In-situ plant hyperspectral sensing for early detection of soybean injury from dicamba

• Hyperspectral sensing can differentiate treated from untreated soybean.
• VIs useful in differentiating treated from untreated soybean from 24 to 72 HAT.
• Differentiation after treatment can be made regardless of spray rates.
• Differentiation after treatment can be made at early stages.
• Difficult to clearly differentiate dose response of soybean to dicamba <72 HAT.

Drift of dicamba onto non-target crops is a major concern because it is highly active on susceptible crops even at low doses. Early detection of crop injury is critical in crop management. A field study was conducted to determine spectral characteristics of soybean (Progeny P4819LL) treated with dicamba. Drift deposition of dicamba was simulated by direct application at 0.05 to 1.0 times of the recommended label rate (0.56 kg [ai] ha−1) to soybean at the 5- to 6-trifloliolate leaf stage, approximately 6 weeks after planting. The canopy spectral measurements were taken at 24, 48, and 72 h after treatment (HAT) using a portable spectroradiometer in the 325–1075 nm spectral range on 3 randomly selected plants within each plot with device optimisation and data calibration. The results indicated that it was difficult to clearly differentiate the dose response of soybean to different dicamba spray rates within 72 HAT. Regardless of spray rates the soybean treated with dicamba could be clearly differentiated from untreated soybean from 24 to 72 HAT through spectral vegetation index analysis with anthocyanin reflectance and photochemical reflectance indices with accuracies at 24, 48, and 72 HAT ranging from 76 to 86%. Simulated dicamba drift injured soybean and reduced its yield by 71 and 90% at 0.05 and 0.1 times recommended rate, respectively. This study demonstrated that hyperspectral remote sensing has a potential in early detection of soybean injury from exposure to off-target dicamba drift at sub lethal rates in the field.