Original ArticleTrait-based metrics as bioindicators: Responses of stream fish assemblages to a gradient of environmental degradation

- The responses of indicators of fish diversity to an environmental gradient were assessed.
- Functional trait indicators have greater predictive power compared to taxonomic indicators.
- The performance of functional diversity indicators vary among ecoregions.
- Indicators of functional composition are the most consistent across ecoregions.

The development of biomonitoring tools is increasingly appealing in light of the increasing degradation of aquatic ecosystems. In this context, we investigated the responses of stream fish communities to the gradient of environmental degradation in different basins using a variety of indices based on functional diversity, functional composition, and taxonomic diversity. We used datasets from three Brazilian ecoregions. In order to describe the gradient of environmental degradation we scored streams based on local and landscape variables. The functional structure of the assemblages was described in relation to seven functional traits related to habitat use, swimming capacity, and maneuverability. We described assemblages in each ecoregion separately in relation to 20 potential indicators, grouped in four families: indices of functional diversity based on presence/absence (FDp/a); indices of functional diversity weighted by abundance (FDabund); indices of functional composition (mT); and, indices based on taxonomic diversity (TDiv). The relationship between the indicators and the gradient of environmental degradation were evaluated using linear regression. We found a significant interaction effect (ANOVA, p = 0.006) between group of indices and ecoregions for the performance of indicators, suggesting a context-dependent response. The indices, on an individual basis, had variable performance and consistency among ecoregions. Four mT indicators demonstrated the highest average performance and consistency. Taxonomic diversity indicators consistently had the lowest average performance, while FDp/a and FDabund indicators had low average performance and variable inter-regional consistency. The differential inter-regional performance of indicators was due to the differences in the lengths of the gradients of environmental degradation. Our results indicate that functional traits have greater predictive power compared to taxonomic indicators for fish responses to a gradient of environmental degradation. Although indicators of functional composition are the most promising, we emphasize that caution is needed when generalizing functional diversity indicators across ecoregions because most of them are context-dependent.