Efeito do ambiente e da posição longitudinal sobre a distribuição de peixes em uma microbacia da Amazônia Central

Abstract

Cross-scale ecological studies can complement traditional single-scale studies by elucidating important interactions between scales that affect the distribution of species. The distribution of fish species in streams is known to be independently affected by both the segment-scale changes in physical, chemical and biological characteristics associated with increasing width and depth along the water course (longitudinal gradient), and the finer mesohabitat-scale variation of stream characteristics (local environments). However, it is not yet clear whether and how the longitudinal gradient and local environment characteristics may interact when both effects are viewed together. Knowledge of possible interactions between site-occupancy/detectability along the longitudinal gradient and site-occupancy/detectability between local environments at a given level of the gradient could reveal important process that affect species distributions and sampling methods. We quantified possible interactions between mesohabitat and segment scale with a site-occupancy sampling design, applied over a microbasin in a non-floodable forest of the Central Amazon, Brazil. Our analysis explicitly accounts for the possibility of detection failure, estimating occupancy and detection probabilities in two types of local environments (main channel and adjacent temporary ponds) in segments of stream that span from first to 4th order. We modelled the whole fish assemblage and each 18 fish species as hierarchical levels of a multispecies model. We found evidence of interaction between longitudinal position and local environment occupancy for the assemblage and for eight species, with a downstream increase in occupancy in both environments. Furthermore, we found evidence of interaction between position and the environment in detectability, with a downstream decrease in detection in channel for the assemblage and for 15 species. The faster downstream increase in channel occupancy is assigned to major changes in hydrological characteristics, as water velocity and depth, which varies more along the gradient in channel than in ponds. Moreover, we observed for almost all species opposite relationships between site occupancy and detection probabilities. So, we argue that changes in species detection along ecological gradients, as longitudinal limitations of fish sampling caused by depth, can easily bias results based on species detection/nondetection data and