Validation of model generality: Can habitat models of Hyalesthes obsoletus, vector of bois noir, transferred in time, space and to a different pest insect, Scaphoideus titanus, vector of flavescence dorée?

Abstract: Besides emerging evidence of new Stolbur phytoplasma vectors, Hyalesthes obsoletus is still considered the main vector of bois noir in Baden region, SW-Germany. Recently, habitat models were developed to study the habitat requirements of the planthopper vector. Vectoring the yellow disease flavescence dorée, surprisingly little knowledge is available about the specific habitat requirements of the leafhopper Scaphoideus titanus.In this study, we address the question of how robust and applicable are the developed models, particularly to new invasive species, such as S. titanus, which is expected in Baden region in near future. To answer this question, we first transferred H. obsoletus models in time. Secondly, we fitted models with biomonitoring data from S. titanus; and thirdly, we developed new habitat models for S. titanus to compare habitats requirements between the two phytoplasma vectors.For model validation we used datasets from i) a biomonitoring dataset of H. obsoletus with presence/absence records of four different years in Baden region; and ii) data derived from a long-term biomonitoring network of S. titanus in the Libourne region, France. Depending on the dataset and the available predictor variables, the habitat model with highest AUC was used for prediction. All statistical models are based on logistic regressions.Besides variations in discriminatory ability, four H. obsoletus models were successfully transferred in time. We found that models including direct variables such as vegetation structure may be more appropriate to predict time-shifted incidences of Auchenorrhyncha species. Visual comparison of spatio-temporal transferred models to S. titanus occurrences showed discrepancies in model predictions. Moreover, new developed habitat models for S. titanus within the Libourne region including four bioclimatic variables achieved higher model performances in comparison to transferred models.Results demonstrate the practical use of previously developed habitat models of H. obsoletus for current and future pest management in the Baden region. We highlight that H. obsoletus models are of limited use to predict S. titanus occurrences in the Libourne region. We argue that ecological insights as well as risk maps of the leafhopper are best realized developing new habitat models based on ecologically relevant predictors, a study area covering an ampler range of environmental gradients and simultaneously accounting for previous insecticide sprayings.