Multispectral characterization, prediction and mapping of Thaumastocoris peregrinus (Hemiptera: Thamascoridae) attack in Eucalyptus plantations using remote sensing

Abstract

Thaumastocoris peregrinus (Hemiptera: Thaumastocoridae), the bronze bug, is among the many species of insect pests affecting commercial Eucalyptus forests in Brazil. T. peregrinus reduces the photosynthetic ability of trees and in some cases can lead to the complete death of trees. This research aims to assess the potential of freely available medium-resolution Landsat 8 imagery in predicting and mapping attacks caused by the bronze bug in Eucalyptus plantations in Brazil using partial least squares discriminant analysis (PLS-DA). The PLS-DA model selected three principal components with a cross-validated error rate of 0.248 for the prediction and mapping of attacked T. peregrinus stands. Important bands were selected from the PLS-DA model using variable importance in the projection (VIP). The VIP bands predicted healthy and attacked stands with an accuracy of 76.7 percent on an independent validation dataset. This study demonstrates the potential of freely available medium-resolution Landsat OLI imagery as an alternative to often expensive high-resolution datasets to successfully characterize and map T. peregrinus attacks in plantation forests in Brazil.