Prediction of macrobenthos distribution using the Maxent and simplified estimable environmental variables in the Seto Inland Sea, Japan

Abstract

Benthic invertebrates are an important biome in ecological processes and are useful indicators for assessing benthic environments. The maximum entropy model (Maxent) is a common species distribution modeling tool for predicting species distribution from appearance records and environmental predictors. In this study, to overcome the requirement for a large amount of benthic environmental data with sediment collection to apply the Maxent to the marine benthos, the use of simplified estimable environmental variables such as water depth and grain size was evaluated. The habitat suitability of 10 representative species of macrobenthos in the 1990s, 2000s, and 2010s were successfully estimated by the Maxent using seven physicochemical parameters of the sediment in the sea. The parameter with the highest contribution rate in Ampelisca naikaiensis was total organic carbon content (44%), while it was mud content for Leptochela gracilis (57%), water depth for Glycela sp. (30%) and Chaetozone sp. (32%), and gravel content for the other six species (29–63%). The mean values of habitat suitability in the whole sea increased for five species, whereas those of the remaining five showed a flat trend from the 1990s to 2010s. The AUC values were >0.70 even when only water depth and grain size were used for the Maxent as simplified estimable environmental variables, and the analysis was sufficiently accurate.