Using hyperspectral reflectance-based predictive models for early Phytophthora infestans and Alternaria solani detection in potato
Kaitlin Gold: University of Wisconsin-Madison
<div>Late blight of tomato and potato (<em>Phytophthora infestans</em>) continues to be one of the most challenging diseases to sustainably and proactively manage. Potato early blight (<em>Alternaria solani</em>) occurs annually in Wisconsin. Growers spend upwards of $19 million to control these two destructive diseases. Our previous work established a non-destructive method of early late blight detection based on hyperspectral reflectance that can identify infected plants with >85% accuracy 2-4 days before visual symptoms appear during its biotrophic phase. Our objective was to use this methodology to distinguish latent <em>P. infestans</em> infection from latent and symptomatic <em>A. solani </em>infection. We conducted two growth chamber experiments using <em>P. infestans</em>, <em>A. solani</em>, co-inoculation, and control, non-inoculated plants to identify significant wavelengths for detection and differentiation. We measured continuous visible to shortwave infrared reflectance (400-2500 nm) on leaves using a portable spectrometer with contact probe at 12-24hr intervals. We could both detect and distinguish early blight from late blight infection with >80% accuracy at 24hr post-inoculation, and upwards of 90% accuracy after 48hr post-inoculation. Shortwave infrared wavelengths (>1300 nm) were important for disease detection and differentiation. Our results support the potential use of hyperspectral reflectance-based predictive models as tools for rapid, early, real-time detection of two important diseases of potato.</div>
|
