|Van Vleck, Lloyd|
Submitted to: Crop Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/1/2000
Publication Date: N/A
Citation: N/A Interpretive Summary: Estimates of genetic variances should consider genetic relationships among all individuals. Consideration of all relationships is computationally expensive and with many sires or lines might be impossible. A common practice has been to ignore all relationships except for those due to having a common sire or being a member of a line. This pilot analysis of crosses of inbred lines of corn compared estimates when relationships amon lines were used or were ignored. Over all nine traits, estimates of genetic variance ignoring relationships were only 51 to 87% as large as estimates using all relationships for one group of lines and from 62 to 100% for the other group of lines. These results show that potential gain from selection would be underestimated if genetic variance was estimated ignoring relationships among inbred lines. The same result has been found with animal analyses.
Technical Abstract: Phenotypic data were obtained for eight traits from design II matings of seven unique sets of four fully inbred lines of each of two different heterotic groups of corn with a total of 112 crosses grown in up to 20 locations in two years. In the first model, lines within each heterotic group were considered random and unrelated. This model was compared with a model with numerator relationships among lines included. The second set of comparisons was similar to the first, but with effects of lines in the first heterotic group considered as fixed in both models. For the third set of comparisons, effects of lines in the second heterotic group were considered as fixed in both models. Estimates of variance components were obtained with a restricted maximum likelihood algorithm. Estimates of variance due to line effects with lines assumed unrelated divided by estimates taking relationships into account were as follows: 1) with both sets of lines considered random, ratios ranged from 0.50 to 0.87 with average of 0.67 for SSS lines and from 0.62 to 1.00 with average of 0.82 for NSS lines, 2) with effects of lines in the NSS heterotic group considered fixed, ratios ranged between 0.59 and 0.89 with average of 0.68 and 3) with effects of lines in SSS heterotic group considered fixed, the ratios ranged between 0.68 and 0.94 with average of 0.82.This reduction indicates potential gain from selection would be more than predicted from estimates of variance due to line effects ignoring relationships among lines.