METHODS TO EVALUATE POPULATIONS FOR ALLELES TO IMPROVE AN ELITE HYBRID

Elite hybrids can be improved by the introgression of favorable allele s not already present in the hybrid. Our first objective was to evalua te several estimators derived from quantitative genetic theory that at tempt to quantify the relative number of useful alleles in potential d onor populations. Secondly, we wanted to evaluate two proposed ways of determining relatedness of donor populations to the parents of the el ite hybrid. Two experiments, each consisting of 21 maize populations o f known pedigree, were grown at three and four environments in Minneso ta in 1991. Yield and plant height means were used to provide estimate s of each of the following statistics: (1) LPLU, a minimally biased st atistic, (2) UBND, the minimum estimate of an upper bound, (3) NI, the net improvement, (4) PTC, the predicted three-way cross, and (5) TCSC , the testcross of the populations. These statistics are biased estima tors of the relative number of unique favorable alleles contained with in a population compared to a reference elite hybrid. Based on rank co rrelations, all statistics except NI ranked populations similarly. The percent novel germplasm relative to the single cross to be improved w as positively correlated with the estimates of favorable alleles excep t when NI was used as the estimator. The relationship estimators agree d with the genetic constitution of the donor populations. Strong posit ive correlations existed between diversity, based on the relationship rankings, and all the estimator rankings, except NI. Potential donor p opulations were effectively identified by LPLU, UBND, PTC, and TCSC. N I was not a good estimator of unique favorable alleles.