Detecting the severity of woodwasp, Sirex noctilio, infestation in a pine plantation in KwaZulu-Natal, South Africa, using texture measures calculated from high spatial resolution imagery

This study was undertaken to determine whether image texture derived from high spatial resolution imagery could successfully predict Sirex noctilio infestation levels in a pine plantation forest in the eastern part of South Africa. The woodwasp, S. noctilio, damages trees by releasing toxic mucus into the wood during oviposition. A 50 × 50 m grid was generated over the study area and 65 cells were randomly selected. Within each cell, a 10 m circular plot was created and the level (%) of S. noctilio infestation was visually assessed. Using high spatial resolution imagery (0.5 m), 13 texture measures were calculated with various window sizes. Correlation tests were then performed to determine the relationship between image texture and S. noctilio infestation levels. Individual correlations were average in strength. Consequently, a stepwise multiple linear regression analysis was undertaken to determine whether a selection of texture images could result in a more accurate prediction of S. noctilio infestation levels. A strong correlation of r = 0.7 between the predicted and the observed levels of S. noctilio infestation using the developed multiple linear regression model showed that image texture is a promising tool for the detection, and ultimately mapping, of S. noctilio infestation in plantations. The result is critical for improving the management of insect infestation in plantations of South Africa.