The quenched fraction of satellites around simulated Milky Way-mass galaxies

TL;DR

The paper tackles how satellite quenching around Milky Way–mass hosts depends on satellite stellar mass and distance from the host, by comparing three state-of-the-art simulations (FIREbox, FIRE-2 zoom-ins, TNG50) with SAGA and ELVES observations. It finds a robust, mass-dependent increase in quenching toward lower-mass satellites, consistent across all datasets, but reveals that radial quenching trends are sensitive to host environment and halo physics. The FIRE-2 zoom-ins show a pronounced environmental imprint: paired MW–M31 analogs yield suppressed inner quenching and a reversed radial trend, unlike isolated hosts and the other simulations. These results underscore the role of host environment and assembly history in shaping satellite evolution and motivate forward-modeling and deeper, environment-focused studies to interpret observed quenching patterns reliably.

Abstract

We compare satellite quenched fractions across three cosmological simulation suites (FIREbox, the FIRE-2 zoom-ins, and IllustrisTNG50) and observational datasets from SAGA, ELVES, and the combined satellite population of the Milky Way and M31. To enable consistent comparisons, we select Milky Way-mass hosts with $M_{\rm halo} = 10^{11.9}$ - $10^{12.2} \, M_{\odot}$ and satellites with stellar masses of $10^{7}$ - $10^{10}\, M_{\odot}$, applying uniform projected apertures and a common quenching definition. All three simulations reproduce the strong observed trend that lower-mass satellites are more likely to be quenched, closely matching the stellar-mass dependence seen in SAGA, ELVES, and the MW+M31 system. This agreement indicates that the mass dependence of satellite quenching is a robust outcome of contemporary galaxy formation models. Radial trends, however, show meaningful differences. SAGA and ELVES exhibit gently declining quenched fractions with projected distance, reflecting strong environmental quenching at small radii. TNG50 most closely matches this behavior, FIREbox, remains consistent with with a nearly flat trend within uncertainties, and the FIRE-2 zoom-ins show suppressed inner quenched fractions driven almost entirely by their paired MW-M31 hosts, which lack high-mass satellites and show strong radial segregation between star-forming and quenched systems. This environmental imprint suggests that host environment and assembly history can influence satellite quenching outcomes and may contribute to diversity across simulations. Overall, while the simulations consistently recover the stellar-mass dependence of quenching their radial trends vary, highlighting the influence of host-halo conditions and motivating deeper exploration of how host environments shape satellite quenching.

Figures (7)

• Figure 1: Stellar mass–halo mass (SMHM) relation for host samples drawn from FIREbox (maroon), the FIRE-2 zoom-ins (blue), TNG50 (indigo), and the SAGA survey (green). All simulated hosts lie within the adopted Milky Way–mass range of $10^{11.9} \leq M_{\rm halo}/M_\odot \leq 10^{12.2}$, while the full SAGA sample is retained to preserve statistical completeness and account for uncertainties in cataloged halo masses. The Milky Way and M31 are plotted as a yellow star and pentagon using $M_{\star,\rm MW}=5\times10^{10}\,M_\odot$, $M_{\rm halo,MW}=10^{12}\,M_\odot$, $M_{\star,\rm M31}=10^{11}\,M_\odot$, and $M_{\rm halo,M31}=2\times10^{12}\,M_\odot$. Histograms of host halo mass (top) and stellar mass (right) share axes with the main panel. All datasets with reported halo masses have medians, $\log_{10}(M_{\rm halo}/M_\odot)\sim12.0$ (FIREbox, FIRE-2 zooms, TNG50, and SAGA). Median stellar masses are $\log_{10}(M_\star/M_\odot)=10.7$ (FIREbox), $10.8$ (FIRE-2 zooms), $10.6$ (TNG50), $10.5$ (SAGA), and $10.7$ (ELVES). Because ELVES does not provide halo-mass estimates, its hosts appear only in the stellar-mass histogram in golden.
• Figure 2: Kernel density estimates of host halo radii, $r_{\rm halo}$, for Milky Way–mass systems drawn from FIREbox (maroon), the FIRE-2 zoom-ins (blue), and TNG50 (indigo). The distributions agree closely, with median radii of 304.6 kpc, 320.3 kpc, and 320.2 kpc, respectively.
• Figure 3: Projected radial distributions of satellites for all five datasets: FIREbox (maroon), FIRE-2 zoom-ins (blue), TNG50 (indigo), SAGA (green), and ELVES (golden). For every dataset, we plot the mean cumulative fraction of satellites per host as a function of projected distance, normalized to the number of satellites within 300 kpc. Shaded bands show the $1\sigma$ host-to-host scatter for the simulations. Following the SAGA survey strategy, we exclude the inner 10 kpc (gray region), where host-galaxy light severely limits satellite detectability. All five samples exhibit broadly similar radial trends, with only mild differences in the degree of central concentration.
• Figure 4: Satellite stellar mass functions (SMFs) for all hosts in FIREbox (maroon), FIRE-2 zoom-ins (blue), TNG50 (indigo), SAGA (green), and ELVES (golden). Thin lines show the SMF for each individual host, while thick lines indicate the median SMF of each dataset. The inset panel shows the distribution of satellite counts per host. The median numbers of satellites in the adopted mass range ($10^{7}$--$10^{10}\,M_\odot$) are 1 for FIREbox, 3 for the FIRE-2 zooms and TNG50, 3 for SAGA, and 4.5 for ELVES.
• Figure 5: Satellite quenched fractions as a function of stellar mass (left) and projected distance from the host (right). Green circles and golden squares show measurements from the SAGA and ELVES surveys with Poisson uncertainties, and black stars mark the stacked quenched fractions of the combined Milky Way + M31 satellite sample from Wetzel2015, with error bars giving the 68% confidence interval. Simulation predictions from FIREbox (maroon), the FIRE-2 zoom-ins (blue), and TNG50 (indigo) are shown with shaded regions representing Poisson uncertainties on the stacked quenched fractions; for the radial trends, the simulation curves incorporate all three orthogonal sightline projections discussed in Section \ref{['sec:satellite_selection']}. All three simulation suites successfully reproduce the strong rise in quenched fraction toward low stellar masses, in excellent agreement with SAGA, ELVES, and the Local Group hosts. Radial trends show greater diversity: TNG50 closely matches the observed gradual decline with projected distance, FIREbox remains statistically consistent with a nearly flat profile once uncertainties are considered, and the FIRE-2 zoom-ins exhibit a suppressed quenched fraction at small radii ($\sim 50$ kpc), a behavior traced in Section \ref{['sec:results_environment']} to the paired FIRE-2 host systems.