Satellite imagery and the prediction of tsetse distributions in East Africa.
This paper briefly reviews the debate over whether the aim of tsetse suppression is eradication or control, concluding that control is the only viable solution for many areas of Africa. This approach requires a fuller understanding of the natural dynamics of tsetse populations and the environmental limits to their distribution than does the alternative approach aimed at eradication. Various types of remotely sensed satellite data are correlated with the meteorological records that have in the past been used to interpret tsetse distribution and abundance patterns, and have the advantage over such records of complete spatial coverage at acceptable resolution for country, regional or continental studies. The paper describes both the temporal Fourier processing of such data from the NOAA and Meteosat series of meteorological satellites and the application of these processed data to describing the distribution of 5 species of tsetse in East Africa, Glossina morsitans s.l., G. pallidipes, G. austeni, G. longipennis and G. fuscipes fuscipes. Observed distributions are described with accuracies of between 85 and 97% (G. morsitans and G. austeni, respectively) when, within non-linear discriminant analysis, variables are selected on the basis of maximizing the minimum multivariate distance between alternative categories of presence and absence training set data. Inaccuracies in the predicted maps arise for several reasons: the distribution maps used to define tsetse and non-tsetse areas are now rather out of date; some of the areas for which predictions are made are very different from any of the training set areas, although the analysis may be forced to assign such areas to one or other category; and, finally, non-tsetse areas may indeed be climatically suitable for flies, but not inhabited by them for a variety of reasons. Examples are given of each for the case of G. longipennis, a species unrecorded from relatively large areas of Kenya which appear to be suitable for it. The conclusion highlights the need to test many of the ideas presented in this paper using contemporary tsetse and satellite data, to establish a real time monitoring system for the changes that lie ahead.
Rogers, D.J. Satellite imagery and the prediction of tsetse distributions in East Africa. In: Towards livestock disease diagnosis and control in the 21st century: proceedings of an International Symposium on Diagnosis and Control of Livestock Diseases Using Nuclear and Related Techniques, Vienna, Austria, 7-11 April 1997. (1998) 397-420. ISBN 92-0-102498-3