Flexible Models for Galaxy Star Formation Histories Both Shift and Scramble the Optical Color-M/L Relationship

TL;DR

This study compares simple and sophisticated galaxy models to understand the origin of the tight optical color–mass-to-light ratio relationship, revealing that complex models introduce more scatter and redshift evolution, impacting galaxy property estimates.

Contribution

It demonstrates how nonparametric star formation histories affect the color–$M_*/L$ relationship and introduces hierarchical Bayesian modeling to refine galaxy property inferences.

Findings

Prospector-$ m extalpha$ yields higher $M_*/L$ and redder colors.

Complex models increase scatter and show redshift evolution.

Hierarchical Bayesian model reduces uncertainties for faint galaxies.

Abstract

The remarkably tight relationship between galaxy optical color and stellar mass-to-light ratio ($M_*/L$) is widely used for efficient stellar mass estimates. However, it remains unclear whether this low scatter comes from a natural order in the galaxy population, or whether it is driven by simple relationships in the models used to describe them. In this work we investigate the origins of the relationship by contrasting the derived relationship from a simple 4-parameter SED model with a more sophisticated 14-dimensional Prospector-$\alpha$ model including nonparametric star formation histories (SFHs). We apply these models to 63,430 galaxies at $0.5<z<3$ and fit a hierarchical Bayesian model (HBM) to the population distribution in the $(g-r)$--$\log(M/L_g)$ plane. We find that Prospector-$\alpha$ infers systematically higher $M_*/L$ by 0.12 dex, a result of nonparametric SFHs producing older ages, and also systematically redder rest-frame $(g-r)$ by 0.06 mag owing to the contribution from nebular emission. Surprisingly, the combined effects of the $M_*/L$ and $(g-r)$ offsets produce a similar average relationship for the two models, though Prospector-$\alpha$ produces a higher scatter of 0.28 dex compared to the simple model of 0.12 dex. Also, unlike the simple model, the Prospector-$\alpha$ relationship shows substantial redshift evolution due to stellar aging. These expected and testable effects produce overall older and redder galaxies, though the color--$M_*/L$ relationship is measured only at $0.5<z<3$. Finally, we demonstrate that the HBM produces substantial shrinkage in the individual posteriors of faint galaxies, an important first step toward using the observed galaxy population directly to inform the SED fitting priors.