Abstract
Celosia (Celosia argentea), is an important tropical vegetable for households in sub-Saharan Africa. Despite the multifaceted usefulness, available genotypes are low-yielding, and the vegetable faces dangers of genetic erosion due to poor research attention. The magnitude and pattern of variability will guide the choice of breeding methods for improvement. Twenty-one celosia accessions were evaluated in 2018 and 2019 to determine study genetic variability and heterotic patterns among clusters. Accessions and clusters differed significantly (p ≤0.05/0.01) for plant height, number of leaves/plant, stem weight, harvest index and dry matter content. Genotypic coefficients of variation; ranging from 37.89 to 0.12, were lower than phenotypic coefficients of variation which ranged from 114.55 to 0.12, both for number of leaves/plant and harvest index respectively, indicating the importance of environment in the variability. Discriminant analysis indicated low (8.12%) classification error rate, indicating the possibility of heterotic patterns among clusters. Principal component (PC) analysis controlled 73% of the observed variability among accessions and identified all measured traits as important contributors with loadings ranging from 0.30 (in PC 1) to 0.63 (in PC 2) for harvest index and stem weight respectively. Useful levels of association were also observed among measured traits. The study concluded that there was sufficient genetic variability for effective selection. Discriminant and principal component analyses identified plant height, number of leaves/plant and dry matter content as major contributors to variation among accessions. Weight of edible parts of Celosia can be simultaneously improved with plant height and number of leaves/plants.