Estimation of synthetic variety yields in pearl millet through parental line evaluation, per se and in tester combinations.

摘要

Three pearl millet (Pennisetum glaucum (L.) R. Br.) synthetic varieties were synthesized each from a different designated set of unrelated parental lines. Each set of parental lines was evaluated for general combining ability in two ways: (a) through a sample of paired testcrosses in a central circulant partial diallel; and (b) through topcrosses using a common unrelated variety tester. The two methods agree in identifying parental lines with good and poor combining abilities.;Yield for each of all possible single cross hybrids in each set were predicted by summing the actual mean yield of the F;Predicted yields of the synthetic varieties using Wright's and Busbice's formulae were similar. Predicted synthetic variety yields using combining abilities derived from topcrosses were 12% higher than predicted yields using testcrosses. Comparison of predicted and actual yield or Syn 3 showed that predictions using testcrosses were similar to the actual yield and those using topcrosses were 17% higher than the actual yield.;Although topcrosses and testcrosses are both useful in selecting parental lines for a synthetic variety, based on general combining ability, the use of heterosis derived from topcross F;The proportion of selfing in pearl millet populations influences the predictions, estimates at 20% selfing seem to be close to the actual performance. Average synthetic variety yields were 30% over the mean yield of the parental lines, showing a considerable amount of heterosis. The highest yielding synthetic variety was made from a set of parental lines that had a high mean yield and a high mean yield of the F