Accessions from the germplasm bank were grouped by race and/ or adaptation and evaluated for agronomic and plant and ear traits. The data were analyzed using multivariate duster analysis. The best 20% of the entries in each duster, based on agronomic potential (Taba et al. 1998), were used to create a breeding core subset for that group of material. The breeding core subset limits the number of entries for further evaluation while maintaining the genetic diversity present in the germplasm bank. The subset was evaluated with testers from both heterotic patterns ( one flint and one dent tester) and then divided into two groups, one for each heterotic pattem, following the prindples of Dudley's theories (1982,1984). The single crosses from the two groups (flint donor x flint tester and dent donor x dent tester) were then crossed to another elite inbred of the same heterotic pattem to form triple cross enhanced germplasm containing 25 % accession and 75% elite germplasm. Backcrossing to an elite inbred is also important, since enhanced material undergoes relatively limited selection before being incorporated into the breeding poo1s, and it is necessary to prevent dilution of favorable attributes of the pool with unfavorable a1leles present in the donor populations. The use of two different elite inbreds provided greater potentia1 for recombination of favorable a1leles from different germplasm sources. Selection for important agronomic traits (plant height, ear height, etc.) was performed during the development of S2lines from the enhanced germplasm, which were then testcrossed to elite testers or the breeding pool from the opposite heterotic group and evaluated in at least two locations for several plant, ear, and agronomic traits. The best entries were included in evaluation trials of the breeding poo1s into which they were to be incorporated. A description of these trials follows. 51 and 52 lineswere formed in each population, under selection for ear and plant traits and resistance to insects and pathogens. Genera1ly, 300 to 400 So families were se1fed and 600 to 800 51 ears were selected and planted ear-to-row, wi!4 300 to 400 52 ears selected and planted ear-to-row for testcross formation. These 52 lines were crosseato testers of the opposite heterotic group and evaluated for: agronomic traits such as: yield and agronomic aspect; plant traits such as days to anthesis and silking; and ear traits such as ear rot and ear aspect. Usua1ly 250 to 350 entries were evaluated in the trial (plus new enhanced germplasm) and 100 to 125 were selected for recombination in the next cycle. Multivariate duster analysis was performed on the plant and ear trait data to create dusters representative of the genetic diversity within the pool, and selection of the best entries in each duster was based on agronomic performance. Previously, duster ana1ysis was performed using the data from each environment separately, but in future the ana1ysis wi1l be performed using data combined across environments, with each trait in each environment representing a separate variable in the ana1ysis. The recombination was performed in a diallel format with forced recombination between the different dusters. For example, pollen from the male rows representing dusters "b" and " c" was bu1ked by cluster and used to po1linate individual plants in female rows representing duster " a" i pollen from the male rows representing dusters II a" and I' c" were bu1ked by duster and used to po1linate individual plants in female rows representing duster Ilbll i and so on.

Text in English

0311|AGRIS 0301|AL-Maize Program

CIMMYT Staff Publications Collection