Understanding intraspecific variation may enable a more accurate prediction of species’ response to current and predicted climate change. Big bluestem (Andropogon gerardii) is the ecologically dominant grass in tallgrass prairies. With a wide distribution across a west/east precipitation gradient (40-119 cm/yr), and a south/north temperature gradient (15-5oC mean annual temperature), we expect intraspecific variation in drought and thermal tolerance. Our study utilized phenotypic and genetic data from geographically distributed populations across the Midwest to explore phenotype-climate and genotype-climate relationships for current and future climate scenarios. For each of the populations, we grew plants from seed and measured blade width, height, biomass, and chlorophyll absorbance. We determined genotypes using Genotyping-by-Sequencing to investigate genetic divergence, and to associate climate variables to outlier SNPs. Conventional methods use species distribution modeling to predict response to climate change but fails to incorporate intraspecific variation. To improve on modeling, we used phenotype and genotype data in generalized additive models and a generalized joint attribute model to predict current and future phenotypes and genotypes across the Midwest.
Results/Conclusions
Distinct population phenotypes were characterized by short stature, low biomass, and narrow leaves in dry regions with height, biomass, and leaf width increasing along a longitudinal cline corresponding with increased precipitation. Genotyping-by-Sequencing identified 6618 SNPs and outlier analysis identified 197 SNPs under divergent selection. Population structure showed evidence for four genetic groups and genetic outliers corroborated phenotypic variation of A. gerardii across the Midwest. Phenotype, genotype, and allele frequency of SNPs in outlier genes were used in a GAM/GJAM approach to predict current and future distributions of phenotype and allele frequency of outlier SNPs. Phenotypes from dry areas were predicted to expand through the Midwest, eclipsing phenotypes from wet areas, provided adequate migration. Similarly, outlier candidate gene SNPs, potentially responsible for phenotype differences, show clear clines across the Midwest in response to climate. The novel GAM/GJAM approach greatly refines species distribution models that assume no intraspecific variation and may more accurately predict species’ response to climate change.