Biogerontology · Oct 21, 2026
Environmental variability versus extrinsic mortality as correlates of mammalian lifespan: a pre-specified comparative analysis with phylogenetic and deep-time evidence
By Li D
What explains the large variation in lifespan across species? I examine an information-turnover proposition-that ancestral environmental variability should shorten captive lifespan after controlling for body mass, temperature and extrinsic mortality (P1), with a possible lineage-fate interaction (P2)-against the classical extrinsic-mortality prediction that safer niches permit longer life. I state the decision rules in advance rather than register a third-party protocol. Integrating AnAge (4,219 records), PanTHERIA, WorldClim/CHELSA climate and a 4,100-species mammal phylogeny with Pagel's-λ phylogenetic generalized least squares (PGLS), together with Paleobiology Database fossils across two mass extinctions, the lifespan-mass scaling reproduces (slopes 0.158 mammals, 0.212 birds; P < 10⁻¹⁴⁰). P1 is not supported under the most conservative specifications: weak proxy signals attenuate or reverse at λ = 1 and under family fixed effects, and the set of analyses is best read as overlapping sensitivity and robustness checks rather than independent falsifications; equivalence intervals for the climate coefficients exclude large (arboreality-sized) effects but not small ones. A basal-metabolic-rate correlate likewise dissolves at λ = 1. Turning the same pipeline to niche safety, flight, arboreality, aquatic life and hibernation carry positive mass-corrected coefficients in raw and maximum-likelihood-λ contrasts; at λ = 1 in a joint model (n = 652) only multiply-origin arboreality is significant (b = + 0.106 dex, 95% CI 0.065-0.147, P = 5.8 × 10⁻⁷), whereas flight and aquatic life are inconclusive with wide confidence intervals. Ancestral reconstruction recovers dozens of independent arboreal gains (42 by parsimony; 45-50 by maximum-likelihood; stochastic-mapping median 70-81, 95% range ≈ 57-94) versus one to a few for flight or aquatic life. Bivalve geographic breadth is associated with survival across the K-Pg and end-Permian extinctions, although the effect weakens after controlling for sampling completeness and is a genus-level extinction phenomenon rather than an individual-lifespan mechanism. Turtles live ≈ 1.8 × longer than same-mass squamates (turtle-squamate comparison n = 179, P = 8 × 10⁻⁸). Human lifespan is an extreme outlier on the scaling plane. The comparative evidence is therefore most consistent with a strong association between extrinsic-mortality regime (niche safety) and lifespan, and with no detectable independent role for environmental fluctuation rate at the effect sizes this study could resolve; I report where inference is inconclusive rather than claim a null effect or a decisive causal setting. The focal analyses comprise 977 mammals and 1,063 birds for the scaling law, 798 species with climate data, 878 species with trait contrasts, 652 species in the joint phylogenetic model, and 349 species with basal metabolic rate (298 in the tree-matched model); each result reports its own sample size.