| Summary: | Computer-based simulation models of changes occurring within crop population when subjected to agents of phenotypic change, have been developed for use on commonly available personal computer equipment. As an underlying developmental principle, the models have been designed as general-case, mechanistic, stochastic models, in contrast to the predominantly empirically-derived, system-specific, deterministic (predictive) models currently available. A modelling methodology has evolved, to develop portable simulation models, written in high-level, general purpose code, allowing for use, modification and continued development by biologists with little requirement for computer programming expertise. The initial subject of these modelling activities was the simulation of the effects of selection and other agents of genetic change in crop populations, resulting in the computer model, PSELECT. Output from PSELECT, specifically phenotypic and genotypic response to phenotypic truncation selection, conformed to expectation, as defined by results from established analogue modelling work. Validation of the model by comparison of output with the results from an experimental-scale plant breeding exercise was less conclusive, and, owing to the fact that the genetic basis of the phenotypic characters used in the selection programme was insufficiently defined, the validation exercise provided only broad qualitative agreement with the model output. By virtue of the predominantly subjective nature of plant breeding programmes, the development of PSELECT resulted in a model of theoretical interest, but with little current practical application. Modelling techniques from the development of the PSELECT model were applied to the simulation of plant disease epidemics, where the modelled system is well characterised, and simulation modelling is an area of active research. The model SATSUMA, simulating the spatial and temporal development of diseases within crop populations, was developed. The model generates output which conforms to current epidemiological theory, and is compatible with contemporary methods of temporal and spatial analysis of crop disease epidemics. Temporal disease progress in the simulations was accurately described by variations of a generalised logistic model. Analysis of the spatial pattern of simulated epidemics by frequency distribution fitting or distance class methods was found to give good qualitative agreement with observed biological systems.
|
|---|
