Predicting bird species occurrences for conservation planning in Uganda

In Uganda, lack of adequate data on species' distributions poses a problem in laying strategies for conservation planning. Some of the data that exist are old and are therefore of uncertain value, given the pace of environmen tal changes. To overcome these problems, a logical prediction model was bui lt based on species-habitat relationships to help predict the occurrence of bird species in Uganda. This computer model makes use of field observation s, from which species-habitat relationships are built; these are based upon biogeographical zones, an 80-category vegetation map, and human population densities. The model is then run to determine 'suitable habitats' for spec ies of interest. The model has been tested for accuracy for the existing na tional parks, and high degrees of accuracy reaching around 90% have been ob tained for those parks with many observations (unpublished data). In future , it is planned to include rainfall as an alternative to biogeographical zo nes. This paper examines the methods used in this research, gives a descrip tion of the model, the results that have been obtained; and assesses the us efulness of the model in conservation planning.