Two Layers of Galaxy Aging in MaNGA DynPop: A Principle-Level Proof of Concept

Version note: Version 3 corrects the size control used in version 2, which measured galaxy size in arcseconds rather than kiloparsecs; at fixed stellar mass, the younger galaxies in this sample lie farther away, so angular size partly encoded distance. With physical size the controlled residual is weaker, depends on the control specification, and in Table 1 vanishes with intrinsic-SPS controls (Section 2). Version 2's statement that it holds in 7 of 8 bins across all three control sets overstated its robustness. Abstract We test a principle-level prediction from an age-dependent gravity framework using the public MaNGA DynPop DR17 catalogs. The prediction is that younger galaxies, at fixed stellar mass, should show larger mass-to-light discrepancies between dynamical and stellar-population estimates. We join three public DynPop products (JAM dynamical catalogs, stellar-population/star-formation-history catalogs, and circular-velocity-curve tables), apply quality cuts, and split 5,952 galaxies into eight stellar-mass bins, then compare the youngest and oldest quartiles within each bin. The raw mass-to-light discrepancy (DML) is larger for young galaxies in 8/8 mass bins (observed SPS) and 6/8 bins (intrinsic SPS). A component decomposition shows that this raw result is mostly driven by the SPS denominator: older stellar populations have higher stellar mass-to-light ratios, as standard stellar-population synthesis predicts. Under a reduced control model (SPS mass-to-light, metallicity, and structural variables, with galaxy size in kpc), a smaller dynamical residual persists in 7/8 mass bins with observed-SPS controls and 6/8 with both SPS terms, but not with intrinsic-SPS controls (3/8). The bin counts depend on the control specification, including whether distance is controlled (Section 2), but a linear age term stays significant, with younger galaxies higher, in every full-sample fit. Parametric dark-matter fractions do not show the same signal. We present this as a proof of concept and invitation to specialist replication, not a detection claim.