Validity of the model

The relevance of this discussion of the practical and theoretical implications of the responses of G. m. morsitans and G. pallidipes to vegetation depends on the extent to which the modelling of movement in relation to bushes and baits can be used to supplement results of field experiments. The model simplifies the real patterns of movement, uses crude maps and employs rough estimates for important parameters, i.e. f, k, interval duration and hop length. No variations are made to the rate or pattern of cell evacuation, to cover the possibility that tsetse take longer, faster or straighter hops in certain vegetation types, because there is no experimental evidence to show whether they do. Moreover, the model seems unable to predict the assessed distribution of tsetse between vegetation types, perhaps because of no allowance for effects of the uneven distribution of hosts and dry-season resting sites. There is no experimental indication of how to allow for these matters. The problems identify important matters for further research Despite the problems, the model does highlight some of the simpler ways in which bushes can contribute to tsetse distribution and invasion. For example, it seems that movement is impeded mainly at the junction of dense and lightly bushed habitats, not by resistance within the main expanse of the dense habitat. This suggests that invasion could be restricted most in areas where flies must cross many narrow bands of alternating vegetation types, i.e. where there are many junctions. Furthermore, the field data available as checks on the modelling of sampling suggest that the model is not far wrong in its indications for matters involved in this, i.e. persistence, catching-out, and the effects of odours, hedges and site manipulation for both species. The modelled rate of removal of flies by traps and targets fits well with the rates observed in the field. The model’s demonstration of the concentration of tsetse at the junction of paths matches Nash’s (1933) field evidence for G. m. morsitans, albeit that he studied paths through tall grass, not bushes. Since the model gives fair approximations for those sampling matters that can be checked, it seems reasonable to accept its indications for other sampling matters, provided the emphasis is on the principles exposed, rather than precise quantitative interpretations.