To determine the genetic diversity of quinoa (Chenopodium quinoa Willd.), the crop's morphological traits and agronomic performance were evaluated during 1992-1993 and 1993-1994, using germplasm preserved in Bolivia and IBPGR descriptors as modified by the Fundacion PROINPA's Quinoa Project. Three multivariate methods were used to analyze diversity, allowing sets of individuals to be described and grouped, while considering multiple characteristics and their interrelationships. Principal component analysis was used to quantify the contribution of the three components most associated with total variance, and to identify the variables characterizing each of these components. The nonhierarchical procedure (k-means) of cluster analysis was used to group accessions according to genetic diversity. The resulting seven clusters were combined with passport data to provide useful descriptions of the gerrnplasm. Multiple discriminatory analysis complemented the study, deriving, through the step-by-step procedure, six statistically significant (P = 0.001) functions. The discriminatory ability of each independent variable was quantified through the index of potency. About 92% of the accessions were readily classified into the seven groups identified by cluster analysis. The Mahalanobis distance (D-2) method was used to show their statistical differences. Even incorrectly classified accessions were identified and reassigned to a more appropriate group.
