Validation of a geographic weighted regression analysis as a tool for area-wide integrated pest management programs for Ceratitis capitata Wiedemann (Diptera: Tephritidae) on Terceira Island, Azores

Extrapolation from quantitative sampling to the surrounding areas is an essential feature of many aspects of real world applications in pest management decision-making. However, these decisions are only as good as the accuracy of the methods that provided the information. The problem of estimating Mediterranean fruit fly population densities from trap grids is a specific case. The efficiency of three methods to estimate fruit flies trapped per day values for nonsampled areas in Terceira Island is evaluated, the inverse distance weighted, ordinary Kriging and the geographic weighted regression (GWR). Each method has its own specificities and merits. The results demonstrate that the GWR method is capable of estimating hotspots for the next season and can be used to identify ecological corridors over a non-sampled area. The high spatial heterogeneity and topographical conditions present on Terceira Island may explain why a more mathematically complex method is more reliable than simpler methods for use in possible future wide-area control program for medfly.