From Cluster Cores to the Low-Density Field: Strong Environmental Quenching of Galaxy Star Formation at Low Redshift

This paper investigates how galaxy star formation activity depends on environment using a sample of 81,647 SDSS galaxies selected over 0.03 ≤ z ≤ 0.075 and 9.7 ≤ log10 (M⋆ /h−2 M⊙) ≤ 11.0, including 18,426 members from 572 clusters in the GalWCat19 catalog. The authors characterize environment in two complementary ways: (1) nearest-neighbor density for the full sample, and (2) clustercentric radius and host halo mass for GalWCat19 clusters.

The sSFR distribution remains bimodal across all environments, with distinct quenched and star-forming components. As local density increases, the quenched component becomes more prominent, while the characteristic sSFR of the star-forming component decreases by approximately 0.29–0.35 dex from the lowest- to highest-density classes. Within clusters, the quenched fraction decreases with increasing projected clustercentric radius, while the star-forming peak shifts by approximately 0.42 dex toward lower sSFR from the outskirts to the inner cluster region. This extends the picture from previous studies, in which environmental trends are primarily associated with changes in the quenched fraction, by showing that galaxies remaining in the star-forming population also exhibit systematically suppressed sSFR in denser environments. The dependence on host halo mass is weaker and is most apparent among lower-stellar-mass galaxies in the inner cluster regions. By measuring the environmental quenching efficiency at fixed stellar mass, the authors find excess quenching in cluster environments beyond that expected from stellar-mass quenching alone. These results show that environment is associated not only with an increased probability of quenching, but also with suppressed star formation among galaxies that remain star forming, with local density and clustercentric radius showing the strongest associations.

The analysis addresses three main questions: how the mean SFR, mean sSFR, quenched fraction, and full sSFR distribution vary with local galaxy density, and in particular whether increasing density changes only the relative weights of the quenched and star-forming populations or also shifts the characteristic sSFR of galaxies that remain star forming; how the star formation properties of cluster members depend on projected clustercentric radius, host halo mass, and projected phase-space position; and whether cluster galaxies exhibit excess quenching beyond that expected from internal stellar-mass quenching alone.

Galaxies are classified as star-forming or quenched using a threshold of log10 (sSFR/yr−1) = −11.09, the minimum between the two peaks of the bimodal sSFR distribution. The approximately volume-limited sample is defined by 0.03 ≤ z ≤ 0.075 and 9.7 ≤ log10 (M⋆ /h−2 M⊙) ≤ 11.0. All binned measurements and distribution fits use 1/Vmax weighting to reduce residual selection-related biases.

For the statistical large-scale environment, the authors use a projected nearest-neighbor density estimator ΣN = N/(πd2N) for N = 5, 10, and 15, dividing galaxies into four density classes using the 20th, 50th, and 80th percentiles. The weighted mean log10 (SFR) and log10 (sSFR) decrease monotonically with increasing density for all three values of N, while the quenched fraction rises from FQ ≃ 0.5 in the lowest-density bins to FQ ≳ 0.8 in the highest-density bins.

The Gaussian-mixture decomposition of the sSFR distributions shows that from the lowest-density class, C10,1, to the highest-density class, C10,4, the quenched component becomes progressively more prominent, with its mixture weight increasing from wQ = 0.46 to wQ = 0.62. The peak of the quenched component shifts from µQ ≃ −11.85 to µQ ≃ −12.02 (∆µQ ≃ −0.17 dex), while the star-forming peak shifts more strongly from µSF ≃ −10.30 to µSF ≃ −10.62 (∆µSF ≃ −0.33 dex). The same qualitative behavior is recovered for the Σ5 and Σ15 estimators.

For the cluster-scale environments, the radial classes show that the inner cluster region, R1, has a quenched mixture weight of wQ = 0.66, decreasing to wQ = 0.64 in R2 and wQ = 0.52 in R3. The star-forming peak is located at µSF ≃ −10.94 in R1 compared to µSF ≃ −10.52 in R3, giving ∆µSF ≃ −0.42 dex. The quenched-component mean changes by only ∆µQ ≃ −0.08 dex. The halo-mass classes show weaker variation: the quenched mixture weight increases from wQ = 0.56 in M1 to wQ = 0.63 in M3, but the peak positions of both Gaussian components vary only mildly and non-monotonically.

The joint radial–halo-mass analysis confirms that at fixed halo mass, the star-forming peak moves systematically toward lower sSFR from R3 to R1. For example, µSF changes from −10.53 in R3 M1 to −10.79 in R1 M1, from −10.54 in R3 M2 to −11.06 in R1 M2, and from −10.48 in R3 M3 to −11.04 in R1 M3. The shift in the star-forming peak from R3 to R1 is ∆µSF ≃ −0.26, −0.53, and −0.56 dex for M1, M2, and M3, respectively, while the quenched-component mean changes by only approximately −0.05, −0.09, and −0.08 dex.

The radial profiles of quenched fraction, mean SFR, and mean sSFR as functions of Rp /R200 show that the highest values of FQ are found in the inner cluster regions, where FQ ≳ 0.8, declining to approximately 0.5–0.6 in the cluster outskirts and infall region. The inner cluster region also shows a tendency for the quenched fraction to increase with host halo mass. The mean sSFR increases outward in all halo-mass classes, indicating that the radial trend is not driven solely by differences in stellar mass.

The projected phase-space analysis shows that galaxies with the lowest sSFR are preferentially concentrated toward small Rp /R200, while galaxies with higher sSFR become more common at larger projected radii. The highest quenched fractions occur predominantly in the inner cluster regions. Comparing the M1, M2, and M3 samples cell by cell, 67% of common cells have a higher quenched fraction in M3 than in M1, with a median difference of ∆FQ = 0.066. Within R1, 74% of common cells have FQ (M3) > FQ (M1), increasing to 83% in R2, compared with 61% in R3, with median differences of 0.086, 0.132, and 0.034, respectively.

The environmental quenching efficiency, defined as ϵenv = (FQ,env − FQ,ref)/(1 − FQ,ref) using the lowest-density class C10,1 as the reference population, is positive across all radial and halo-mass classes. The ordering ϵenv (R1) > ϵenv (R2) > ϵenv (R3) is maintained across the full stellar-mass range. In the inner cluster region, R1, the environmental quenching efficiency generally increases with host halo mass, particularly over the three lowest stellar-mass bins. A similar halo-mass dependence is present in R2, while in R3 the efficiencies of the three halo-mass classes are more similar and do not follow a consistent ordering.

The authors conclude that environmental suppression of star formation is already detectable within the star-forming galaxy population, rather than appearing only through an increase in the quenched fraction. This extends the picture emphasized by Wetzel et al. (2012, 2013), in which environmental trends are primarily reflected in changes in the quenched fraction. The results show that local density and projected clustercentric radius show the strongest associations with both effects, while host halo mass provides a weaker secondary dependence, most apparent near cluster centers. The authors note that successful models should simultaneously reproduce the bimodal shape of the sSFR distribution, the environmental variation in the quenched and star-forming components, and the shift of the star-forming peak toward lower sSFR in denser and more central environments.

Improvements to AI Systems:

1. Environment-Aware Galaxy Evolution Models: AI systems can now incorporate both local density (nearest-neighbor density) and clustercentric radius as explicit input features, enabling more accurate predictions of star formation rates, sSFR distributions, and quenching probabilities for galaxies across diverse environments.

2. Bimodal sSFR Decomposition: AI models can be trained to decompose galaxy populations into quenched and star-forming components using Gaussian mixture models, with the ability to predict not only the relative weights of these components but also the shift in the star-forming peak (e.g., 0.3–0.4 dex suppression) as a function of environment.

3. Phase-Space Quenching Predictor: Improved AI systems can predict the quenched fraction and sSFR based on projected phase-space position (Rp/R200 and velocity offset), enabling real-time classification of galaxies as recently infalling, virialized, or in the outskirts of clusters.

4. Halo-Mass-Dependent Corrections: AI models can now account for the secondary dependence of star formation suppression on host halo mass, particularly for lower-stellar-mass galaxies in inner cluster regions, improving predictions of environmental quenching efficiency (ε env) at fixed stellar mass.

5. Residual-Selection-Bias Correction: AI systems can apply 1/Vmax weighting techniques learned from this study to correct for selection biases in large spectroscopic surveys, improving the fidelity of inferred galaxy property distributions.

6. Multi-Scale Environmental Feature Extraction: AI systems can be enhanced to simultaneously process local (nearest-neighbor density), intermediate (clustercentric radius), and global (halo mass) environmental scales, capturing the full hierarchical nature of galaxy environment.

7. Predictive Models for Quenching Efficiency: AI systems can now estimate environmental quenching efficiency as a function of stellar mass, radius, and halo mass, allowing for more precise forecasts of galaxy evolution in cosmological simulations.

8. Synthetic Galaxy Population Generators: Generative AI models can be trained to produce realistic sSFR distributions that match the observed bimodality, including the environmental-dependent shifts in both quenched and star-forming peaks, for use in mock catalogs and survey simulations.

9. Anomaly Detection in Galaxy Surveys: AI systems can flag galaxies whose sSFR deviates significantly from the expected environmental trends (e.g., star-forming galaxies in dense cluster cores), identifying rare objects that may represent recent mergers, AGN feedback, or other stochastic processes.

10. Transfer Learning for Higher-Redshift Studies: The quantitative relationships between environment and star formation suppression established here can be used to calibrate AI models applied to higher-redshift surveys (e.g., JWST, Euclid), where similar bimodal sSFR distributions are expected but with different environmental scaling.

We investigate how galaxy star formation activity depends on environment using a sample of 81,647 SDSS galaxies selected over 0.03 at most z at most0.075 and 9.7 at most 10(M/h-2M) at most11.0, including 18,426 members from 572 clusters in the GalWCat19 catalog. We characterize environment in two complementary ways: (1) nearest-neighbor density for the full sample, and (2) clustercentric radius and host halo mass for GalWCat19. The sSFR distribution remains bimodal across all environments, with distinct quenched and star-forming components. As local density increases, the quenched component becomes more prominent, while the characteristic sSFR of the star-forming component decreases by approximately 0.29 -- 0.35 dex from the lowest- to highest-density classes. Within clusters, the quenched fraction decreases with increasing projected clustercentric radius, while the star-forming peak shifts by approximately 0.42 dex toward lower sSFR from the outskirts to the inner cluster region. This extends the picture from previous studies, in which environmental trends are primarily associated with changes in the quenched fraction, by showing that galaxies remaining in the star-forming population also exhibit systematically suppressed sSFR in denser environments. The dependence on host halo mass is weaker and is most apparent among lower-stellar-mass galaxies in the inner cluster regions. By measuring the environmental quenching efficiency at fixed stellar mass, we find excess quenching in cluster environments beyond that expected from stellar-mass quenching alone. These results show that environment is associated not only with an increased probability of quenching, but also with suppressed star formation among galaxies that remain star forming, with local density and clustercentric radius showing the strongest associations.