Circulant partial diallel in an interpopulation of maize: efect of the number (s) of crosses.

Author(s)

Sandro Ricardo Fuzatto

• Document type: Doctoral Thesis
• Press: Piracicaba.
• Institution: Universidade de São Paulo (USP). Escola Superior de Agricultura Luiz de Queiroz (ESALA/BC)
• Defense date: 2003-04-15

Examining board members

Jose Branco de Miranda Filho; Edward Madureira Brasil; Antonio Augusto Franco Garcia; Eliana Patricia Gorgulho; Pedro Hélio Estevam Ribeiro

• Advisor: Jose Branco de Miranda Filho

Abstract

In the partial mating scheme, each line of one population is crosses with a number (s) of lines of the opposite population. The objective of this work was to study the effect of the number s in the estimates of general combining ability (GCA) and specific combining ability (SCA) of S1 lines from two contrasting populations (GN-03 and GN-04) of maize (Zea mays L.) crossed according to the partial diallel. Two groups (GA and GB) of fifth S1 lines were randomly crossed (GN-03 x GN-04) using s=6. From the 300 possible crosses in each group, some were last resulting 297 in GA and 282 in GB. Crosses were evaluated in three locations (Piracicaba-SP and Uberlandia-MG with three replications; and Jatai-GO with two replications); group GB was evaluated only in the first two locations. The whole set of crosses was divided into six experiments for each group. The analyses of variance were performed for each experiment and then grouped over experiments. The traits ear yield (t/ha) and plant height (cm) were analysed. The analyses were performed for varying number of crosses per line; i.e., for s=3, 4, 5, 6. Ear yield of S1 xS1 crosses in both groups varied from 94% to 99% in relation to the mean of both checks in Uberlandia and Piracicaba. In Jatai, only group GA in midseason planting yielded 17% more than the average of checks. Specific combining ability didn’t showed significance for every traits and locations. While general combining ability was always significant. Decreasing s led to a higher variation in the estimates of GCA and also to a decrease in the precision of SCA estimates. The coefficient of correlation between observed means ( ij ij j i ij e + s + g + g + = Y m ) and predicted means through the reduced model ( j i ij gˆ + gˆ + ˆ = Y ˆ m ) was calculated for boths groups in all locations. The higher coefficients of correlations were observed for smaller s values, leading to a miss interpretation. That SCA is of low importance in the phenotypic expression. Consistent results were observed by using s=5 allowing the detection of significance and good of the estimates of GCA and SCA. The estimates of the interpopulation variance components were obtained for ear yield using s=6. The estimates of the additive genetic variance expressed in (g/plant)2 were 120.56, 61.92 and 38.44 in Piracicaba, Uberlandia and Jatai, respectively. The dominance variance were higher than the additive variance in all locations; the ratios ) / ( 2 A 2 D s s were 6.1, 4.3 and 7.2, respectively, indicating the importance of the non additive effects in the interpopulation. (AU)