Spectral estimation of carnosic acid content in in vivo rosemary plants

Rosemary (Salvia rosmarinus (L.) Schleid., Handb. syn. Rosmarinus officinalis L.) extracts are widely used as natural preservatives due to their antimicrobial and antioxidant properties, which are attributed to the phenolic diterpenoid carnosic acid (CA). Growers are rewarded based on CA content in their rosemary leaf harvested. Conventional methods for estimating leaf CA content are destructive and often time-consuming. This preliminary study presents a spectral non-destructive approach for in vivo estimation of CA content in different rosemary cultivars based on the reflectance spectra of their canopy. The proposed approach is based on the characteristic rosemary absorption features along the visible and shortwave infrared spectral regions at 550 nm, 1200 nm, and 1690 nm, respectively, attributed to leaf color, the oxygen-hydrogen bond bending in water molecules, and distinctive carbon-hydrogen bond features typical for terpenes and phenolic compounds. Correlations between measured CA content by high-performance liquid chromatography (HPLC) and leaf reflectance spectra, normalized spectral indices, and latent components obtained by genetic algorithm-based partial least squares regression (GA-PLSR) were assessed using data collected from 79 rosemary cultivars. The GA-PLSR model successfully predicted the CA content among the various cultivars, further providing evidence of high weightage to the above-mentioned absorption features also obtained from two best-wavelength combination selections. Randomly selected canopy spectra were used to calibrate and simultaneously cross-validate 100 iterations, using the ‘leave-k-out’ approach. The root mean squared error (RMSE) obtained for calibration and cross-validation were 0.86% and 1.15% CA content from the dry leaf matter, and the residual prediction deviation (RPD) were reported to be 2.71 and 2.05, respectively. This work will set the stage for precise planning of harvesting time to ensure increased yield and income for the farmers and improved utilization of resources.