Phenotypic Variability of tarwi (Lupinus mutabilis S.) in Peruvian Germplasm Collections

Given the increasing loss of genetic diversity worldwide, phenotypic characterization becomes indispensable to identify resilient varieties capable of diversifying and strengthening agricultural production of lesser-known crops such as tarwi (Lupinus mutabilis Sweet). In this context, the objective of the present investigation was to characterize the phenotypic diversity of 41 tarwi accessions, conserved in the germplasm bank of the National Institute of Agrarian Innovation of Peru. The accessions were evaluated during two consecutive growing seasons between 2022 and 2024 at the Santa Ana Experimental Station, under local conditions. Thirty morphological descriptors (17 qualitative and 13 quantitative) were used, following the guidelines of the International Board for Plant Genetic Resources. Data were analyzed using descriptive statistics, principal component analysis (PCA), hierarchical clustering, and correlation analysis for quantitative descriptors, and a frequency table and Shannon-Weaver diversity index (Hʹ) for qualitative descriptors, which allowed for the classification and understanding of variability among accessions. The results revealed high phenotypic variability, particularly in characteristics linked to yield, plant architecture, and floral attributes. From this, the accessions were grouped into three morpho-agronomic types: (1) highly productive genotypes, (2) accessions with vigorous vegetative development, and (3) short-cycle plants with moderate yields. Likewise, positive correlations were identified between characteristics such as the number of primary branches and the number of leaflets per leaf. Yield per plant showed a significant association with the total number of pods, 100-seed weight, and seed thickness. Conversely, a negative correlation was observed between plot yield and days to harvest. It is concluded that there is significant agromorphological diversity, with differentiated functional groupings and useful associations for selection processes. Although the results are promising, the study needs to be expanded with more accessions to validate and strengthen these inferences.