Using community-level modeling to understand and map current and future spatial patterns of adaptive genetic variation

Background/Question/Methods

A major challenge in predicting the impacts of climate change on biodiversity is moving beyond species-level models and towards greater consideration of intraspecific variation in climatic tolerances due to local adaptation. Overcoming this challenge ultimately hinges on the development of spatial modeling frameworks capable of (1) linking genetic and environmental variation to identify loci having adaptive significance and (2) using these gene-environment relationships to map the geographic distribution of adaptive genetic variation under current and future climate. While landscape genomics and high throughput sequencing are beginning to facilitate the study of adaptive genetic variation in non-model organisms, modeling techniques that can translate vast quantities of genomic data into spatial predictions remain in their infancy. In this contribution, we demonstrate how community-level modeling methods typically used to simultaneously model all species in an assemblage can be powerfully applied to the problem of analyzing and mapping genomic variation by modeling all SNPs in a genome. Using balsam poplar (Populus balsamifera) as a case study, we draw on the long history of using trees as model systems to understand adaptive genetic variation relation to geography and climate. 

Results/Conclusions

Our results suggest community-level modeling methods can be powerfully applied to the challenge of scaling from population-level genomic variation to landscape scale impacts of global change. We demonstrate that these new techniques can (1) accommodate non-linearity in the exploration of gene-environment relationships, (2) handle large genomic datasets that include numerous rare, low frequency alleles, (3) provide insight into regions of the genome ostensibly under selection, and (4) generate maps of how adaptive genomic diversity is predicted to vary across the landscape. When projected to future climates, these models identify the potential impacts of climate change at the gene level and how these impacts vary spatially between populations. This represents an advance beyond the application of species-level SDMs, which assume all populations within a species respond identically to environmental change, and existing linear techniques for identifying molecular markers associated with climate adaptation. Collectively, we demonstrate that novel application of community-level modeling methods offer unique strengths for exploring and mapping adaptive genomic variation that are complimentary to or improve upon the currently available techniques in landscape genomics.