Diseases cause enormous losses of yield and quality for crop producers worldwide. To meet future food demands, crops are bred for resistance to as many of these maladies as possible. One such disease, anthracnose [Colletotrichum sublineola], is a fungal disease of great importance to sorghum [Sorghum bicolor, L. Moench] production because it causes significant annual economic losses in the crop. Breeding for anthracnose resistance requires time-consuming phenotyping, which is subjective and conditional to the evaluator. It is possible that quantitative assessment using high-throughput methodologies to estimate the trait may be more effective. In this study, we present an in-depth statistical analysis of fixed-wing, unmanned aerial system (UAS) evaluation of anthracnose incidence and severity in sorghum using normalized difference vegetation index (NDVI). In early phases of infection, correlations between ground-truth and UAS estimates of anthracnose were moderate but they increased substantially by the end of the season (r = −0.55 to −0.95). Additionally, both metrics had moderate-to-high repeatabilities throughout the growth period (R = 0.60–0.90), indicating they were consistently able to differentiate genotypes. Finally, we found that the UAS-derived measurements (R2 = 0.377, 0.473) were better associated with ground-truth measurements (R2 = 0.278, 0.347) for grain yield under anthracnose pressure. The results of this study indicated that fixed-wing UAS could potentially be effective for evaluating anthracnose disease present in sorghum, and the greater range of the UAS allowed the effective evaluation of larger numbers of plants than ground truth or traditional remote sensing methods.
Supplementary notes can be added here, including code and math .