Variation in vegetation structural complexity explains evapotranspiration in a tropical dryland ecotone

• Vegetation structure, including structural complexity (Db), plant area index, and effective number of layers, changed across a dryland ecotone in Ecuador. • Structural complexity (Db) explained a large share of the spatial variation in evapotranspiration, especially in the dry season. • During the dry season in particular, the vegetation-atmosphere feedback was tight. • Preserving structural complexity emerges as important to ensure drylands' continued functionality. In natural drylands, the spatial distribution of vegetation characteristics is shaped by environmental conditions, with aridity being a key factor. Aridity can be quantified using an aridity index, which relates precipitation to potential evapotranspiration (ETp). However, the extent to which variations in vegetation characteristics affect vegetation-atmosphere fluxes through actual evapotranspiration (ETa) or the ratio of actual to potential evapotranspiration (ETa/ETp) remains less clear. We hypothesized that vegetation structure features significantly contribute to the spatial variation in ETa and ETa/ETp in a dryland ecotone. In western Ecuador’s lowlands, characterized by a 210-day dry season, we assessed 3D vegetation structure using a ground-based mobile laser scanner on 75 plots across different forest types (xerophytic forest, dry forest, and evergreen forest) during both, wet and dry seasons. The mean plant area index (PAI) and stand structural complexity, measured as box dimension (D b ), were similar in xerophytic and dry forests but higher in the evergreen forest. The effective number of layers (ENL2D) was similar between the dry forest and evergreen forest. PAI slightly decreased from the wet to the dry season across all vegetation types, while D b and ENL2D remained stable. Satellite-derived ETa (from MODIS data) was similar across forest types during the wet season and higher in the evergreen forest during the dry season. ETa decreased from the wet to the dry season, despite similar ETp in the dry season. Models incorporating spatial autocorrelation, based on Moran’s Eigenvector Map, explained 35 % and 36 % of the spatial variability in ETa by PAI and D b , respectively, in the wet season, and very little in ETa/ETp. In the dry season, models with predictors PAI, D b or ENL2D explained over 70 % of the spatial variation in both ETa and ETa/ETp. Our findings reveal that differences in vegetation structure, particularly as measured by D b , significantly contribute to explaining spatial variation in ETa and ETa/ETp, especially during the dry season. We suggest that dry-season aridity likely drives adaptations in vegetation structure that reduce ecosystem evapotranspiration. By utilizing high-resolution 3D vegetation structure data, we can better understand the intricate relationship between forest structure and land–atmosphere feedback. Our findings emphasize the importance of preserving the complex structural diversity of dryland ecosystems to ensure their continued functionality.