Selecting a binomial regression model on the predation rate of Euseius concordis (Chant, 1959)

Abstract

The red spider Oligonychus ilicis (McGregor, 1917) (Tetranychidae) has proven to be a major cause of problems in coffee (Coffea spp.). Chemical control has been widely used to control this mite, however, the existence of the adverse effects of its application has generated concern regarding its use. An alternative to this method is the biological control of pests by natural predators, among them mites of the genus Euseius (Phytoseiidae) have been reported as efficient predators in the biological control of various species of phytophagous mites providing high rates of predation. Given the above, it’s of great importance to the accuracy of information obtained by a mathematical model, since they will condition decisions, and generalized linear models offer a wide diversity of models front predation rate. In this sense, the present study investigated the proportion of Euseius concordis (Chant, 1959) (Phytoseiidae) predatory mite attack on demand of prey O. ilicis, selecting the most appropriate binomial regression model to represent it, according to the AIC and BIC criteria, Deviance test and simulated envelope. The results indicate that the Logit model showed the best values AIC, BIC and Deviance test for suitability, therefore, the most suitable model to explain the relationship of predation/prey of the mite E. concordis in control of O. ilicis. The E. concordis performance in densities exceeding 4.3 mites/cm² is less than 50%. From the 6.4 density mites/cm² the predation rate of the mite E. concordis showed slight decrease and trend of constancy.