Environmental & Spatial Datasets
We retained four environmental PCs, explaining 92.6% of variance in the raw environmental data (Table 2; see also Table S5 for estimations of variance, eigenvalues, and loadings). The first PC described winter AT (mean winter AT and minimum AT of the coldest month). PC2 was best described by summer AT and SST (mean and minimum summer SST, maximum summer AT, and maximum AT of the warmest month). The third PC was largely representative of wave exposure, but also included annual temperature range. PC4 was described by a combination of summer AT and SST (maximum summer SST, maximum AT of the warmest month), as well as annual temperature range.
Estimations of spatial (dbMEM) and dispersal (AEM) distances produced eight dbMEM vectors representing nonlinear gradients in spatial structure (Table 2, Fig. S3), and three AEM vectors representing current-mediated larval connectivity (Table 2).