TY - GEN A1 - Limberger, Oliver A1 - Homeier, Jürgen A1 - Gonzalez-Jaramillo, Victor A1 - Fries, Andreas A1 - Murkute, Charuta A1 - Trachte, Katja A1 - Bendix, Jörg T1 - Foliar trait retrieval models based on hyperspectral satellite imagery perform well in a biodiversity hotspot of the SE Ecuadorian Andes T2 - International journal of remote sensing N2 - The determination of the spatial distribution of canopy traits is crucial for understanding the spatio-temporal dynamics of ecosystem functions, such as carbon sequestration, water and energy fluxes in tropical montane forests. Especially in remote areas such as the Andes of south-east Ecuador, remote sensing using satellites has been proven to provide valuable information on canopy traits. However, the performance of the multispectral models to date is limited. Here, we analyse the potential of DESIS (DLR Earth Sensing Imaging Spectrometer) hyperspectral surface reflectance data for the prediction of specific leaf area (SLA), foliar toughness, nitrogen (N) and phosphorus (P) content by calibrating a PLSR model for each foliar trait. Model validation showed a high explanation of variance (R2 : 0.78–0.92) and a low model error (mean-normalized RMSE: 8–13%) for all traits, showing an improved performance compared to models based on multispectral imagery and ancillary data within the same forest type. Hyperspectral models predicting foliar traits from individual tree crowns showed a lower performance for most traits, probably related to the difference in spatial resolution compared to this study. Given the high species richness of the Andean biodiversity hotspot and the complex topography, the high quality of the DESIS models is remarkable, considering that the multispectral models from the region require topographic predictors in addition to reflectances to achieve a good performance. The spatial trait maps generated by our models based on the reflectances only show clear relationships with terrain elevation. Thus, the hyperspectral trait maps from satellite orbit are capable of supporting large-scale ecological analyses between features, diversity and ecosystem functions. Y1 - 2025 U6 - https://doi.org/10.1080/01431161.2025.2511211 SN - 1366-5901 SP - 1 EP - 19 PB - Taylor & Francis CY - London ER -