Correlation between clone values in the germplasm collection and hybrids produced by these clones at CEPEC, with special emphasis on the yield x vigour relationship

The prediction of the general combining ability (GCA), from the performance of accessions per se in germplasm collections, and the methodologies of analyses that can improve the efficiency of evaluation, are important points to be investigated in the optimisation of breeding programmes. Aiming to understand these aspects, data from 544 accessions in CEPEC's germplasm collection were analysed and compared with GCA values calculated from eight progeny trials (one diallel and seven NCII designs). The accessions were each represented by one row of 10 trees, except in the case of 21 accessions that were represented by two rows, each of 10 trees. The following variables were observed for mature trees over six years in the collection, and over one to eight years in the hybrid trials: wet bean weight per tree (WBW), number of healthy pods (NHP) and trunk sectional area (TS). Parent-offspring correlations were calculated based on the clonal performance in the collection and the value of the clones as parents (GCA), in several sets of data obtained in the different hybrid trials. The average coefficients of linear correlation between the replicate data, obtained from the accessions that were repeated twice in the collection, were significant for WBW and for NHP, and the correlations were slightly improved when the yield data were corrected using TS as a co-variable or as a block effect, which indicates that the correction reduces the fixed effects of the environment. In general, highly significant parental and interaction effects were observed in the progeny trials. However, parent-offspring correlations were non-significant for WBW and NHP for relatively uniform parental clones (e.g. local Comun clones). However, positive parent-offspring correlations were often observed for genetically variable clones, but only when NHP data from the parental clones were corrected by using TS. The same tendency was observed for WBW, again when corrected using TS, though this was less clear than for NHP. It is concluded that yield data, corrected for vigour, might represent a more stable trait in germplasm collections than yield itself. Furthermore, NHP data of parental clones that are corrected by TS appear to be a more reliable predictor of the number of pods produced by the progenies than uncorrected parental data. (Résumé d'auteur