Estimating the probability of bird mortality from pesticide sprays on the basis of the field study record.

The outcome of avian field studies was examined to model the likelihood of mortality. The data were divided into clusters reflecting the type of pesticide application and bird guilds present on site. Logistic regression was used to model the probability of a bird kill. Four independent variables were tested for their explanatory power: a variable reflecting acute oral toxicity and application rate; a variable reflecting the relative oral to dermal toxicity of the pesticides; Henry's law constant; and a variable reflecting possible avoidance of contaminated food items, the hazard factor (HF). All variables except for HF significantly improved model prediction. The relative dermal to oral toxicity, especially, was shown to have a major influence on field outcome and clearly must be incorporated into future avian risk assessments. The probability of avian mortality could be calculated from a number of current pesticide applications and the conclusion was made that avian mortality occurs regularly and frequently in agricultural fields.