Environmental Invariance of the Galaxy Size-Mass Relation

TL;DR

This paper investigates whether the galaxy size–mass relation depends on environment by analyzing a large, homogeneous DESI Legacy Imaging Survey sample that extends to $M_\star\sim10^{7}\,M_\odot$ and uses distance to the nearest cluster as the environmental metric. Sizes are derived from LS Tractor fits, with volume corrections applied via $1/V_{\max}$, and masses drawn from GSWLC and WH2024; the study separates galaxies by color and Sérsic index to track subpopulations. The key finding is that the environmental variation of the size–mass relation on Mpc scales arises from changing subpopulation mixes rather than direct size transformations, implying assembly histories primarily govern galaxy size. The results support environment-insensitive size–mass relations for subpopulations and highlight the usefulness of population-specific calibrations for baryon-cycle physics in simulations. The work also discusses the potential of alternative size metrics like $R_1$ and the implications for comparing observations to theoretical models.

Abstract

The galaxy size-luminosity and size-stellar mass relations are important constraints on the galactic baryon cycle of gas accretion, star formation, and feedback. There are conflicting claims in the literature regarding how environment influences size: both direct transformative effects and `assembly bias' may contribute to observed variations with environment. We construct a large homogeneous sample of size measurements to M*~10^7 Msun. Our sample fills a gap in field galaxy size measurements around 10^7-10^8 Msun; the literature at these masses is biased towards satellites of L* galaxies and members of galaxy clusters. We use sizes from the DESI-LS, together with a published catalog that contains stellar masses and cluster positions derived from DESI-LS photometry. Our sample extends to z<0.3 and comprises 540,228 galaxies with spectroscopic redshifts and 9,513,732 galaxies with photometric redshifts. We explore the environmental dependence of size for a mass-limited subset of our sample at z<0.05, based on distance to the nearest cluster center. We obtain size-luminosity and size-mass relations in good agreement with previous studies. By separating galaxies according to color and morphology, we show that the environmental variation of the overall size-mass relation on Mpc scales can be understood as the consequence of a changing mixture of subpopulations, rather than direct size transformation. For example, at fixed mass, quiescent (red) late-type galaxies within 2Mpc of a cluster have the same size as quiescent late-type galaxies 30Mpc from the nearest cluster. Our results support individual galaxy assembly histories as the primary determinant of galaxy size. The existence of significantly different, environment-insensitive size mass relations for subpopulations separated by color and Sersic index provides a clear target for calibration of the baryon cycle in cosmological simulations.

Figures (16)

• Figure 1: A summary of the two main galaxy samples we use in this work. Normalized histograms of (a) $R_\mathrm{eff}$, (b) $R_\mathrm{maj}$, (c) $\mu_r$, (d) $r$-band magnitude, (e) $M_r$, and (f) $M_\star$ for data with spectroscopic redshift and data with photometric redshift. Filled blue and filled orange show the unweighted distribution of data with spectroscopic redshift and data with photometric redshift. The dashed blue and orange histograms represent the $V/V_\mathrm{max}$ weighted distributions of the spectroscopic and photometric data samples, respectively. The density scale on the vertical axis is arbitrary.
• Figure 2: Upper row: color magnitude diagrams of the sz sample (panel a) and the pz sample (panel b). Lower row: contours show the distribution of galaxies with $n>2.5$ (red) and $n<2.5$ (blue) for the sz sample (panel c) and the pz sample (panel d). All the contours are drawn at [0.1, 0.5, 3, 10, 30, 80, 99]% of the peak density. The dashed line indicates a fiducial separation between red sequence and blue cloud galaxies, $g-r = -0.023\times M_r+0.15$, determined by the eye. Although most galaxies with high Sérsic index are located in the red sequence, the two definitions of the type clearly select different samples of galaxies.
• Figure 3: Distribution of effective semi-major axis for sz galaxies. The blue-dashed lines show the distribution without applying $V/V_\mathrm{max}$ weighting. The solid orange lines show the weighted distribution. The black lines show the results of maximum likelihood fits to the weighted distributions.
• Figure 4: Magnitude--size relations for our samples. From left to right, panels show: (a) the $M_r$ vs. circular effective radius $R_\mathrm{eff}$ for our sz sample; (b) comparison of the results in panel (a) to those of shen_size_2003; (c) the $M_r$ vs. semi-major axis radius $R_\mathrm{maj}$ for our sz sample; (d) $M_r$ vs. $R_\mathrm{maj}$ for our pz sample. The black circles show results for all the galaxies in a particular sample. Dark red and blue circles show results for the color-selected red and blue subsamples, respectively. Pink and light blue circles show results for the high and low Sérsic index subsamples, respectively. The error bars represent the dispersion of the maximum likelihood fits. In panel (b), the red and gray triangles are data points from shen_size_2003. For ease of comparison, the shaded gray region in (c) reproduces the $R_\mathrm{eff}$ relation for the sz sample from panel (a). The shaded gray region in panel (d) reproduces the result for all the galaxies in the sz sample using $R_\mathrm{maj}$ from panel (c). We ensure that each $M_r$ bin contains at least 100 galaxies. The bottom panels (e-h) show how the corresponding dispersion around each relation varies with $M_r$. In panel (f), the dispersion found by shen_size_2003 is shown with a thin gray line.
• Figure 5: Size-mass relation for the (a) sz and (b) pz samples. Masses for sz galaxies are taken from the GSWLC catalog salim_galexsdsswise_2016. Masses for pz galaxies are taken from WH2024. The line styles and symbols have the same meaning as those in Fig. \ref{['fig:Mr_size']}. The error bars represent the dispersion, $\sigma$, of the log-likelihood. The bottom panels show $\sigma$ vs. $M_\star$ for (c) sz and (d) pz samples. The gray region in panel (b) reproduces the result for all the galaxies in the sz sample from panel (a).