FIGURE 2 - Compositional turnover functions from GF models fit to different sets of outlier and randomly selected SNPs. Each bold, colored curve in each panel is an aggregate turnover function from a different GF model representing the R 2-weighted average of the individual turnover functions (shown in Figure 3) fitted to the number of SNPs indicated in the legend. For example, the green line indicates the overall turnover function from a GF model fit to minor allele frequencies of 310 significant outliers identified bylfmm analyses, 146 of which had an R 2greater than zero in the GF model and therefore contributed to the aggregate function for this model. The thin black lines indicate aggregate turnover functions from GF models fit to 999 different sets of 500 randomly selected SNPs and the teal line represents the combined aggregate of these 999 models. The location of where each population occurs on each gradient is labeled along the x-axes (for clarity only a subset of populations are shown). The vertical dashed and dotted lines indicate where the VT and IH common gardens respectively fall along each gradient given the local conditions during the common garden experiments in 2014-2015.