Habitat models for a riparian carabid beetle: their validity and applicability in the evaluation of river bank management

In order to assess the management success of river rehabilitation measurements it is necessary to have representative target species and objective statistical methods. In this study we, tested the validity of habitat suitability models for the riparian carabid beetle Bembidion velox in the evaluation of river bank management along the River Elbe, Germany. On the basis of seven independent data sets from different sites and years we have proven the robustness of logistic regression models with respect to their explanatory and predictive power and their applicability in the field. All models had robust explanatory power and described a strong association of B. velox with semi-terrestrial sandy open soil habitats. Transfers of model results for adult beetles to their larvae and vice versa were highly significant with sand content and stem distance as the main habitat factors for both life stages. To broaden the local explanatory power towards general predictions we performed model cross-validation in space and time. Spatial transfers produced models with excellent discrimination properties, measured by Area Under Curve (AUC) values of Receiver Operating Characteristics (ROC) plots, independent of sampling designs and trapping methodology. However, the applicability of habitat models for B. velox is defined by the validity period, as the availability of suitable habitats for this species is highly temporally variable and dependent on water level. Model transfers between species also demonstrated that the chosen target species is representative for carabids with similar distribution patterns, as the single species model had high predictive power for the occurrence of a multi-species carabid group.