Environmental determinants of functional diversity of stream fish assemblages in agroecosystems

Background/Question/Methods

Some anthropogenic impacts, such as agriculture expansion, are related to the impairment of habitat integrity in streams. These environmental alterations can influence patterns of fish diversity not only in a taxonomic, but also in a functional perspective. We investigated, in river basins used for pasture and agriculture in Southeastern Brazil, the importance of environmental variables on the taxonomic (richness) and functional diversity of stream fish assemblages. Using a standardized electrofishing protocol, we collected fish in 77 randomly selected stream reaches. We determined functional traits for each species (with abundance higher than nine) according to habitat use (morphological attributes), size, and feeding biology (gut analysis). We used a Principal Coordinate Analysis (PCoA) to select environmental variables from an initial set of 28 variables. We described fish functional variability by plotting species in a two-dimensional ordination (PCoA), based on the functional distance among them. We determined the functional diversity by calculating functional richness (FRic), functional evenness (FEve), functional divergence (FDiv), functional dispersion (FDis), mean pairwise distance index (MPD), mean nearest taxon distance index (MNTD), net relatedness index (NRI), and nearest taxon index (NTI). We tested the importance of environmental variables on the functional diversity indexes with multiple regressions.

Results/Conclusions

We evaluated 37 fish species and retained seven environmental variables for inferential analysis. The PCoA first axis differentiated species according to their swimming capacity in fast waters and occupancy of the vertical strata. The second axis represented the species size. Species richness, FRic, and FDis were positively associated to the distance to the streams’ springs (t_richness=2.511, P=0.014; t_FRic=2.670, P=0.009; t_FDis=2.021, P=0.047), in accordance to the longitudinal gradient. Species richness was positively related to the amount of grass covering marginal banks (t_FRic=2.375, P=0.020), in contrast to MPD, MNTD, and NRI (t_MPD=-3.419, P=0.001; t_MNTD=-2.650, P=0.010; t_NRI=2.674, P=0.009). This grass comes from surrounding pasture and represents a degraded condition. Despite the fact of a higher taxonomic diversity in grassy streams, assemblages tended to be functional clustered (i.e., coexisting species tend to be functionally redundant). MNTD was also related to sand proportion in the stream bottom (t=2.229, P=0.029), suggesting absence of species functionally redundant. These results emphasize the importance of local variables on the functional structure of stream fish assemblages. Furthermore, the occurrence of clustered assemblages in degraded streams suggests that the physical habitat degradation acts as environmental filter, and contributes to the functional homogenization of fish fauna in agricultural basins.