The influence of external environment at cosmic noon on the subsequent evolution of galaxy stellar mass

TL;DR

This work addresses how high-redshift environments influence the later stellar-mass growth of galaxies and whether including environmental information improves progenitor–descendant matching. Using random forest regression on two state-of-the-art cosmological simulations, EAGLE and MAGNETICUM, the authors find that the spherical overdensity within 1 cMpc, $δ_{1,\mathrm{sp}}$, is the strongest predictor of a galaxy’s $z\sim0$ stellar mass across a range of progenitor masses, and that environment can rival progenitor mass in predictive power for $z\gtrsim2$. Tracking galaxies by environment reveals that overdense regions yield descendants with $M_{*}$ up to 3–8 times larger than those from underdense regions, with growth driven by a mix of in-situ star formation and ex-situ accretion, depending on the mass regime. Incorporating $δ_{1,\mathrm{sp}}$ into number-density based predictions reduces the residual scatter by about 20–35% at $z\sim0$, demonstrating tangible gains in progenitor–descendant matching and offering observational pathways to account for environmental effects in cosmic evolution studies.

Abstract

Connecting high-redshift galaxies to their low-redshift descendants is one of the most important and challenging tasks of galaxy evolution studies. In this work, we investigate whether incorporating high-redshift environmental factors improves the accuracy of matching high-redshift galaxies to their $z\sim0$ descendants, using data from the EAGLE and MAGNETICUM simulations. Using random forest regression, we evaluate the relative importance of a set of environmental metrics at $z\sim3$ in determining the stellar mass of descendant galaxies at $z\sim0$. We identify the spherical overdensity within 1 cMpc ($δ_{1,\mathrm{sp}}$) as the most important environmental predictor. Tracking galaxies at $z\sim3$ with similar initial stellar masses but different $δ_{1,\mathrm{sp}}$ values, we find that, across all mass bins in both simulations, high-density environments produce $z\sim0$ descendants with median stellar masses up to eight times higher than the descendants of galaxies in low-density environments. For galaxies with $M_{*}\lesssim10^{10}M_{\odot}$, the difference is attributable to more merger-induced mass growth in high-density environments, whereas for higher-mass galaxies, it results from a combination of enhanced in-situ star formation and greater external mass accretion. By assessing the importance of overdensity across multiple scales and redshifts, we find that at $z\gtrsim2$, environmental factors become as important as stellar mass in predicting the stellar mass of $z\sim0$ descendants. Compared to using stellar mass at $z\sim3$ alone, incorporating $δ_{1,\mathrm{sp}}$ reduces the scatter in the residuals between the predicted and actual stellar masses by approximately 20% in EAGLE and 35% in MAGNETICUM.

Figures (11)

• Figure 1: The relative importance (R.I.) of both the $z\sim3$ environment, as characterised by different metrics, and the stellar mass of the progenitor galaxies themselves, in determining the stellar mass of the $z\sim0$ descendants of galaxies with $9.0<\log (M_{*}/M_{\odot})\leq9.5$ (left), $9.5<\log (M_{*}/M_{\odot})\leq10.0$ (middle), and $\log (M_{*}/M_{\odot})>10.0$ (right) in Eagle (top) and Magneticum (bottom). The height of each bar denotes the R.I. of each feature, with the error bars representing the standard deviation from 10 independent runs. The bars are colour-coded by feature categories: nearest neighbour measures (blue), aperture measures (yellow), annulus measures (green), sphere measures (pink), and the stellar mass of the progenitor galaxies (gray).
• Figure 2: Same as Figure \ref{['fig:rfr mass bins']}, but for all galaxies with $\log (M_{*}/M_{\odot})>9.0$ in Eagle (top) and Magneticum (bottom).
• Figure 3: The cumulative mass functions (CMFs) of all galaxies with $\log (M_{*}/M_{\odot})>9.0$ at $z\sim0$ (black), $z\sim1$ (blue), $z\sim2$ (green) and $z\sim3$ (red) in Eagle (solid lines) and Magneticum (dashed lines).
• Figure 4: The number density evolution of $z\sim3$ galaxies with $\log (M_{*}/M_{\odot}) \sim 9.1$ (blue), 9.6 (green), 10.1 (yellow), and 10.6 (red) in Eagle (top) and Magneticum (bottom). The horizontal solid lines indicate a constant number density for reference, while the dashed lines show the median number density evolution within each mass bin. The shaded regions represent the scatter between the 20th and 80th percentiles of the number density distribution at each redshift.
• Figure 5: From left to right: the number density evolution of $z\sim3$ galaxies with $\log (M_{*}/M_{\odot}) \sim 9.1$, 9.6, 10.1, and 10.6 in different environments, characterised by the 1-cMpc spherical overdensity($\delta_{1,\mathrm{sp}}$). The top row shows results from the Eagle simulation, while the bottom row corresponds to the Magneticum simulation. In each panel, the solid and dashed black lines have the same meaning as those in Figure \ref{['fig:CND All']}. The median number density evolution tracks and the scatters between the 20th and 80th percentiles of the number density distribution of galaxies belonging to each environmental density bin are indicated by dotted-dashed lines and shaded areas in different colours, respectively.