Multivariate clustering of curves on dehydration in sweet potato roots

Minimizing post-harvest losses during sweet potato storage is imperative for the producer and consumer. Therefore, the use of statistical tools to select the superior genotypes that have good post-harvest characteristics will contribute to breeding programs. The aim of this study was to demonstrate the applicability of the multivariate clustering technique as an alternative in the post-harvest study on root dehydration among sweet potato accessions. A total of 74 accessions of sweet potatoes were evaluated in a randomized complete block design in four replicates. The roots were stored in plastic boxes at room temperature and the loss of fresh matter was measured on times 0, 4, 8, 12 and 16 days after harvest. The multivariate clustering was based on the Euclidean distance and the Tocher optimization method on adjusted curves using linear and nonlinear models. The non-linear best fit model was Brody, which allowed the discrimination of accesses with higher and lower dehydration. The multivariate grouping of curves was efficient in the post-harvest study on sweet potato access roots.