Macrobenthic species response surfaces along estuarine gradients: prediction by logistic regression

This study aims at contributing to the development of statistical models to predict macrobenthic species response to environmental conditions in estuarine ecosystems. Ecological response surfaces are derived for 10 estuarine macrobenthic species. Logistic regression is applied on a large data set, predicting the probability of occurrence of macrobenthic species in the Schelde estuary as a response to the predictor variables salinity, depth, current velocity and sediment characteristics. Single logistic regression provides good descriptions of the occurrence along 1 environmental variable. The response surfaces obtained by multiple logistic regression provide estimates of the probability of species occurrence across the spatial extent of the Schelde estuary with a relatively high degree of success. Results from subsampling 50% of the original data 10 times indicate that final models were stable. A visual geographical comparison is presented between the mapped probability surfaces and the species occurrence maps. We conclude that where patterns of distribution are strongly and directly coupled to physicochemical processes, as is the case at the estuarine macro- and meso-scale, our modelling approach was capable of predicting macrobenthic species distributions with a relatively high degree of success, although processes controlling estuarine macrobenthic distribution cannot be determined using this method. However, the models and predictions could be used for evaluation of the effects of different management schemes within the Schelde estuary. [KEYWORDS: Benthic macrofauna · Estuarine, spatial gradients · Logistic regression · Response surfaces · Schelde estuary]