Non-destructive prediction of pigment content in lettuce based on visible-NIR spectroscopy

Background: Lettuce (Lactuca sativa L.) is one of the most important salad vegetables in the world, with a number of head shapes, leaf types and colors. The lettuce pigments play important physiological functions, such as photosynthetic processes and light stress defense, but they also benefit human health because of their antioxidant action and anticarcinogenic properties. In this study three lettuce cultivars were grown under different farming systems, and partial least squares models were built to predict the leaf chlorophyll, carotenoid and anthocyanin content.

Results: The three proposed models resulted in high coefficients of determination and variable importance for the projection values, as well as low estimative errors for calibration and external validation datasets. These results confirmed that it is possible to accurately predict chlorophyll, carotenoid and anthocyanin content of green and red lettuces, grown in different farming systems, based on the spectral reflectance from 500 to 1000 nm.

Conclusion: The proposed models were adequate for estimating lettuce pigments in a quick and non-destructive way, representing an alternative to conventional measurement methods. Prediction accuracies were improved by using the detrending, smoothing and first derivative pretreatments to the original spectral signatures prior to estimating lettuce chlorophyll, carotenoid and anthocyanin content, respectively. © 2016 Society of Chemical Industry.