Habitat selection ecology of the aquatic beetle community using explainable machine learning.

The aim of our work is to determine the importance of habitat features for the selection of the aquatic beetle community. Insects are represented by their general ecological traits such as body size, ecological element and trophic type, which are categorised into four body size ranges, four ecological groups and four trophic types. To determine the importance of habitat selection of the studied insects, we analysed the relationships between the above categories and the set of habitat features of the lake and its surroundings. Ensemble machine learning modelling (XGBoost-SHAP) revealed the mechanism of habitat feature selection in relation to the general ecological traits. We found strong interactions between the body size, ecological element and trophic type of beetles, suggesting that these general traits control the structure and functioning of the beetle community studied. The area of the lake and the features of beetle occurrence in the aquatic environment play an important but secondary role, and the importance of the characteristics of the lake's riparian zone was minimised. We found several categories of beetles as they select the number of the same habitat features. The study can provide valuable information for the practical conservation and management of lake ecosystems.