Intraspecific variation has been found, in some cases, to exceed that of interspecific variation. We examine intraspecific variation and local adaptation in Andropogon gerardii, the dominant grass of the Great Plains and representing ~70% of biomass. It has wide geographic distribution across the Great Plains precipitation gradient from western Kansas (dry) to Illinois (wet). It is crucial to understand big bluestem responses to climate for conservation, restoration, and agricultural cattle production. Ecotypes (xeric, mesic, wet) were reciprocally planted as ecological communities spanning a rainfall gradient of 500 to 1,200 mm (wet to east) in Colby, Hays, and Manhattan, KS, and Carbondale, IL. We tested for evidence of local adaptation over 5 years using single ecotype plots (community plots seeded with other prairie plants) and plots with all three ecotypes mixed together (community plots containing all three ecotypes to test for experimental selection). Planting of ecotypes as a community and over multiple years is rarely done, but offers the most realistic test of local adaptation. We utilized Genotyping-By-Sequencing to identify SNP markers in plants of known ecotypes to predict ecotype identity from mixed ecotype plots to ascertain which ecotype is winning in each of the sites across the climate gradient.
Results/Conclusions
Principal component analyses and population structure show strong genetic differentiation between xeric and wet ecotypes. Outlier analysis in Bayescan identified 64 markers under divergent selection, including GA1 (a gene known to control internode length and height in plants), in which we observe strong ecotype height differences between xeric and wet ecotypes. Single ecotype community plots suggest local adaptation to drought with the plants from central KS having higher cover in Hays, KS and plants from Illinois having greater cover in its home site of Carbondale, IL. To determine which ecotype is winning in mixed ecotype plots, we used the GBS genotype information from plants of known ecotype, then used this to train a random forest model that could assign unknown individuals from the mixed plots to one of our three ecotypes. Random forest modeling confirms local adaptation as observed in the single ecotype plots for most sites. These multi-year, community plantings show evidence of local adaptation of dry and wet ecotypes in reciprocal gardens across the Great Plains. Ultimately these results will provide recommendations to land managers on which climate-adapted source populations of big bluestem is best suited for conservation and restoration planting in future warmer and drier climates.