Use of leaf hyperspectral data and different regression models to estimate photosynthetic parameters (Vcmax and Jmax) in three different row crops

By 2050 future global food demand is projected to require a doubling of agricultural output, and climate change will exacerbate this challenge by intensifying the exposure of field crops to abiotic stress conditions, including rising temperature, increased drought, and increased CO₂ concentration ([CO₂]). One of the keys to improving crop yield under different stresses is studying is photosynthesis. Photosynthetic parameters, such as the maximum rate of carboxylation of RuBP (V_c,max), and the maximum rate of electron transport driving RuBP regeneration (J_max) vary in response to climate conditions and have been identified as a target for improvement. However, the techniques used to measure these physiological parameters are very time consuming, ranging from 30 to 70 min per measurement and require specialized personnel. Therefore, breeding or genetic mapping for these traits under these conditions is prohibitively time-consuming. Spatial and temporal variation in plant photosynthesis can be estimated using remote sensing-derived spectral vegetation indices. Spectral estimates of green vegetation biomass and vigor, including vegetation indices such as the Normalized Difference Vegetation Index (NDVI), are widely used to estimate vegetation productivity across spatial and temporal scales but are unable to provide assessments of specific photosynthetic parameters. For that reason, hyperspectral remote sensing shows promise for predicting photosynthetic capacity based on more detailed leaf optical properties. In this study, we developed and assessed estimates of Vcmax and Jmax through four different advanced regression models: PLS, BR, ARDR, and LASSO based on leaf reflectance metrics measured with an ASD FieldSpec4 Hi-RES of different crops under different environmental conditions such as (1) different varieties of soybean under high [CO₂] and high temperature, (2) different varieties of peanut under drought stress and (3) 20 varieties of cotton diverse origin and grown under field conditions. Both phenotypic variability and varying levels of stress were employed with each crop to ensure adequate ranges of responses. Model sensitivities were assessed for each crop and treatment separately and in combination in order to better understand the strengths and weaknesses of each model in all the different conditions. For the combination of three species, all the models suggest a robust prediction of Vcmax around R²:0.67 and the same for the Jmax R²: 0.55.