RESUMO
Environmental heterogeneity is one of the most influential factors that create compositional variation among local communities. Greater compositional variation is expected when an environmental gradient encompasses the most severe conditions where species sorting is more likely to operate. However, evidence for stronger species sorting at severer environment has typically been obtained for less mobile organisms and tests are scarce for those with higher dispersal ability that allows individuals to sensitively respond to environmental stress. Here, with the dynamics of fish communities in a Japanese bay revealed by environmental DNA metabarcoding analyses as a model case, we tested the hypothesis that larger environmental heterogeneity caused by severe seasonal hypoxia (lower concentration of oxygen in bottom waters in summer) leads to larger variation of species composition among communities. During summer, fish species richness was lower in the bottom layer, suggesting the severity of the hypoxic bottom water. In contrast to the prediction, we found that although the environmental parameters of bottom and surface water was clearly distinct in summer, fish species composition was more similar between the two layers. Our null model analysis suggested that the higher compositional similarity during hypoxia season was not a result of the sampling effect reflecting differences in the alpha or gamma diversity. Furthermore, a shift in the species occurrence from bottom to surface layers was observed during hypoxia season, which was consistent across species, suggesting that the severe condition in the bottom adversely affected fish species irrespective of their identity. These results suggest that larger environmental heterogeneity does not necessarily lead to higher compositional variation once the environmental gradient encompasses extremely severe conditions. This is most likely because individual organisms actively avoided the severity quasi-neutrally, which induced mass effect-like dispersal and lead to the mixing of species composition across habitats. By showing counter evidence against the prevailing view, we provide novel insights into how species sorting by environment acts in heterogeneous and severe conditions.