The influence of spatial processes on fish community structure: using a metacommunity framework for freshwater bioassessment.

The use of biological indicators in a bioassessment approach is important for inferences of anthropogenic stress in routine monitoring programs. One of the primary assumptions of bioassessment is that biological indicators observed at specific sampling sites will allow for inferences of local environmental quality; however, this assumption requires a reliable understanding of dispersal processes, which is particularly relevant in river systems due to their dendritic network. Inter-stream dispersal between different points of the river network could bias bioassessment, especially for highly mobile organisms like fish. Here, we examine standard biological metrics used in routine biomonitoring to determine how spatial variables, including dispersal, influence inferences between fish populations and environmental gradients. Using redundancy analysis (RDA) and variation partitioning, we tested the relative influence of both environmental and spatial variables on fish community structure and related community metrics. Fish were collected from 99 sampling sites distributed across 44 rivers and streams of the Great Morava River Basin, Serbia. Electroconductivity, the percentage of agricultural areas, dissolved oxygen, ammonia, and nitrate-nitrogen were found to be significant environmental variables, while ten spatial predictors from broad- to small-scales were found to influence fish community structure and community metrics. Our results suggest that contemporary dispersal among streams influences fish community composition, but that trait-based metrics are less sensitive than basic measures of diversity to spatial processes. Our results highlight the need for spatially independent sampling, as well as the need to consider dispersal-based processes in routine biomonitoring.