Improving the performance of macroinvertebrate based multi-metric indices by incorporating functional traits and an index performance-driven approach

Human-driven multiple pressures impact freshwater ecosystems worldwide, reducing biodiversity, and impacting ecosystem functioning and services provided to human societies. Multi-metric indices (MMIs) are suitable tools for tracking the effects of anthropogenic pressures on freshwater ecosystems because they incorporate various biological metrics responding to multiple pressures at different levels of biological organization. However, the performance and applicability of MMIs depend on their metrics' selection and their calibration against natural environmental gradients. In this study, we aimed at unraveling i) how incorporating functional trait-based metrics affects the performance of MMIs, ii) how disentangling the natural environmental gradients from anthropogenic pressures effects affects the performance of MMIs, and iii) how the performance of MMIs developed using a metric performance-driven approach compares with MMIs developed using an index performance-driven approach. We carried out a field survey measuring abiotic and biotic variables at 53 sites in the Karun River basin (Iran) in 2018. For functional trait-based metrics, we used 15 macroinvertebrate traits and calculated community-weighted mean trait values and functional diversity indices. We used random forest modeling to account for the effect of natural environmental gradients on each metric. Based on our results, incorporating functional traits increased the MMI performance significantly and facilitated ecological interpretation of MMIs. In the sense of naturally occurring spatial variability in macroinvertebrate assemblages and its confounding effect with effects of anthropogenic pressure, both taxonomic and functional components of macroinvertebrate assemblages co-varied strongly with natural environmental gradients, and accounting for these covariations increased the performance of MMIs. Finally, we found that index performance-driven MMIs had a higher performance in terms of precision, bias, sensitivity, and responsiveness than metric performance-driven MMIs.