LAUSR

ABSTRACT Amaranth (Amaranthus spp.), although primarily used as a leafy vegetable, could have additional worth in that the seed could represent raw materials in the production of protein concentrates. However, seed yield is generally low due to the continuous use of unimproved cultivars. A breeding program focused on developing improved varieties requires knowledge of selection criteria for seed yield improvement within, and among, cultivated Amaranthus species. Stepwise multiple regression, genotypic correlation, and path analysis for yield component characters affecting seed yield were carried out in genotypes belonging to species of Amaranthus to identify the best selection criteria for increased seed yield. Stepwise multiple regression analysis indicated different parameter models were able to predict seed yield in A. hypochondriacus L, A. cruentus L, and A. dubius (Mart. Ex Thell.). Path coefficients indicated one thousand seed weight and stem diameter should be used in selection aimed at improving seed yield in A. hypochondriacus. Thousand seed weight and the number of leaves could be effective selection criteria for seed yield improvement in A. cruentus. The number of days to flowering could serve as an indicator for indirect selection for high seed-yielding genotypes in A dubius. Improve seed yield in Amaranthus brings with it increased micronutrient content.