European Journal of Sustainable Agroecosystems

The inability of crops to access an adequate amount of water is the largest abiotic limitation on crop yields across the globe. Understanding how crops behave when subjected to both gradual soil and atmospheric drying is a major issue in agronomy, plant physiology, and agricultural engineering. Ecophysiological modeling, which uses mathematics to characterize the processes that help plants obtain water under duress, has witnessed substantial progress over the past ten years. This review of literature in major databases, such as Scopus, Web of Science, PubMed, ScienceDirect, MDPI, Frontiers, as well as reports by FAO and WMO, published primarily between 2015 and 2026, is about the state of ecophysiological modeling of plants under water deficit scenariosThe aim of the study was to delineate the important paradigms used in modeling such as the models of stomatal conductivity, crop simulation models based on process technology (such as DSSAT, AquaCrop, and APSIM), indices of water stress based on remote sensing, and predictive models based on artificial intelligence, and compare them in terms of their postulates, effectiveness, and limitations. The research results show that the physiological models of stomatal activity that take into account several factors (temperature, water vapor pressure deficit, and soil moisture) outperform empirical models that consider one factor only, that none of the modeling methods is superior regardless of the culture and environmental conditions, and that using unmanned aerial vehicles for thermal and multispectral imaging together with machine learning makes it possible to reach the accuracy level of physiological measurements in water stress detection. Some of the persistent gaps are having limited ability to develop cross-model and cross-scale validations, having poor integration of subsurface hydraulic features into canopy-scale models, limited representation of genotype-specific stomatal behavior that cannot be accounted for, and a lack of studies linking mechanistic modelling with data-driven approaches using data obtained under actual field conditions. The authors of the review say that hybrid physics-assisted machine-learning systems, combined with multi-source sensing technologies and standardized validation procedures, are the best approach towards development of accurate and transferable ecophysiological drought response models for `climate-resilient agriculture`.