Search for a command to run...
Sustainable cocoa production underpins rural livelihoods across the tropics and offers significant potential for both climate change adaptation and mitigation. Although the biodiversity value of cocoa agroforestry systems is well documented, their functional capacity to buffer drought stress and enhance climate resilience remains insufficiently understood. In particular, scalable methods for monitoring plant physiological responses to water stress across structurally heterogeneous agroforestry landscapes are urgently needed. In this study, we assess the potential of multispectral drone imagery to detect drought-related physiological dynamics in cocoa agroforestry systems, with a specific emphasis on the role of shade tree leaf phenology.We integrated high-resolution multispectral drone imagery with in situ physiological measurements across ten smallholder cocoa plantations of comparable age in the northern cocoa belt of Ghana. Thirteen shade tree species, representing distinct functional groups based on leaf phenology, were selected. For eight individuals per species, we quantified structural traits (diameter at breast height, tree height, and canopy area) and phenological status, and measured leaf-level transpiration and stomatal conductance using a LI-600 porometer. Multispectral imagery acquired during the late wet, mid-wet, and peak dry seasons between 2021 and 2023 was used to derive the Green Normalized Difference Vegetation Index (GNDVI), a spectral proxy sensitive to chlorophyll content and photosynthetic activity. We observed pronounced seasonal and functional-group-specific differences in canopy reflectance, with significant interactions between season and shade tree phenology. GNDVI was strongly correlated with key physiological traits, particularly stomatal conductance, and exhibited consistent responses to seasonal climatic variation. These results demonstrate that drought-induced physiological stress, expressed as reductions in stomatal conductance, can be reliably predicted from spectral traits derived from high-resolution multispectral drone imagery, highlighting its potential as a scalable tool for assessing drought resilience in cocoa agroforestry systems.