Recurrent Selection for Kernel Weight in Spring Wheat 1

Successful recurrent selection in cross‐pollinated species is well documented, but few reports of its use in self‐pollinated species are available. Our objectives were to determine the gain from stuctured recurrent selection for kernel weight of wheat ( Triticum aestivum L.) and to determine its indirect effects on other traits. Ten lines, selected for high kernel weight from 100 lines screened, were intercrossed to form the initial C 0 population. Selection among S 1 plants for high kernel weight was used with approximately 22 plants selected each cycle. The selection intensity varied from 1.5 to 2.9% per cycle and averaged 2.2%. Eighty random S 3 and S 4 lines each from C 0 and C 2 , respectively, were used to compare variances and means. In addition, C 1 , C 2 , C 3 , and C 4 were grown as populations for evaluation of gain. Kernel weight increased 3% per cycle from C 0 through C 2 as measured by the lines, and 7% per cycle from C 1 through C 4 as measured by the populations. Genetic variance for most measured traits remained as high in the C 2 as the C0 population. Two cycles of selection resulted in lines with kernel weight higher than any lines in the C 0 population. These results were further confirmed by the results of the population tests which indicated continued gain for kernel weight through Cycles 3 and 4. Negative indirect effects of selection for kernel weight were observed for days to head, spikelets per spike, kernels/cm of spike, and spikelets/ cm of spike. Grain protein increased slightly while plant height, grain yield, and test weight did not differ over cycles of selection. Structured recurrent selection produced a population with much higher kernel weight than did single crossing and selection among segregating progeny, even though the original population was selected for high kernel weight and a high selection intensity was used.