Longevity is heritable and negatively genetically correlated between the sexes in yellow-bellied marmots

Longevity, a major fitness component, is heritable in multiple species, including both captive and wild populations, and often varies widely between the sexes. The sex-specific genetic architecture of longevity, however, has rarely been estimated in wild populations, despite its potentially large implication for the evolutionary dynamic of a species. Using a long-term study of wild yellow-bellied marmots, a hibernating rodent, we estimated sex-specific additive genetic variance V_A and the cross-sex genetic correlation r_fm of longevity. Given the challenges associated with accurately measuring longevity in the wild, we used a new analytical approach based on a Censored Poisson distribution allowing us to integrate measurement errors on longevity in the model. Our approach revealed moderate and comparable V_A in both sexes and a strongly negative r_fm , albeit with large credible intervals. This contrasts with the results from a classic model with a restricted dataset for which V_A in males was estimated as zero, rendering the r_fm inestimable and uninterpretable. Our results suggest that studying selection and evolution while focusing on only one sex can lead to erroneous predictions given that, in marmots, selection pressures increasing longevity in one sex would inherently select for the reverse effect in the other sex. Taken together, this suggests the possible presence of a self-reinforcing feedback loop for the development of different life-history strategies among sexes in marmots, with long-lived females producing short-lived males who must maximize early life reproductive success ("live-fast die-young" strategy) and vice versa. Our study provides rare evidence of heritable longevity in a wild population and highlights how genetic conflicts between the sexes may constrain evolution and help maintain sex-specific genetic variance in fitness.