Genomic Offsets Predict Survival With Low Accuracy in a Marine Common Garden

Predicting how populations will respond to climate change is an urgent priority in evolutionary and conservation biology. Genetic and environmental influences have long been known to impact responses to shifting climates, giving rise to the development of genomic offset methods that attempt to predict the degree of maladaptation following environmental change. Although these methods have the potential to improve our understanding of populations' responses to climate change, questions remain about how well they predict fitness in a diversity of systems. Here, we pair genomic offsets with a large-scale common garden experiment to investigate how fitness of a marine mollusc, Crassostrea virginica, varies across environmental gradients and to validate offset predictions against empirical fitness. We find that at two common garden sites in the Chesapeake Bay, Northern oysters had the lowest survival and growth, whereas Southern oysters had similar survival and growth to local wild oysters and selection lines. We compared survival and length to offset predictions, finding higher method performance at the lower salinity, moderate disease site than at the higher salinity, higher disease site. The inclusion of disease in the offset model training additionally impacted offset method performance, though the effect varied between common garden sites, fitness proxies and across methods. Despite a nearly 50% difference in survival across sites, the magnitude of offset predictions to both common gardens was similar, highlighting issues with extrapolating offset predictions across species ranges. Performance of genomic offset methods in this system was not adequate to make real-world predictions to environmental stressors.