Modelling habitat preference,abundance and species richness of alien macrocrustaceans in surface waters in Flanders (Belgium) using decision trees

The introduction and the spread of alien invasive species are a worldwide phenomenon causing global ecological and economic damages. Among the invaders, alien macrocrustaceans are known to be very successful invertebrates that colonise new habitats rapidly. Data from different fresh and brackish waters gathered by the Flemish Environment Agency (VMM) were used to build data-driven models predicting habitat preference, abundance and species richness of alien macro-Crustacea present in surface waters in Flanders. Different techniques such as regression and classification trees in combination with several optimisation methods (e.g. pruning) were used to construct the models. The performance of the models was moderate, because a balance between performance, ecological relevance and complexity was strived for. When using a three-fold cross validation it was found that the variation between the folds was limited, which is an indication of the robustness and the good reliability of the constructed models. Based on a sensitivity analysis the importance of conductivity, Kjeldahl nitrogen and shipping were stressed as well as graphically illustrated. Alien macrocrustaceans were predicted as present under brackish water conditions as well as in fresh waters with intensive ship traffic and low levels of organic pollution. The alien species richness was higher in rivers with intensive ship traffic and increased with increasing conductivity. Especially in brackish waters, alien macrocrustaceans reached high abundances. In fresh water, the abundance of alien species was generally lower. An integrated model that combined our habitat suitability model with a water quality model was used to predict the future distribution of alien macrocrustaceans. The predictions indicated that the prevalence and the species richness of alien macrocrustaceans are likely to increase with improving chemical water quality, whereas their abundance will probably decrease slightly. From our analysis, it is clear that models are a useful tool and that decision makers should focus on vulnerable areas such as brackish water areas and areas with intensive ship traffic in order to prevent the further introduction and spread of alien species.