Hyperspectral soybean reflectance affected by soybean aphid (Hemiptera: Aphididae)

Monday, June 1, 2015: 9:15 AM
McDowell + Tuttle (Manhattan Conference Center)
Tavvs Alves , Entomology, University of Minnesota, Saint Paul, MN
Robert Koch , Entomology, University of Minnesota, Saint Paul, MN
Ian MacRae , Dept. of Entomology, University of Minnesota, Saint Paul, MN
Spectral reflectance can potentially provide estimates of plant physiological stress caused by soybean aphid Aphis glycines Matsumura (Hemiptera: Aphididae). Wavelength-specific spectral changes, however, may be required to characterize aphid-induced stress and distinguish it from other stressors. The objective of our study was to identify optimal spectral wavelengths associated with cumulative pressure of aphid feeding on soybean plants. Field trials were conducted in 2013 and 2014 using caged-plots arranged in a randomized complete block design. Early-, late-, and non-infested treatments with eight and seven replications (2013 and 2014, respectively) were used to create a gradient of soybean aphid pressure. Whole-plant soybean aphid densities were recorded weekly and used to calculate cumulative aphid-days (CAD). Plant reflectance on ultraviolet, visible and infrared spectral ranges was recorded using a ground-based hyperspectral spectrometer on 2,151 wavelength channels. Nine linear regression model equations were used to find the best fit for the relationship between CAD and plant reflectance at each wavelength for each evaluation date. Goodness of fit and complexity of the regression models were compared using Akaike Information Criterion (AIC). Our preliminary results indicated that significant models considering narrow wavelengths within near-infrared spectral range provided relatively better fit (lower AIC) than the significant regression models using ultraviolet, visible or far-infrared wavelengths. Future research is required to incorporate the use of soybean spectral response into a sampling plan to improve efficiency of soybean aphid management.
Previous Presentation | Next Presentation >>