Generalized estimating equations to evaluate the dependence of lifetime in confined Africanized bees

Abstract

Research with honeybee needs experimentation in cages. In laboratory bioassays to assess the lifetime of Apis mellifera, each sampling unit contained a group of bees confined in a cage. Since current is a longitudinal analysis to assess the effect of time dependence on the survival of bees fed on different diets, generalized estimating equations applied to the logit model were used to assess the viability of incorporating correlation structures that related the survival of bees observed in each cage over time, so that the mortality of a given number of bees might influence the mortality of the others. Results showed that there was time dependence between the observations of bees inside the cages, and the working correlation matrix adjusted by longitudinal analysis was `Non-stationary-M-dependent´ in which the number of periods of dependence must be specified. Correlations are not the same for all observations of all periods. The solution with granulated sugar and the solution with granulated sugar enriched with lemon juice proved to have the lowest level of mortality during the assessed periods. Assay showed that bees tended to die during daytime.