. (2017) took the upon the halo growth model based on the simulations only.
cluster halo mass at high redshift to track the expected growth Moreover, some QGs at z = 0 might have evolved from the
trajectory of clusters at lower redshifts. Similarly, we take the SFGs at higher redshifts, and the progenitors of some SFGs
halo mass region, log Mhalo M ~ 11.6–12.5, as the galaxy- might be quiescent. For simplicity, we neglect the cases where
sized halos at z = 0.5 to trace the corresponding ranges of the type of progenitors changes from z = 0.5 to 0.1. Therefore,
galaxy halo mass at lower redshifts. Based on the known stellar this strategy to create galaxy samples with evolutionary links
mass, the corresponding halo mass of a galaxy can be estimated should be further validated through a detailed comparison
by the stellar-to-halo mass relation (SHMR) that was recently between simulations and observations.
established by Shuntov et al. (2022) with the COSMOS2020
catalog. For the galaxies at z < 0.5, the SHMR can be modeled 5. Conclusion
as
In this paper we explore the size–mass relations of quiescent
M (M* M*,0)d and star-forming galaxies in three distinct environments (i.e.,
log Mhalo = log M1 + b log * + - 0.5, BCGs, cluster galaxies, and field galaxies), using a large
M*,0 1 + (M* M*,0)-g
sample of 32,039,360 galaxies with r < 22 and z < 0.5 selected
where all of log M1, log M*,0 , β, δ and γ are the best-fitting from the DESI Legacy Imaging Surveys. Based on the
color (g - z) and sSFR criteria, these galaxies are divided into
parameters given in Shuntov et al. (2022).
QGs and SFGs. The BCGs are found to be dominated by QGs.
The upper and lower limits for halo mass (log Mhalo M) at The difference in size–mass relation for the galaxies within
z = 0.5 are set to be 11.6 and 12.5 (for a typical galaxy-sized
three environments has been investigated, and the main
halo), respectively. The stellar mass (log M* M) corresp-
conclusions are the following:
onding to the lower limit of halo mass is 9.8, which is the lower
limit of stellar mass for the mass-complete sample at 1. Regardless of galaxy environment, QGs exhibit a steeper
0.4 < z < 0.5. The derived upper limit of stellar mass at size–mass relation compared with SFGs, based on both
z = 0.5 is log M* M = 10.8. mass-complete and mass-matched subsamples. The slope

12
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

difference between QGs and SFGs is significant at of the University of California under contract to the U.S.
0.1 < z < 0.5 for various environments. Department of Energy.
2. The galaxies with greater stellar masses demonstrate a
steeper size–mass relation. For mass-complete subsam- Appendix
ples, BCGs show the highest slope in the size–mass The Accuracy of Photometric Redshift
relation when compared to member and field galaxies. No
We adopt three parameters to evaluate the quality of
significant difference in slopes is observed between
photometric redshift at 0 < z < 0.5 as follows:
cluster members and field galaxies.
3. The slope variations of the size–mass relation with 1. Bias (!NMAD): the median of systematic deviation
redshift exhibit distinct characteristics for QGs and SFGs. between the spectroscopic (zspec) and photometric red-
The slope of QGs reaches its maximum at 0.2 < z < 0.3, shifts (zphoto),
while the slope of SFGs is higher at z < 0.2. Compared to
the mass-complete subsamples, the mass-matched sub- z photo - zspec ⎞
NMAD = median ⎛⎜ ⎟.
⎝ 1 + zspec ⎠
samples exhibit a more consistent trend of slope variation
with redshift for the QGs and SFGs within three
environments. 2. Dispersion (sNMAD ): the dispersion of !NMAD, and the
4. Based on the mass-matched subsamples of QGs and median !z(=zphoto − zspec) is used to calculate the normal
SFGs, no significant slope difference for the size–mass distribution of median absolute deviation,
relation is found among BCGs, cluster members, and
field galaxies. It suggests that, though mass growth mode ∣  z - median (  z)∣ ⎞
sNMAD = 1.48 ´ median ⎛⎜ ⎟.
and mass distribution for QGs and SFGs can be affected ⎝ 1 + zspec ⎠
by the galaxy environment, stellar mass is the most
fundamental factor driving the size evolution at z < 0.5. 3. Outlier rate (η0.15): the fraction of galaxies exhibiting
significant deviations between their photometric redshifts
and spectroscopic redshifts, satisfying
Acknowledgments ∣zphoto - zspec∣
h= > 0.15.
This research is supported by the National Key R&D 1 + zspec
Program of China No. 2023YFA1607800 and the National
Natural Science Foundation of China (NSFC, Nos. 12273013,
12373010, 12120101003, and 11890691). Y.G. acknowledges The accuracy of photometric redshifts in the range of
the support from the China Postdoctoral Science Foundation z < 0.5, is sNMAD» 0.013, as shown in Figure 11. Compared
(2020M681281) and the Shanghai Postdoctoral Excellence with the accuracy of about 0.017 for the total sample in Zou
Program (2020218). H.Z. acknowledges the support from the et al. (2019), the photometric redshift accuracy is better at
National Key R&D Program of China (grant No. z < 0.5. To verify the accuracy of photometric redshift as a
2022YFA1602902) and China Manned Space Project (Nos. function of magnitude, we divided the magnitude into six bins,
CMS-CSST-2021-A02 and CMS-CSST-2021-A04). as shown in Figure 12. Three parameters (bias, dispersion, and
The DESI Legacy Imaging Surveys consist of three
individual and complementary projects: the Dark Energy
Camera Legacy Survey (DECaLS), the Beijing-Arizona Sky
Survey (BASS), and the Mayall z-band Legacy Survey
(MzLS). DECaLS, BASS, and MzLS together include data
obtained, respectively, at the Blanco telescope, Cerro Tololo
Inter-American Observatory, NSFʼs NOIRLab; the Bok tele-
scope, Steward Observatory, University of Arizona; and the
Mayall telescope, Kitt Peak National Observatory, NOIRLab.
NOIRLab is operated by the Association of Universities for
Research in Astronomy (AURA) under a cooperative agree-
ment with the National Science Foundation. Pipeline proces-
sing and analyses of the data were supported by NOIRLab and
the Lawrence Berkeley National Laboratory (LBNL). Legacy
Surveys also uses data products from the Near-Earth Object
Wide-field Infrared Survey Explorer (NEOWISE), a project of
the Jet Propulsion Laboratory/California Institute of Technol-
ogy, funded by the National Aeronautics and Space Admin-
istration. Legacy Surveys was supported by: the Director,
Office of Science, Office of High Energy Physics of the U.S.
Department of Energy; the National Energy Research Scientific
Computing Center, a DOE Office of Science User Facility; the Figure 11. Photometric redshift (zphoto) as a function of spectroscopic redshift
U.S. National Science Foundation, Division of Astronomical (zspec). The color gradients in the figure represent the density of the galaxies.
The black solid line in the figure indicates that zspec = zphoto, and the red solid
Sciences; the National Astronomical Observatories of China, lines show the outlier rate of photometric redshifts with η > 0.15. The values of
the Chinese Academy of Sciences and the Chinese National the three parameters (bias, dispersion, and outlier rate) are marked in red in the
Natural Science Foundation. LBNL is managed by the Regents lower right corner.

13
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

Figure 12. Photometric redshifts vs. spectroscopic redshifts in six different r-band magnitude bins. The color gradients in the figure represent the density of the
galaxies. The black solid line in the figure indicates that zspec = zphoto, and the red solid lines show the outlier rate of photometric redshifts with η > 0.15. The values of
the three parameters (bias, dispersion, and outlier rate) are marked in red at the lower right corner.

14
The Astrophysical Journal, 961:253 (15pp), 2024 February 1 Chen et al.

outlier rate) are labeled at the lower right corner of each panel. Kluge, M., Neureiter, B., Riffeser, A., et al. 2020, ApJS, 247, 43
For the galaxies with r > 22, both dispersion and outlier rates Kravtsov, A. V. 2013, ApJL, 764, L31
exceed 0.03. One possible reason is the lack of spectroscopic Lani, C., Almaini, O., Hartley, W. G., et al. 2013, MNRAS, 435, 207
Lavoie, S., Willis, J. P., Démoclès, J., et al. 2016, MNRAS, 462, 4141
data as a training sample at the faint end (see also Figure 2 of Leja, J., Johnson, B. D., Conroy, C., van Dokkum, P. G., & Byler, N. 2017,
Zou et al. 2022). The large uncertainty of redshift would ApJ, 837, 170
enhance the contamination in the membership of clusters. To Lidman, C., Iacobuta, G., Bauer, A. E., et al. 2013, MNRAS, 433, 825
ensure the purity of cluster members, we set the r-band Lidman, C., Suherli, J., Muzzin, A., et al. 2012, MNRAS, 427, 550
Lin, Y.-T., Brodwin, M., Gonzalez, A. H., et al. 2013, ApJ, 771, 61
magnitude limit to 22 in further sample selection. Maltby, D. T., Aragón-Salamanca, A., Gray, M. E., et al. 2010, MNRAS,
402, 282
ORCID iDs McBride, J., Fakhouri, O., & Ma, C.-P. 2009, MNRAS, 398, 1858
McDonald, M., Allen, S. W., Bayliss, M., et al. 2017, ApJ, 843, 28
Zhaobin Chen https://orcid.org/0009-0006-9628-7318 McLure, R. J., Pearce, H. J., Dunlop, J. S., et al. 2013, MNRAS, 428, 1088
Yizhou Gu https://orcid.org/0000-0003-3196-7938 Mo, H. J., Mao, S., & White, S. D. M. 1998, MNRAS, 295, 319
Hu Zou https://orcid.org/0000-0002-6684-3997 Mowla, L., van der Wel, A., van Dokkum, P., & Miller, T. B. 2019, ApJL,
872, L13
Qirong Yuan https://orcid.org/0000-0002-9244-3938 Muzzin, A., Marchesini, D., van Dokkum, P. G., et al. 2009, ApJ, 701, 1839
Naab, T., Johansson, P. H., & Ostriker, J. P. 2009, ApJL, 699, L178
References Naab, T., Johansson, P. H., Ostriker, J. P., & Efstathiou, G. 2007, ApJ,
658, 710
Afanasiev, A. V., Mei, S., Fu, H., et al. 2023, A&A, 670, A95 Nersesian, A., Xilouris, E. M., Bianchi, S., et al. 2019, A&A, 624, A80
Arnouts, S., Moscardini, L., Vanzella, E., et al. 2002, MNRAS, 329, 355 Newman, A. B., Ellis, R. S., Bundy, K., & Treu, T. 2012, ApJ, 746, 162
Ascaso, B., Aguerri, J. A. L., Varela, J., et al. 2011, ApJ, 726, 69 Oemler, A. J. 1974, ApJ, 194, 1
Bait, O., Barway, S., & Wadadekar, Y. 2017, MNRAS, 471, 2687 Oliva-Altamirano, P., Brough, S., Lidman, C., et al. 2014, MNRAS, 440, 762
Baldry, I. K., Glazebrook, K., Brinkmann, J., et al. 2004, ApJ, 600, 681 Paspaliaris, E. D., Xilouris, E. M., Nersesian, A., et al. 2023, A&A, 669, A11
Beck, R., Dobos, L., Budavári, T., Szalay, A. S., & Csabai, I. 2016, MNRAS, Pedregosa, F., Varoquaux, G., Gramfort, A., et al. 2011, JMLR, 12, 2825
460, 1371 Peng, Y.-j., Lilly, S. J., Kovač, K., et al. 2010, ApJ, 721, 193
Bekki, K., & Couch, W. J. 2003, ApJL, 596, L13 Pichon, C., Pogosyan, D., Kimm, T., et al. 2011, MNRAS, 418, 2493
Bird, J. C., Kazantzidis, S., Weinberg, D. H., et al. 2013, ApJ, 773, 43 Poggianti, B. M., Calvi, R., Bindoni, D., et al. 2013, ApJ, 762, 77
Bluck, A. F. L., Conselice, C. J., Buitrago, F., et al. 2012, ApJ, 747, 34 Pozzetti, L., Bolzonella, M., Zucca, E., et al. 2010, A&A, 523, A13
Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000 Salim, S. 2014, SerAJ, 189, 1
Bundy, K., Ellis, R. S., Conselice, C. J., et al. 2006, ApJ, 651, 120 Salim, S., Boquien, M., & Lee, J. C. 2018, ApJ, 859, 11
Calvi, R., Vulcani, B., Poggianti, B. M., et al. 2018, MNRAS, 481, 3456 Saracco, P., Gargiulo, A., Ciocca, F., & Marchesini, D. 2017, A&A, 597, A122
Carnall, A. C., Leja, J., Johnson, B. D., et al. 2019, ApJ, 873, 44 Schawinski, K., Urry, C. M., Simmons, B. D., et al. 2014, MNRAS, 440, 889
Cebrián, M., & Trujillo, I. 2014, MNRAS, 444, 682 Shen, S., Mo, H. J., White, S. D. M., et al. 2003, MNRAS, 343, 978
Chabrier, G. 2003, PASP, 115, 763 Shibuya, T., Ouchi, M., & Harikane, Y. 2015, ApJS, 219, 15
Cooper, M. C., Griffith, R. L., Newman, J. A., et al. 2012, MNRAS, 419, 3018 Shuntov, M., McCracken, H. J., Gavazzi, R., et al. 2022, A&A, 664, A61
Delaye, L., Huertas-Company, M., Mei, S., et al. 2014, MNRAS, 441, 203 Smith, G. P., Kneib, J.-P., Smail, I., et al. 2005, MNRAS, 359, 417
De Lucia, G., & Blaizot, J. 2007, MNRAS, 375, 2 Strazzullo, V., Pannella, M., Mohr, J. J., et al. 2023, A&A, 669, A131
Dey, A., Schlegel, D. J., Lang, D., et al. 2019, AJ, 157, 168 Suess, K. A., Bezanson, R., Nelson, E. J., et al. 2022, ApJL, 937, L33
Dressler, A. 1980, ApJ, 236, 351 van der Wel, A., Franx, M., van Dokkum, P. G., et al. 2014, ApJ, 788, 28
Dubinski, J. 1998, ApJ, 502, 141 van Dokkum, P. G., Nelson, E. J., Franx, M., et al. 2015, ApJ, 813, 23
Gao, J., Zou, H., Zhou, X., & Kong, X. 2018, ApJ, 862, 12 van Dokkum, P. G., Whitaker, K. E., Brammer, G., et al. 2010, ApJ, 709, 1018
Gao, J., Zou, H., Zhou, X., & Kong, X. 2020, PASP, 132, 024101 Webb, K. A., Villaume, A., Laine, S., et al. 2022, MNRAS, 516, 3318
Greene, J. E., Murphy, J. D., Comerford, J. M., Gebhardt, K., & Adams, J. J. Webb, T. M. A., Muzzin, A., Noble, A., et al. 2015, ApJ, 814, 96
2012, ApJ, 750, 32 Weinmann, S. M., Kauffmann, G., van den Bosch, F. C., et al. 2009, MNRAS,
Gu, Y., Fang, G., Yuan, Q., et al. 2019, ApJ, 884, 172 394, 1213
Gu, Y., Fang, G., Yuan, Q., Cai, Z., & Wang, T. 2018, ApJ, 855, 10 Weinmann, S. M., Kauffmann, G., von der Linden, A., & De Lucia, G. 2010,
Gu, Y., Fang, G., Yuan, Q., Lu, S., & Liu, S. 2021, ApJ, 921, 60 MNRAS, 406, 2249
Gunn, J. E., & Gott, J. R. I. 1972, ApJ, 176, 1 Whiley, I. M., Aragón-Salamanca, A., De Lucia, G., et al. 2008, MNRAS,
Huang, K.-H., Fall, S. M., Ferguson, H. C., et al. 2017, ApJ, 838, 6 387, 1253
Huertas-Company, M., Mei, S., Shankar, F., et al. 2013b, MNRAS, 428, 1715 White, S. D. M., & Frenk, C. S. 1991, ApJ, 379, 52
Huertas-Company, M., Shankar, F., Mei, S., et al. 2013a, ApJ, 779, 29 White, S. D. M., & Rees, M. J. 1978, MNRAS, 183, 341
Ilbert, O., Arnouts, S., McCracken, H. J., et al. 2006, A&A, 457, 841 Whitney, A., Conselice, C. J., Bhatawdekar, R., & Duncan, K. 2019, ApJ,
Ilbert, O., McCracken, H. J., Le Fèvre, O., et al. 2013, A&A, 556, A55 887, 113
Ilbert, O., Salvato, M., Le Floc’h, E., et al. 2010, ApJ, 709, 644 Williams, R. J., Quadri, R. F., Franx, M., et al. 2010, ApJ, 713, 738
Jones, C., & Forman, W. 1984, ApJ, 276, 38 Williams, R. J., Quadri, R. F., Franx, M., van Dokkum, P., & Labbé, I. 2009,
Karunakaran, A., Sand, D. J., Jones, M. G., et al. 2023, MNRAS, 524, 5314 ApJ, 691, 1879
Kaushal, Y., Nersesian, A., Bezanson, R., et al. 2024, ApJ, 961, 118 Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, AJ, 140, 1868
Kawinwanichakij, L., Silverman, J. D., Ding, X., et al. 2021, ApJ, 921, 38 Yoon, Y., Im, M., & Kim, J.-W. 2017, ApJ, 834, 73
Kelkar, K., Aragón-Salamanca, A., Gray, M. E., et al. 2015, MNRAS, Zou, H., Gao, J., Xu, X., et al. 2021, ApJS, 253, 56
450, 1246 Zou, H., Gao, J., Zhou, X., & Kong, X. 2019, ApJS, 242, 8
Khullar, G., Bayliss, M. B., Gladders, M. D., et al. 2022, ApJ, 934, 177 Zou, H., Sui, J., Xue, S., et al. 2022, RAA, 22, 065001
Kim, K. J., Bayliss, M. B., Noble, A. G., et al. 2023, ApJ, 955, 32 Zou, H., Zhou, X., Fan, X., et al. 2017, PASP, 129, 064101

15