Application of simulation models in the epidemiology of pests and diseases: an introductory review.

Abstract

Watt (196la, b, 1963, 1964) was one of the first people to realise the full potential of agricultural pest modelling. Watt's main interest was to use models to evaluate different control strategies, and he was able to show theoretically that pest populations could reach higher levels after the application of a pesticide than in its absence (Watt, 196la). Watt (1961b) was also concerned with modelIing field populations and later extended his ideas to resource management.

Conway (1973, 1977) criticised many of the existing mathematical pest models for being too general (producing only obvious or trivial results), for ignoring the economic aspects of control, for failing to initiate the interaction between modelling and experimentation and for their irrelevance to pest management. systems. The first criticism is now less important as systems teams start to tackle specific problems with specific objectives. The analytical approach however is still interested in general models e.g. to account for the searching strategies of predators and parasitoids. The second point is a reflection of two considerations; most modellers are trained in one subject, in this case biology, and know little of economics and the economic aspects of crop protection have not often been calculated. These considerations are  changing as entomologists and phytopathologists realise the importance of damage levels, economic thresholds and action levels. The third criticism is possibly unjustified. Model builders quickly realise what experimental work needs to be done to provide the missing data for their models. Very often however with an analytical approach the importance directly unmeasurable variables and parameters are stressed, e.g. the mutual interference constant of predators.  These values are usually calculated from graphs and are very often an incorporation of many biological processes into one variable or parameter.  Thus they are purely descriptive with little explanatory value. Analytical models are, because of their abstract character, usually difficult to validate in the real world. Very often the author gives examples where his model fits observations in the field but this does not answer the question as to whether his theoretical considerations are valid. This makes the value of these analytical models for interpretation of a specific field situation limited, and conclusions on the mode of action of a system speculative and dangerous. Finally, models are now being employed in management systems.