Predictive dispersal models of Hypothenemus hampei (Coleoptera: Curculionidae) in coffee plantations with different uses of adjacent soil in Turrialba, Costa Rica

Abstract

Statistical models were developed that describe the dispersion of the borer (Hypothenemus hampei) in land uses adjacent to coffee plantations (grass, cane and forest). Sampling was conducted with a methodology of transects in farms located in the canton of Turrialba, within the Central Volcanic Corridor Talamanca (CBVCT) in Costa Rica. In each transect BrocapTM traps were placed every 10 m, capturing the berry borer for six months. With the purpose of studying the distance at which the insect is dispersed in different adjacent land uses, three statistical modeling approaches were used 1) non-linear models, 2) general linear mixed models and 3) generalized linear mixed models. These different modeling approaches within each model type were evaluated with the information criteria AIC and BIC. To compare these different modeling approaches, the Pearson correlation coefficient between the predicted values for each model and the observed values was used. The results of this work suggest that the best fit for the description of the insect dispersion was according to the generalized linear mixed model. Based on this model, the approximate maximum distance of dispersal occurred in the adjacent sugar cane land use at 30 m and the minimum distance was in the adjacent forest land use at 10 m, suggesting that the cane cultivation serves as a transit for the coffee borer to move between the coffee plantations, and the soil use that best controls the movement of the coffee borer is the forest.