Integrated multivariate analysis of morphological and yield traits in native Capsicum chinense ecotypes grown in acidic soils of the Peruvian Amazon

Introduction The comprehensive characterization of native Capsicum chinense ecotypes represents a strategic priority for genetic improvement, germplasm conservation, and the sustainable use of Amazonian crops. The objective of this study was to evaluate morphological, phenological, and productive variability among 12 ecotypes from the Peruvian Amazon by integrating multivariate análisis and machine learning with soil physicochemical characterization. Methods The research was conducted on acidic tropical soils with low organic matter content and limited availability of exchangeable bases, conditions representative of degraded Amazonian agroecosystems, which enabled the assessment of soil–plant interactions and their influence on phenotypic expression and crop yield. Results The results revealed a broad, well-structured range of phenotypic variability, with fruit diameter, fruit length, fruit weight, and seed weight identified as the primary morphological determinants of yield and adaptive capacity under low-fertility soil conditions. Principal component analysis indicated that four components explained more than 70% of the total variance, primarily associated with productivity, fruit morphometry, and phenological traits. Cluster análisis identified groups with high internal consistency, while linear discriminant analysis validated the phenotypic structure, achieving a classification accuracy of 91.8%. The ecotypes JEB-028 and LAG-022 exhibited superior productive performance, whereas BAL-012 and YUR-001 demonstrated greater phenotypic stability under restrictive soil conditions. Discussion Overall, these findings confirm the strategic value of native Amazonian germplasm and underscore the importance of integrating edaphic diagnostics into genetic selection programs and into strategies for the sustainable management and restoration of degraded agricultural soils in the Amazon.