Author(s): Laura Plichard; Maxence Forcellini; Yann Le Coarer; Herve Capra; Sylvie Merigoux; Jean-Michel Olivier; Nicolas Lamouroux
Linked Author(s):
Keywords: No Keywords
Abstract: The Habitat concept is one of the main concepts in ecology. Describing and understanding the process of habitat selection by organisms is crucial to better quantify biological responses to natural and anthropogenic habitat alterations. Habitat selection models in rivers, developed at the scale of microhabitats around organisms, are frequently coupled with hydraulic models of stream reaches to predict potential changes in abundance due to modified flow regimes. However, many fish and macroinvertebrates habitat models are built without sufficient temporal and spatial replication, which limit their transferability from one river to another and consequently their predictive performance. Using current statistical developments, we developed habitat selection models for 31fish species and 206 macroinvertebrates taxa using very large data sets collected in from 10 rivers for fish (34sites, 145 surveys) and 11 rivers for invertebrates (22 sites, 90 surveys). We linked abundance to hydraulic key variables (e. g. water depth and current velocity for fish; shear stress for macroinvertebrates) using a negative binomial distribution hypothesis to account for the overdispersion of abundance counts, and using generalized linear mixed effects models to deal with the spatial and temporal variability of abundance and habitat selection between surveys. Finally, we tested the transferability of the models between rivers.
Year: 2018