Population Structure and Ancestry Coefficients
The clustering of individuals into ancestral populations was estimated with sNMF using cluster numbers from 1 to 20. The optimal number of clusters was unclear as the cross-entropy continuously decreases up to K=11, although the relative change in cross-entropy values is smaller after K=6 (Figure S3 ). In the K=3 model, clusters are consistent with morphotypes (N. riversi , intermediate, andN. ingens ), but with some populations showing some admixture among these major lineages. These admixed populations are located in a geographical transition zone where separate lineages could have experienced historical gene flow (Figure S1 ). In the K=4 model, the southern cluster splits into two clusters, revealing a distinct ancestral subpopulation located in the Kings drainage. In the K=5 model, the southernmost cluster further subdivides into west (Kaweah) and east (Kern) ancestral populations. In the K=6 model, the northern sample sites (sites 3 and 4) split from other northern populations and thus subdivide N. riversi (sites 1 to 6) into two clusters. From K=7 to K=11, new clusters mostly emerge around the contact zones of major drainages (Figure S1 ).