Fig. 3: Effects of species richness on community stability and its underlying components.

In a the richness-community stability (CV_net⁻¹), relationships are sown for each 5-year window indicated by different colors (1 = 2003, 17 = 2019). b The change in the slope of the log–log relationship between richness and community stability (power exponent b of curves shown in a for each consecutive 5-year rolling window. The solid regression line was fit using the relationship slope~log(window). Similarly, c are the regression coefficients of richness on the five-year temporal mean and SD in community productivity and d on the population stability (CV_pop⁻¹) and asynchrony (async.) of the log–log relationships. These coefficients are relative effects of richness on community stability as b_mean  − b_SD and b_async + b_CVpop⁻¹ are the slope of the log–log relationship between richness and community stability (b_CVnet⁻¹) shown in b (see Methods). Black and dashed regression lines respectively highlight significant and non-significant trends along the rolling windows. Tests for significance are two-sided for a difference from 0.