Author
 |
GUZMAN, PETER - IOWA STATE UNIVERSITY |
|
Lamkey, Kendall |
|
Submitted to: Maydica
Publication Type: Proceedings Publication Acceptance Date: 3/21/1999 Publication Date: N/A Citation: N/A Interpretive Summary: The number of individuals kept and crossed together to start the next round of selection is one of the most important decisions made by a breeder. This number influences the resources required to conduct selection, the amount of gain made from selection, and how quickly genetic variation erodes. If the number kept is too large, then the resources required to maintain a given selection intensity will be large (for example if you keep 20% of th total, then if you want to keep 20 individuals, 100 need to be evaluated versus keeping 10 individuals and evaluating 50), but genetic variation will be maintained. If the number kept is small resource usage is reduced, genetic gain is enhanced, but genetic variation will be sacrificed. Very little empirical information is available to help breeders determine the number of individuals to keep in a selection program. Starting from a common source, four selection programs were conducted where 5, 10, 20, and 30 individuals were kept. Five cycles of selection were conducted for eac of the four selection groups. The purpose of the study was to compare the amount of gain realized by selection with that predicted by quantitative genetic theory. It is commonly recommended in breeding programs to keep 20 to 40 individuals to minimize inbreeding in the population and realize the maximum amount of genetic gain. These results demonstrated that there is no theoretical or practical advantage to keeping more than 10 individuals in a selection program. These results allows breeders to either effectively double their selection intensity and use the same amount of resources or to maintain their current selection intensity and cut resource use in half. These results have important implications for breeders designing breeding programs. Technical Abstract: Random genetic drift and inbreeding depression resulting from the use of small effective population size limit the gain of recurrent selection programs. The objectives of our study were to predict the gains of S2, S1, full-sib, and modified-ear-to-row recurrent selection methods in maize (Zea mays L.) with varying effective population sizes of 5, 10, 20, and 30 and to compare these predicted gains with the realized gains obtained in a related study. Using the variance estimates of the BS11C0 population, predicted gain cycle-1 was computed based on single trait selection and index selection. Predicted gain cycle-1 for S1-progeny selection method in the BS11 maize population was also computed by multiplying the heritabilities and the selection differentials obtained from the original selection. Results showed that the trend was similar for single trait selection and index selection. Relatively higher predicted values were observed for the inbred progeny recurrent selection methods in all traits. Predicted gain cycle-1 for a trait in a recurrent selection method increased with increasing effective population size but differences among the predicted values were agronomically insignificant. While the predictions based on single trait and index selection showed increasing grain yield with increasing effective population size, results from the selection trials of the S1 programs showed otherwise. The comparison between our predictions and the realized gains obtained in a related study indicated that intermating more than 10 individuals would not result in a significant response over a few generations of selection. There is no distinct advantage of using an effective population size of greater than 10 to realize gain in a short-term maize recurrent selection program. |