Effects of land management upon species population dynamics :
[Thesis]
Parry, Hazel Ruth
a spatially explicit, individual-based model
Morgan, Derek
University of Leeds
2006
Ph.D.
University of Leeds
2006
Individual-based approaches in ecology provide a new approach to spatially explicit modelling. They are paralleled by the emergence of agent-based modelling in the field of artificial intelligence (AI) that is manifest in object-based approaches in a number of geographical disciplines, from hydrology to sociology. An individual-based approach to the simulation of organisms in a spatial context allows for a greater understanding of how individual-level behaviour and interactions result in population-level phenomena at the landscape-scale. Such models are inherently flexible and adaptable to other species or systems, and the model can be parameterised from biological behavioural information widely available in the literature. This research constructs, analyses and experiments with an individual-based model of aphid (Rhopalosiphum padi) population dynamics in agricultural landscapes during the autumn and winter. The model combines deterministic equations governing the development of the aphids with stochastic, behavioural rules. Several stages of model assessment validate the model: assessment at the conceptual, developmental and operational stages. The need for a solution for the model to cope with large population sizes led to experimentation with both mathematical and computational solutions to this problem. It was found that parallel computing to distribute the simulation across a 30-node Beowulf cluster was the most effective at increasing model efficiency whilst preserving model behaviour. Key scenarios are presented, that show the power of this approach in predicting potential impacts of agricultural landscape change, including the effects of crop management, marginal habitat configuration and pesticide regime. This research clearly demonstrates the potential of spatially explicit individual-based modelling to predict scenarios that may advise policy decision-makers as a landscape management tool.