USING SINGLE-REPLICATE ON-FARM TESTS TO ENHANCE CULTIVAR PERFORMANCE EVALUATION

Release of mediocre and inferior cultivars can have costly and deleter ious consequences for breeders, crop improvement associations, and gro wers. Multilocational testing of new cultivars in replicated small-plo t trials over several years has traditionally been used to screen new cultivars prior to release. Single-replicate on-farm tests (SROFTs) of spring barley (Hordeum vulgare L.) cultivars were conducted by easter n Washington growers from 1990 to 1992 to compare with the results obt ained from replicated small-plot trials. The comparison of interest wa s the risk of releasing inferior barley cultivars and the similarity o f cultivar response in the two testing systems. Both systems shared fo ur common cultivars in 1990, and five common cultivars in 1991 and 199 2. For comparison purposes, single-replicate tests were grouped with f our replicates nested within groups to compare with four replicates wi thin each small-plot location. The risk of releasing inferior cultivar s was estimated for each testing system x year combination as the prob ability of failing to detect true significant differences among cultiv ars (Type II error risk) while keeping the Type I error fixed at the 0 .05 level. Type II error was lowest for SROFT in 2 of the 3 yr and equ al to small-plot testing in the third year. Cultivar response was simi lar in both systems with the exception of one cultivar in 1990. In the final stage of cultivar selection, SROFT results can be combined with small-plot trial results with small risk of releasing more inferior c ultivars or selecting different cultivars.