Sustainable water management and enhanced irrigation efficiency in the growing macadamia sector in subtropical regions such as South Africa are essential amid severe periodic water scarcity exacerbated by climate change. This requires precise modelling of tree water demand under varying conditions. However, current methods often lack accuracy or require extensive data inputs. In this study, we adopted and evaluated a simple, mechanistic, low data-input transpiration model for macadamia trees. To this end, we conducted a comprehensive experimental study in the sub-humid Levubu region, South Africa, collecting tree sap velocity data from two macadamia cultivars, along with microclimate and soil water data over two seasons. First, the model was calibrated under non-limiting water conditions using data on tree-intercepted radiation, vapor pressure deficit (VPD), and canopy conductance to simulate potential tree transpiration (Td), representing the upper limit of macadamia water use. Secondly, we further developed the model to simulate Td for water deficit conditions by rescaling simulated potential Td based on the observed fraction of transpirable soil water (FTSW). The performance of the calibrated model was validated against observed Td from (spared) independent datasets for both cultivars.
Observed macadamia Td showed pronounced seasonal variability (ranging from 0.6 mm d−1 in winter to 1.3 mm d−1 in summer), largely influenced by varying VPD and FTSW. The model captured the strong response of stomatal closure to increasing VPD, reflecting the conservative water use of macadamia trees. Model performance was satisfactory for both cultivars, and under both non-limiting and water deficit conditions, with lower relative error measures in the latter. This indicates that the improved model under water deficit is well-suited for accurately estimating macadamia Td under heterogeneous environmental conditions, making it a valuable tool for optimizing irrigation practices and conserving water resources in macadamia orchards.