Genomic offset is not predictive of recent demographic trends in Lycaeides butterflies

Le résumé fourni par la source

Genomic offset (GO) is increasingly used to predict population maladaptation risk under climate change, with larger offsets assumed to indicate greater vulnerability. Despite rapid adoption in conservation planning, it remains unclear how sensitive GO estimates are to key methodological choices, including SNP set composition, genotype-environment association (GEA) methods, and the specific GO metric used. Empirical validation against observed population dynamics also remains limited. Here, we evaluate the methodological robustness and predictive performance of GO using multidecadal demographic monitoring data from Lycaeides butterflies, a system with short generation times and high fecundity that may facilitate rapid adaptive responses. GO estimates were broadly consistent across SNP sets, regardless of composition or size, with climate-associated and randomly selected SNPs yielding largely concordant values. Consistency across GEA methods was moderate and depended on the SNP set used. In contrast, GO metrics differed substantially in the magnitude of maladaptation estimated, suggesting they capture distinct biological signals and should not be treated as interchangeable. Crucially, GO was a poor predictor of observed population trends, regardless of SNP set composition, GO metric, or GEA method, both at sites used to fit GEA models and when extrapolated to independent demographic sites. These findings suggest that, while GO provides a valuable conceptual framework for assessing potential maladaptation, its quantitative estimates and predictive power are sensitive to methodological choices and species-specific biological context. We therefore urge careful alignment of GO metric assumptions with conservation objectives, along with rigorous empirical validation, before GO estimates are used to inform management decisions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.