Selection of Vis-NIR wavebands for forecasting apple fruitlet drop in response to chemical treatment

Abstract

Apple trees tend to bear an excess of young fruits (fruitlets). Chemical thinning with an application of bioregulators early in the season is currently the most common solution to adjust fruit load. However, the results of this approach are highly inconsistent between seasons. In the present study, we identified the most informative bands in the visible-near infrared range for forecasting fruitlet destiny in response to the thinner application, which could be used to support grower decisions. “Golden Delicious” apple trees were monitored during two consecutive seasons, following the application of synthetic auxins commonly used in commercial orchards. Simple bands, band differences, and band ratios were investigated as features for forecasting fruitlet destiny from the first week of monitoring. Optimal threshold values and best wavelengths combinations were determined via the receiver operating characteristic (ROC) curve. At the early stage of fruitlet development (4–6 days after treatment, DAT), applying thresholding to the band difference R 973 − R 404 was preferable (accuracies ranging from 66% to 87%), while applying thresholding to the band ratio R 693 / R 674 was preferable at 7–12 DAT (accuracies ranging from 76% to 95%). The present results indicate that the development of a simplified portable device for forecasting apple fruitlet drop seems feasible. Complimentary measurements performed on intact vs. denuded fruitlets revealed that the spectral differences observed can be explained by substantial changes in trichomes density that occur naturally during fruitlet development and affect the apparent pigment absorption. • Spectral wavebands were investigated to forecast apple fruitlet drop. • Optimal bands were identified with receiver operating characteristic curve. • Two to four optimal bands led to 66–95% accuracy depending on monitoring dates. • Later monitoring dates resulted in higher forecast accuracies. • Trichome density changes strongly affected apple fruitlet reflectance.