Abstract
Why mutation rates (μ) and genome sizes (GS) vary among species remains a central question in evolutionary genetics. Two influential models, the drift-barrier hypothesis (DBH) and the mutational-hazard hypothesis, propose that effective population size (N(e)) shapes these traits via the efficiency of selection, predicting higher μ and larger genomes in small populations. A recent comparative analysis of vertebrates reported a significant negative correlation between N(e) and μ, interpreted as support for the DBH. Using phylogenetic path analysis, we reanalyze the same dataset of 55 vertebrate species spanning mammals, birds, reptiles, and fishes, in which μ was estimated from high-coverage parent-offspring trios, while explicitly controlling for six life-history traits within a causal framework that tests model-implied conditional independencies. We show that the reported N(e)-μ association is entirely mediated by generation time (GT), which independently influences both variables; once this "back-door" path is blocked, N(e) has no detectable effect on μ. Once GT is accounted for, mating system shows the largest association with μ. Parallel analyses of GS within the same validated life-history framework reveal that GS is unrelated to N(e), μ, or their interaction and is decoupled from the life-history covariation that strongly structures μ. These results are robust to alternative N(e) estimators and causal model formulations. Together, our findings indicate that N(e) provides little explanatory power for variation in μ or GS across vertebrates, challenging the presumed universality of drift-limited genome evolution.