Employing Spectral Features to Accelerate Sorghum Phenotyping Against Sap-Feeding Aphids.

Current efforts to detect and evaluate crop resistance to insect pests are limited by traditional phenotyping methods, which are time-consuming and highly variable. Sugarcane aphid (SCA; ) is a major pest of sorghum in North America that has emerged over the last decade and negatively impacts plant growth and development. The spectral reflectance data in visible, near infrared and shortwave infrared range (VIS-NIR-SWIR; 400-2500 nm) have been used to measure plant traits related to stress responses, nutrient dynamics, and physiological status. We examined the potential of spectral features (VIS-NIR-SWIR) to improve the current phenotyping methods in monitoring sorghum resistance mechanisms to SCA. We used eight sorghum lines that displayed varied levels of resistance to SCA and collected data from control and aphid-infested plants. Spectral feature data were collected using a leaf spectrometer, while plant physiological and chlorophyll fluorescence parameters were measured with LICOR and MultispeQ devices. The random forest classifier model differentiated the control and aphid-infested plants with a high accuracy of 87.4% with important spectral features in the VIS-NIR spectral range, particularly from 508 to 573 nm and 715 to 728 nm. The spectral indices exhibit significant difference in Greenness Index and Plant Senescence Reflectance Index in aphid-infested susceptible lines (BTx623, SC1345) compared with control plants. In addition, plant physiological parameters, such as stomatal conductance and chlorophyll fluorescence, showed significantly higher value for aphid-infested resistant line (Tx2783) compared with susceptible line (BTx623) in both treatments. Further, a partial least square regression model demonstrated medium predictive capability for plant physiological parameters related to fluorescence. In summary, spectral features at VIS-NIR range demonstrated promising results in differentiating aphid-infested sorghum plants. This is a proof-of-concept study on potential of spectral sensing to develop an effective monitoring and phenotyping plant resistance to aphids.