Lauren Ross

Fruit cracking in pomegranate (Punica granatum L.) is a significant preharvest disorder that reduces yield, quality, and marketability, resulting in losses of up to 75% in Mediterranean climates. Cracking occurs when the peel cannot accommodate internal aril expansion and is influenced by factors such as climate, nutrition, and irrigation. Although pomegranate is considered drought-tolerant, cracking and russeting remain widespread, underscoring the need for improved water management. This study evaluated the effects of irrigation strategies on plant water status (PWS), yield, and cracking prevalence in a commercial ‘Wonderful’ orchard in Tsor’a, Israel, over two seasons (2023–2024). Four irrigation regimes were tested: standard industry control, early irrigation, high water, and low water. PWS was monitored using stem water potential (SWP), trunk growth rate (TGR), stomatal conductance (SC), and plant area index (PAI). Concurrently, UAV-based multispectral, thermal, and LiDAR imagery were integrated with meteorological and physiological data for plot-scale monitoring. Random Forest (RF) models, trained within the caret framework in R with Boruta feature selection, identified spectral and structural indicators as key predictors. Strong predictive performance was achieved for SWP (R² = 0.815, RMSE = 0.145 MPa), TGR (R² = 0.781, RMSE = 6.975 μm), and SC (R² = 0.822, RMSE = 0.028 mol m⁻²s⁻¹), demonstrating that UAV-based remote sensing can effectively capture spatiotemporal dynamics of PWS and cracking incidence. By integrating UAV imagery with machine learning, this study presents a scalable framework for precision irrigation scheduling in pomegranate orchards, supporting strategies to sustain production and reduce yield losses from fruit cracking.