Date of this Version
Metacommunity studies have demonstrated that local macroinvertebrate communities are structured not only by local environmental conditions but also by spatial processes. Effective bioassessment tools should account for spatial processes while doing so with the least amount of cost. In this study, we applied variance partition techniques based on redundancy analysis to assess the performance of three sets of benthic invertebrate metrics in detecting agricultural land-use effects in a SE Brazil rainforest watershed. Macroinvertebrate data were analyzed separately regarding their taxonomic, functional structure and bioindicator metrics developed for the study region. We stipulated that groups of metrics most sensitive to land-use effects should have the highest amount of variance explained by the joint effects of land use and environmental variation, independently of spatial structuring. Statistical analyses were repeated removing rare taxa in order to assess the effects of their inclusion in the responsiveness of each group of metrics. Traditional bioindicator metrics were more responsive to environmental variation associated with agriculture than taxa abundances and functional attributes. Furthermore, a few common taxa drove a high proportion of the variation observed in invertebrate communities, regardless of how invertebrate data were organized. Similar analytic approaches have the potential to be useful in curtailing sorting and identification efforts when developing macroinvertebrate-based biomonitoring protocols, especially in areas where information regarding the taxonomy of benthic communities is still poorly described.