Spatial sensitivity of a generic population model, using wild boar (Sus scrofa) as a test case

To develop a robust, generic approach to modelling uncertain processes that can be used in real landscapes, we constructed two spatial models. Both involved a sub-population model running simultaneously in parcels across a coverage, with movement into randomly chosen neighbours. An exploratory model investigated the dynamics of the sub-population process in a homogeneous (raster) landscape and explored how density dependence and movement interacted with spatial scale to affect model output. The second model applied the same sub-population dynamic across a spatially irregular, heterogeneous landscape based on UK habitat data. Wild boar (Sus scrofa) was chosen as a model species and the spread and abundance predicted by the applied model were compared with the limited field data suitable for this species in the UK.During the analysis we found a series of thresholds for sub-population size, which dictated the range of scales at which the model should be applied. If the thresholds were ignored and sub-populations were modelled at too small a scale, risk of extinction and spatial spread became exaggerated; at too large a scale, problems of spatial representation and low movement were observed. Driven by uncertain vital rates, the population model showed a range of behaviours at different scales, which were primarily explained by density dependent movement. When we applied our model to wild boar in the UK to simulate two historial releases, we achieved encouraging similarity to field observations. We are confident that we could have predicted the successful establishment of wild boar in the UK, and our approach, when refined, could be used to model future growth and spread.We found that spatial population models intended for use in heterogeneous landscapes need to be explored in the absence of confounding factors such as varying carrying capacity, in order to be certain that artefacts due to model structure are not present. We recommend the sensitivity of spatial population models to the interaction of spatial representation, movement and population process should be demonstrated at a variety of scales. We believe that the key to a successful real-world model is coherence between the numerical scale dictated by the sub-population process, the spatial scale of its representation and model stability.