2.4 Differential gene expression analysis
We used DESeq2 to compute log-fold changes in response to environmental contrasts (Love et al., 2014). Prior to running DESeq2, we removed genes with mean read counts less than three to eliminate bias from very low expressed genes. We encoded the non-stressed and stressed populations for each contrast as “control” and “treatment” in DESeq2, respectively. For sequencing runs from Dixon et al., 2014, we ran DESeq2 on all the runs, including the genotyping replicates used for transplanting in the original study, and added “transplant location” as an additional variable to our analysis to remove the effect of transplant location.