Extended Data Fig. 5: Linear display of trends in biodiversity metrics.

This figure is a complement to Fig. 3 in the main text to show intercepts and slopes of trends in biodiversity metrics. Lines represent model predicted trend with 95% confidence interval bands. Statistical output provided in Table S3. Panels A), D), G), J), M), and N) correspond to trends for whole communities. B), E), H), and K) correspond to trends for the local, non-game fish assemblage subsets. C), F), I) and L) correspond to trends for the introduced, game fish assemblage subsets. Significance of trends was determined by two-sided chi-square tests. Temporal trends of M) non-rarefied richness (χ²(1) = 3.39, p = 0.066) did not vary by past temperature regime, but N) evenness (χ²(1) = 7.917, p = 0.005) did vary by past temperature regime. In warm streams, non-rarefied richness nonsignificantly (Z = 1.804, p = 0.071) and evenness significantly (Z = 1.964, p = 0.049) increased. In intermediate streams, non-rarefied richness (Z = −0.424, p = 0.671) and evenness (Z = −0.395, p = 0.693) trends were not significantly different from zero. In cold streams, non-rarefied richness nonsignificantly (Z = −1.772, p = 0.077) and evenness significantly (Z = −2.556, p = 0.011) decreased. By comparing changes relative abundance, rarefied richness, non-rarefied richness, and evenness we can evaluate causes of changes in rarefied richness.