Table 1.

Important terminology and acronyms

Abbreviation	Explanation
IBD	Isolation by distance: the expectation of increased genetic distance with increased geographic distance (3, 4)
CSE	Cavalli-Sforza–Edwards distance (38)
Linearized FST	Measure of genetic distance: FST/(1 − FST)
Full	Complete dataset (38 sites, 703 pairwise genetic distances)
Train	Complete dataset excluding one point and its affiliated pairs (37 sites, 666 pairwise genetic distances)
Test	One point and its affiliated pairs (1 site, 37 pairwise genetic distances)
Rtrain	Pearson correlation between predicted and observed enetic distance for training dataset
Rtest	Pearson correlation between predicted and observed genetic distance for testing dataset
Rfull	Pearson correlation between predicted and observed genetic distance for full dataset
RMSEtrain	Root-mean-square error of model using the training dataset
RMSEtest	Root-mean-square error of model using the testing dataset
RMSEfull	Root-mean-square error of model using the full dataset
RF	Random forest: a nonlinear classification and regression tree analysis
RSQtrain	Pseudo-R-squared (% variance explained by the model) built with the training dataset
RSQfull	Pseudo-R-squared (% variance explained by the model) built with the full dataset

See SI Appendix, Table S3 for more details.