Map Generation in High-Value Horticultural Integrated Pest Management: Appropriate Interpolation Methods for Site-Specific Pest Management of Colorado Potato Beetle (Coleoptera: Chrysomelidae)

Abstract

We describe a sampling program using sample units that might be feasible for mapping Colorado potato beetle, Leptinotarsa decemlineata (Say), densities for site-specific integrated pest management (IPM), an approach that varies the spatial placement of interventions in relation to the variation in pest density within a field. The influence of 5 inter-polation methods (kriging, 4 inverse distance weighted functions. and thin plate spline with tension) on 3 estimators of the error associated with each interpolation method (the overall error sum of squares, the proportion of total variation explained by the interpolated map (R2), and the categorization of cells relative to a threshold) were computed. When the threshold contour was overlaid onto surfaces, different interpolators suggested similar areas for treatment. The error sum of squares from kriging was generally smaller or equal to that achieved withother interpolators. The error sum of squares from inverse distance interpolators generally decreased with increasing weight in the exponent. With the higher exponents, the error sum of squares from inverse distance interpolators was as low as kriging and better than spline with tension. The R2 increased with mean density for all interpolators. At threshold densities, R2 values were ≈0.90 for adults and large larvae, and ≈0.75 for egg masses. All interpolators classified cells with respect to threshold density for all life stages with good to excellent accuracy (>85%). This research shows that kriging or simpler interpolators could be used for implementation of site-specific IPM.