The use of discriminant analysis in predicting the distribution of bluetongue virus in Queensland, Australia.

Abstract

The climatic variables that were most useful in classifying the infection status of Queensland cattle herds with bluetongue virus were assessed using stepwise linear discriminant analysis. A discriminant function that included average annual rainfall and average daily maximum temperature was found to correctly classify 82.6% of uninfected herds and 72.4% of infected herds. Overall, the infection status of 74.1% of herds was correctly classified. The spatial distribution of infected herds was found to parallel that of the suspected vector, Culicoides brevitarsis. This evidence supports the role of this arthropod species as a vector of bluetongue viruses in Queensland. The effect of potential changes in temperature and rainfall (the so-called 'global warming' scenario) on the distribution of bluetongue virus infection of cattle herds in Queensland was then investigated. With an increase in both rainfall and temperature, the area of endemic bluetongue virus infection was predicted to extend a further 150 km in and in southern Queensland. The implications of this for sheep-raising in Queensland are discussed.