Use of interpolated climatic parameters to predict risk of blowfly-strike in Queensland sheep flocks

The risk of blowfly-strike was investigated in 160 sheep flocks in Queensland, Australia. The association between blowfly-strike--measured by the use of pesticides in response to flystrike reported by flock managers during the period 1994-1997--and interpolated rainfall and temperature data was examined using survival analysis and stepwise regression. The statistical model explaining the most variation in risk of blowfly-strike included average maximum temperature and its second-order polynomial 4 months preceding flystrike. The model was validated using reports of blowfly-strike from 87 flocks during the period 1997-1999 (collected using the same method as for model construction) and interpolated maximum temperature data. Although there was overlap of 95% confidence intervals derived for model coefficients estimated using both construction and validation data sets, false-negative and false-positive misclassification percentages were 33.3 and 53.3%, respectively. False-negative misclassification was greater for flocks located in southern Queensland, and false-positive misclassification was greater for northern Queensland flocks.