Bivariate Stochastic Modeling of Functional Response With Natural Mortality

A correction due to Abbott ( 1925 ) is the standard method of dealing with control mortality in insect bioassay to estimate the mortality of an insect conditional on control mortality not having occurred. In this article a bivariate stochastic process for overall mortality is developed in which natural mortality and predation are jointly modeled to take account of the competing-risks associated with prey loss. The total mortality estimate from this model is essentially identical with that from more classical modeling. However, when predation loss is estimated in the absence of control mortality the results are somewhat different, with the estimate from the bivariate model being lower than that from using Abbott’s formula in conjunction with the classical model. It is argued that overdispersion in observed mortality data corresponds to correlated outcomes (death or survival) for the prey initially present, while Abbott’s correction relies implicitly on independence.