Genetic diversity and its relationship to hybrid performance in maize as revealed by RFLP and AFLP markers
--Ajmone-Marsan, P, Castiglioni, P, Fusari, F, Kuiper, M*, Motto, M
*Keygene n.v., P.O. Box 216, Wageningen, The Netherlands

In maize, the prediction of hybrid performance is of considerable importance and has attracted large interest over the years. Recently, genetic linkage maps based on molecular markers have been constructed in this crop (Coe et al., MNL 69:191-267, 1995), with the hope that they will provide effective means for predicting hybrid performance and heterosis. In the present study we surveyed genetic divergence among 13 inbred lines of maize using DNA markers and assessed the relationship between genetic distance and hybrid performance in a diallel set of crosses between them. The parental lines were assayed for DNA polymorphism using 135 restriction fragment length polymorphisms (RFLPs) and 209 amplified fragment polymorphisms (AFLPs).

A total of 508 RFLP bands were detected when considering 13 inbreds tested with all probe-enzyme combinations. Of the 149 probe-enzyme combinations used in this study, 135 (91%) revealed polymorphism across the 13 inbreds assayed. The majority (72%) of the polymorphic probe-enzyme combinations gave single-banded RFLP patterns. The remaining yielded multiple-banded RFLP patterns, suggesting the presence of repeated binding sequences in the genome for the corresponding DNA clones. The number of RFLP variants per probe-enzyme combination ranged from 2 to 5 in the former case and from 2 to 9 in the latter case, with an average of 3.31 and 4.97, respectively.

For AFLP analysis, a total of six primer combinations was used to assay the 13 inbreds. These permitted the production of approximately 500 selectively amplified DNA fragments ranging in size from 60 to 600 nucleotides and the identification of 209 polymorphic markers. On average 30-120 distinguishable bands were observed after selective amplification with each primer combination, and an average of 34.8 of these AFLP bands were found to be polymorphic among lines with a range from 19 to 52. These results suggest that AFLPs are able to detect a larger number of polymorphisms in a more efficient way in comparison to RFLPs, due to the much higher number of loci assayed in a single multiplex PCR reaction.

Estimate of genetic distances (GDs) between lines from BSSS, LSC and miscellaneous heterotic groups calculated with RFLP and AFLP markers gave almost identical mean GD values (45.9 versus 46.2). Moreover, GDs based on AFLP data had a similar range (19.5 - 63.7) to the range of GDs calculated from RFLP data (24.1 - 59.9). For both molecular markers the subset means for GDs, were, as expected, significantly greater for combinations of lines of different origin (51.8 and 50.7) than for the BSSS x BSSS (34.7 and 36.4) and LSC x LSC (31.1 and 37.2).

The dendrograms from UPGMA cluster analyses (Rohlf, New York Exeter Publ., 1989) of GDs based on RFLP and AFLP data are presented in Figure 1. Both have a high cophenetic coefficient (r = 0.91 and 0.90 respectively for RFLPs and AFLPs) and therefore show a good fit with GD values. Clustering based on RFLP data resulted in two major groups (Fig. 1a). One main cluster was comprised of lines derived from or related to BSSS along with H55, while the other was composed of 5 LSC related inbreds along with Pa91. The AFLP-based dendrogram assigned the 13 inbreds to three major groups (Fig. 1b): i) the BSSS-related lines; ii) H55 along with Pa91; and iii) the Lancaster lines. In addition, when compared to the RFLP-based dendrogram, discrepancies in forming subgroups within the major groups were noted. Thus, in the dendrograms obtained from cluster analysis, all lines with defined affiliation to one of the heterotic groups were assigned to the respective main clusters. These results suggested that RFLP and AFLP data clearly separated lines from the BSSS and LSC heterotic groups and detected pedigree relationship among inbreds.

A particular use of genetic markers is the prediction of hybrid performance. These results are consistent with experimental results of previous studies and quantitative genetic expectations.

The estimates of simple correlations (r) of GDs with F1 performance for grain yield (F1P) and SGD with SCA effects are presented in Table 1. The correlation coefficients of GDs calculated for RFLP and AFLP data with grain yields for the entire set of 78 hybrids were highly significant (P( 0.01) but only of moderate size. The r value was 0.36 for the GD based on RFLPs and 0.51 for the GD based on AFLPs. By contrast, for both classes of molecular markers a lack of relationship was noted between these two variables in the three subsets of crosses. It must be emphasized however, that the results obtained from the BSSS x BSSS and LSC x LSC group of crosses should be interpreted with caution because they are based on a small number of crosses. Estimates of r values between specific genetic distances (SGDs) and specific combining ability (SCA) effects were, with both class of markers, for all crosses and individual subsets of crosses positive and in general significant. In particular, a high correlation between the two variables was obtained for the entire set of 78 crosses (0.65 and 0.72), and in the BSSS x BSSS subset for both classes of molecular markers (0.72 and 0.81), whereas in the LSC x LSC subset a relatively high correlation (0.66) was reported only for AFLPs. In addition, significant correlations, although of moderate size, were found also in the subset of unrelated lines (0.38 and 0.47) for RFLP and AFLP, respectively. Finally, it was worth noting that correlation between GD and SGD calculated from AFLP data with F1P and SCA effects were higher than those based on RFLPs.

In summary, results from this study suggest that molecular marker based analysis, and in particular AFLP technology, offers a reliable and effective means of assessing genetic variation and of studying relationships among currently and historically important maize inbred lines. This may provide an alternative way for predicting performance and heterosis of maize hybrids. In particular, correlations between AFLP markers and SCA estimates may have a practical utility in predicting hybrid performance.

Table 1. Simple correlations of genetic distance (GD) and specific genetic distance (SGD) based on RFLP and AFLP data respectively, with F1 performance (F1P) and specific combining ability (SCA) of grain yield for all crosses and in different subsets of maize crosses.


Please Note: Notes submitted to the Maize Genetics Cooperation Newsletter may be cited only with consent of the authors

Return to the MNL 72 On-Line Index
Return to the Maize Newsletter Index
Return to the Maize Genome Database Page