Estimation of genetic parameters in barley (Hordeum vulgare L.) : II. Partial diallel analysis.

Four different sets of partial diallels were analysed for their relative efficiencies for estimating the genetic parameters in barley: (1) partial diallel with 12 parents, each involved in only 5 crosses; (2) partial diallel with 12 parents, each involved in only 3 crosses; (3) partial diallel with 8 parents, each involved in only 5 crosses; and (4) partial diallel with 8 parents, each involved in only 3 crosses. In partial diallel experiments, the estimates of gca effects were higher than in those of full diallel. Ranking pattern of the parents on the basis of gca effects in partial diallels deviated considerably from the ranking in full diallel. With decreasing "s" per parent, the deviation in ranking was also more. This clearly suggests the unsuitability of partial diallel analysis for screening high general combiners. Selection of best cross combinations is also not possible because only a sample of crosses (s out of n) is analysed under partial diallel so that there is every possibility of the best cross being excluded from the sample. In general, overdominance was exhibited, indicating that there is ample scope for heterosis breeding in barley.