Validation and inference of high‐resolution information (downscaling) of ENETwild abundance model for wild boar
Meta data
Disclaimer: The present document has been produced and adopted by the bodies identified above as authors. This task has been carried out exclusively by the authors in the context of a contract between the European Food Safety Authority and the authors, awarded following a tender procedure. The present document is published complying with the transparency principle to which the Authority is subject. It may not be considered as an output adopted by the Authority. The European Food Safety Authority reserves its rights, view and position as regards the issues addressed and the conclusions reached in the present document, without prejudice to the rights of the authors.
Abstract
The ENETWILD consortium provided in August 2019 a map at 10x10 km resolution for wild boar abundance based on hunting data. The availability of prediction maps at a spatial resolution comparable with the one of the home range of wild boar can be useful for further evaluation of risk of spread of African swine fever (ASF). Therefore, predictions of abundance on the basis of the wild boar home range are required. The downscaling procedure needs information on what resolution level is being used for predictions (hunting grounds, municipalities and NUTS3). This report presents the validation of previously produced hunting yield maps (10x10 km resolution) and new model projections downscaled at 2x2 km resolution. A new dataset based on hunting bag numbers was used as external data for validation. These data were arranged at two levels: at country level for the European scenario and at NUTS3 level for a scenario in Spain, where the data availability is higher than the rest of Europe in terms of quantity and quality. Very similar geographical patterns of wild boar abundance were obtained when the models were transferred to 2x2 km grid. The downscaled model predictions were aggregated at country and NUTS3 levels and compared against the external dataset. Our study confirmed that both 10x10 km and 2x2 km resolutions were able to detect spatial variation in wild boar hunting bags (high model performance) and to predict the numbers of wild boar hunted with relative precision (moderate model accuracy). Nevertheless, an overestimation of absolute number of hunted wild boar was observed using both resolutions. Reasons for this overestimation are discussed in this report. The linearity between predictions of hunting yield and external dataset was maintained, indicating that hunting yield predictions can be considered as a good proxy of wild boar abundance. Therefore, updated wild boar hunting yield data, collected at the finest spatial resolution as possible, is needed to correctly recalibrate our model at regional level, an in particular in eastern European countries.