Extended Data Fig. 10: Evaluation of the effects of land use on biodiversity trends.

A) Trends in abundance (catch per unit effort) varied with land use. Abundance did not change in agriculture, grassland/shrub, and urban streams, while abundance increased in forest/wetland streams. For all other biodiversity metrics, land use influenced the intercept but not the slope of temporal trends. B) Rarefied richness was lowest in urban streams compared to all others. C) Site uniqueness is greater in forest/wetland and grassland/shrub streams compared to agricultural streams, and other land use types are not different from each other. D) Functional diversity is greater in agricultural streams compared to forest/wetland and grassland/shrub streams, and other land use types are not different from each other. All pairwise comparisons can be found in Table S7. In A), lines represent model predicted trend with 95% CI bands. Points and errors in B), C), and D) are model-estimates and 95% CIs, holding covariates at their proportional values. For A), n = 4240; for B), n = 3784; for C), n = 4491; and for D), n = 4482. Sample sized varied by endpoint, because of differences in available sampling information, minimum sample size, and number of unique species.