Revisiting wild boar spatial models based on hunting yields to assess their predictive performance on interpolation and extrapolation areas

While reliable estimates of species abundance distribution are required for wildlife management and are greatly needed at broad spatial scales, such information is scarce. In this context, the usefulness of spatial modelling as a tool for predicting game species relative abundance and distribution from hunting yield data was studied. Hunting yield data is affected by several factors related to species management, hunting regulations, and hunting efficacy and some doubts have been raised about the use or reliability of this data for large-scale modelling. Some years ago, Acevedo et al. (2014) calibrated five spatially explicit models (one per bioregion) by using hunting yield data for wild boar Sus scrofa (from hunting seasons 2006 to 2009) for approximately 60% of mainland Spain. After internal validation, the models were extrapolated to produce predictions of species relative abundance for the whole mainland country. Here, we reviewed these previous models to evaluate their predictive performance on new data (from hunting seasons 2014 to 2018) in areas where the models had been calibrated (interpolation areas) and also when projected into new ones (extrapolation areas). Our results showed that the previous models were able to forecast current general patterns of wild boar relative abundance with population growth rates equivalent to those reported by other authors, although differences between bioregions were observed. Performance on interpolation areas was higher than that obtained on extrapolation areas. Accuracy of model predictions decreased when fine resolution assessment at hunting ground level was carried out. Our results suggest that spatial models calibrated on hunting yields could be a good option to predict general wild boar relative abundance distribution patterns, although critical assessment is needed, since models can fail when they are extrapolated to areas for which no information is available and at fine scale resolution. These results represent a step forward in the use of hunting yields for describing ranges of species relative abundance at large spatial scales.