Effectiveness of classes of testers for evaluation of lines as potential parents for grain sorghum (Sorghum bicolor (L.) Moench); hybrids

The choice of appropriate testers, which are still not well defined in sorghum, for initial evaluation of sorghum lines as potential parents for hybrids determine the success of hybrid breeding programs. In two experiments over 2 to 3 locations in 2 to 3 years, effectiveness of inbreds, F1$\sp{\rm \prime S}$ or double cross was compared, and a population (KP9B) as testers in determining general combining abilities (GCA) of new inbred lines as parents for sorghum grain hybrids in the Milo-Kafir cytoplasmic male sterility system. Effectiveness was based on higher GCA\u27s of testers, consistency in classifying lines across environments, simplicity in use, and higher dispersion ability in testcrosses. Trials in experiments one and two included testcross hybrids involving a maximum of twelve male sterilized restorers and sixteen male sterilized maintainers used as testers to evaluate fourteen heterotic maintainer and fifteen restorer lines respectively and the corresponding inbred parental lines. Experiment one was conducted at Mead, Nebraska, and Hesston Kansas, in 1993-1994 and experiment two at Mead, 1992-1994. Regardless of tester types, GCA\u27s were more important than specific combining abilities. GCA\u27s of some high yielding inbreds were greater than or equal to GCA\u27s of related F1 testers. The latter were greater than or equal to GCA\u27s of F1$\sp{\rm \prime S}$ related to low yielding inbred testers and were in general greater than or equal to GCA of low yielding testers of any kind. Rank correlations for similarity of testers in classifying lines within environments or between testcross yields and yields of parental lines, and for consistency of any tester in ranking line s across environments were low. However consistency in identifying the majority of the best 45% lines across environments was slightly in favor of the F1 and low yielding inbred testers. Few high yielding inbred testers tended to discriminate among lines more effectively than other testers. When considering all attributes, no single tester was effective. These results suggest that one more dispersive inbred and one more consistent and less biased F1 tester used simultaneously may effectively identify new B and R hybrid parental lines with good combining abilities before extensive hybrid testing