Predicting Sika deer impacts on animal and plant communities; a comparison of ecological models
Start date: 14 September 2007
The Purbeck area (Dorset, UK) has one of the largest groups of feral Sika deer (Cervus nippon Temmink 1838) in England and Sika are the most important wild vertebrate grazer across a range of habitats in this area. Purbeck also contains important areas of lowland heath, one of the world’s most threatened ecosystems. Sika are abundant on the heaths and can have major positive and negative ecological impacts. Consequently it is vital to understand what factors affect the distribution and impact of Sika deer across Purbeck so that appropriate management can be implemented.
Ecological modelling has become a crucial ecological methodology among the research community and different models have been developed to predict population or individual responses to changes in the environment. Predicting responses of Sika in the Purbeck area to changes in landscape, management regime (i.e. culling or fencing) or changes in patterns of human disturbance is crucial in order to predict future impacts on key habitats.
In this study we compare two different approaches to predicting habitat use by Sika: a new approach using individual-based models vs. more “traditional” spatial/habitat distribution models.
Individual-based models assume that the habitat use of an animal population results from the optimal foraging decisions of all individuals within the population: at each point in time individuals occupy the patch in which their chance of survival and reproduction (i.e. fitness) is maximised. These models have been designed to predict how individuals within a population will alter their behaviour in response to environmental change (e.g. different management regimes), and to predict the population consequences of these optimal decisions from the location and fates of individuals. We use a model named MORPH (Stillman, In press), which is fourth in a succession of individual-based models that have been used successfully on birds but has not being validated on mammals.
There are a wide variety of spatial or habitat distribution models, many involving geographic information systems (GIS) and remote-sensing (Guisan and Zimmerman, 2000, Elith and Burgman, 2003). These models are capable of providing probabilities of presence or abundance of Sika on different patches depending on habitat characteristics (i.e. food resources, vegetation structure).
We analyse the suitability of these two different approaches with regard to the models’ assumptions, available data series, their generality, reality and precision. Also we suggest possible complementary uses of spatial and individual-based models in order to formulate predictions of ecological impacts.