RESULTS
We found evidence of a decline in the species richness of the whole-community over time (Fig. 2a and Table S2). Likewise, top-carnivore, mesocarnivore, and detritivore species richness decreased over time (Fig. 2a and Table S2). We also found temporal declining trends in the abundance of whole-community and all trophic guilds (Fig. 2b and Table S2). Similarly, temporal declining trends in the energy flux were observed in whole-community and single trophic guilds (Fig. 2c and Table S2). Particularly, the energy flux for whole-community, top-carnivores, mesocarnivores, omnivores, and detritivores was reduced by at last 71%, 72%, 67%, 70%, and 78% from the first to the last year, respectively (Fig. 2c). The proportion of energy flux between trophic compartments has also changed systematically over time, becoming highly concentrated in the omnivore and detritivore compartments, with the top- and meso-carnivore compartments losing energy flux (Fig. 2d and Table S2).
At all three river sites, we found positive associations between rarefied species richness and energy flux for the whole community and for single trophic compartments. Increased whole-community species richness was associated with greater energy flux (Fig. 3a and Table S3) (linear mixed-effects model; log(whole-community energy flux): effect size for log(whole-community richness) 2.60 ± 0.47 (mean ± s.e.m.)). Increased top-carnivore species richness was related to greater top-carnivore energy flux (Fig. 3b and Table S3); (the effect size for log(top-carnivore richness) was 0.50 ± 0.09 (mean ± s.e.m.)). Increased mesocarnivore species richness was related to increased mesocarnivore energy flux (Fig. 3c and Table S3); (the effect size for log(mesocarnivore richness) was 0.26 ± 0.05 (mean ± s.e.m.)). Increased omnivore species richness was strongly related to increased omnivore energy flux (Fig. 3d and Table S3); (the effect size for log(omnivore richness) was 1.87 ± 0.50 (mean ± s.e.m.)). Lastly, increased detritivore species richness was also strongly related to increased detritivore energy flux (Fig. 3e and Table S3); (the effect size for log(detritivore richness) was 1.13 ± 0.32 (mean ± s.e.m.)).
Structural equation modeling (SEM) revealed direct and species richness-mediated indirect negative effects of human footprint on energy flux (Fig. 4 and Table S4). The negative effects of human footprint on species richness and energy flux were maintained after accounting for key drivers of diversity and ecosystem functioning, such as precipitation, N:P ratio, and river properties (i.e., water discharge and turbidity). Specifically, human footprint indirectly decreased top-carnivore energy flux by decreasing top-carnivore species richness (Fig. 4a-d; –0.32). Similarly, the human footprint indirectly decreased mesocarnivore energy flux by decreasing mesocarnivore species richness (Fig. 4d-f; –0.27). Furthermore, the human footprint reduced the energy flux of omnivores (β = –0.55) and detritivores (β = –0.14) only directly (no species richness-mediated indirect effects; Fig. 4). There was also a strong positive effect of time on human footprint, which indirectly decreased both diversity and energy flux (Fig. 4). There were direct and diversity-mediated indirect positive effects of precipitation on the energy flux of top-carnivores and detritivores (Fig. 4b,k).