Spectral Classification of Reniform Nematode (Rotylenchulus reniformis) Infested Cotton Plants

摘要

Field tests were performed to compare hyperspectral data from known Rotylenchulus reniformis levels to spectral data from production fields naturally infested with R. reniformis. Hyperspectral measurements were recorded using a hand-held spectroradiometer to determine reflectance properties of R. reniformis infested cotton plants. For hand-held data, a Matlab hyperspectral toolkit (MHTK) was used to extract relevant wavelengths. Relevant wavelengths identified for discriminating R. reniformis populations were within the visible and near-infrared portion of the electromagnetic spectrum. Discriminant models were created using the relevant wavelengths along with R. reniformis population ranges and analyzed with MHTK self-organizing maps and linear discriminant analysis (LDA). A cotton plant's single leaf was observed to be the best reference with up to 100% classification accuracies for in-season analysis of production fields in acquiring reflectance information relative to R. reniformis presence and population estimation on cotton. However, plant canopy was the best reference (100% classification accuracies) for controlled field microplot cotton reflectance analysis suggesting that the controlled field environment may assist in regulating atmospheric effects and background spectral effects normally present in cotton production fields. Forty days after planting was found to be the best time to acquire cotton plant reflectance data and soil samples for R. reniformis analysis. Data acquired during this time using the appropriate target, single leaf in production field analysis or plant canopy in microplot analysis, provide highest classification accuracies in accurately detecting and estimating R. reniformis on cotton plants.