Abstract
Livestock farming in the Colombian Andean-Amazon region is characterized by its low competitiveness and its negative socio-environmental impact, being a factor of deforestation at national level and contributing to the generation of greenhouse gases. Silvopastoral systems (SPS) are a strategy to reduce this impact. This investigation was conducted aiming to identify and perform the nutritional characterization of tree and shrub species relevant to SPS in the Andean-Amazon region of Alto Putumayo. A diagnosis was carried out through semi-structured surveys, field visits, and knowledge dialogue sessions, recognizing potential species and their uses. For the nutritional characterization of the forage, the samples were analyzed in the laboratory using Near-Infrared Spectroscopy- NIRS technology. Subsequently, a statistical analysis was performed using an unsupervised machine learning technique based on principal component analysis. Hierarchical clustering using Ward's method identified 38 woody species with different growth habits (shrubs and trees), distributed across 20 families, mainly Asteraceae, Fabaceae, and Moraceae. Approximately 60% of the identified species are relevant in animal nutrition. The species with the best nutritional quality were Sapium stylare, Sambucus nigra, and Tithonia diversifolia. Likewise, four groups were determined, where the first two groups account for 79.1% of the variance. Collectively, our findings indicated that Alto Putumayo is home to a wide diversity of tree and shrub species with forage potential, suggesting a potential for their use in agroforestry.