Comparative analysis of genetic similarity among maize inbred lines detected by RFLPs, RAPDs, SSRs, and AFLPs
--Pejic, I1, Ajmone-Marsan, P, Morgante, M2, Kozumplick, V3, Castiglioni, P, Taramino, G4, Motto, M

1Dipartimento di Produzione Vegetale e Tecnologie Agrarie, Universit? di Udine, Via delle Scienze 208, 33100 Udine, Italy. Permanent address Faculty of Agriculture University of Zagreb, Dept. of Plant Breeding, Genetics and Biometrics, Svetosimunska 25, HR-10000 Zagreb, Croatia

2Dipartimento di Produzione Vegetale e Tecnologie Agrarie, Universit? di Udine, Via delle Scienze 208, 33100 Udine, Italy

3Faculty of Agriculture University of Zagreb, Dept. of Plant Breeding, Genetics and Biometrics, Svetosimunska 25, HR-10000 Zagreb, Croatia

4DuPont Agricultural Biotechnology, Delaware Technology Park, Suite 200, 1 Innovation Way, PO Box 6104, Newark, DE 19714-6104

Information about germplasm diversity and the relationships among elite breeding materials has a significant impact in the improvement of crop plants. DNA-based fingerprinting technologies have been proven useful in genetic similarity studies. Among them, RFLP was the first and is still the most commonly used in the estimation of genetic diversity in plant species. The recently developed PCR-based marker techniques, which include RAPDs, SSRs and AFLPs, are playing an increasingly important role in this type of investigation. Here we report a direct comparison of DNA based techniques in reference to their informativeness and applicability for genetic diversity study using a set of 33 maize inbred lines.

The inbreds were surveyed for polymorphism with the four different marker systems. All of the molecular markers used in this study were able to uniquely fingerprint each of the inbred lines. The total number of assays ranged from only 6 primer combinations for AFLPs to 94 probe/enzyme combinations for RFLPs. The total number of polymorphic bands identified ranged from 90 for RAPDs to 255 for RFLPs. An average number of 4.8 alleles per locus with an average effective number of 3.2 alleles per locus ranging from 1.2 to 6.5 could be distinguished for each probe/enzyme combination using RFLPs. This value increased to 6.8 with SSRs, with an average number of effective alleles of 4.4 per locus, ranging from 1.1 to 6.6, while for RAPDs and AFLPs these values were lower (1.6 for both). This was reflected also in lower expected heterozygosity values. Overall the highest assay efficiency index was observed for AFLPs (45.7) and the lowest for RFLPs (3.2). RAPDs and SSRs (5.8 and 4.2, respectively) were comparable to RFLPs. In particular, for AFLPs the high assay efficiency index is due to the simultaneous detection of several polymorphic bands in a multiplex amplification per single reaction.

The genetic similarity trees produced from each marker system showed that inbreds were ordered in the four trees, as expected, into major groups of lines derived from BSSS and LSC, although discrepancies in forming subgroups within the major groups were observed as well as in the clustering of inbred lines of miscellaneous origins. Considering the BSSS-related lines the topology of each tree is unique with some evident similarity: the clustering of B14, B37 and B73 types is in general fully conserved across the four trees. On the LSC side cluster of C103 and Mo17 related lines were consistently reported for all methods with the exception of Va22, derived from C103, that for all methods was aggregated with lines of different origins. The Oh43 related lines (Oh43 and A619) were grouped with Lancaster only by RAPDs and AFLPs, while SSRs and RFLPs clustered these with BSSS lines; Oh43 is usually considered a Lancaster type. It is also interesting to note that similarly to the RAPD-based trees, the clustering based on AFLP data produced a tree with a relatively narrow range of similarity values between the more related and the more distant pairs of inbreds. However, all the main clusters within the set of inbreds herein studied were confirmed by clustering based on AFLP data. Four pairs of most similar inbreds (Lo932, Lo944, B14A, CM109, Lo916, Lo999, H55, H96) were clustered together as in the SSR tree, and in two cases (A619, Oh43, and H55, H96) as in RFLP and RAPD trees.

The estimates of correlation coefficients (r's) among available coancestry coefficients (f's) and genetic similarity (GS) data obtained from the four molecular marker systems showed that all r's were highly significant (P < 0.01). It is worth noting that RAPDs showed the lowest correlation (r = 0.40) with f values, while AFLPs showed the highest value (r = 0.62). The r's among similarity data obtained with the different molecular marker techniques were also significant. Correlation coefficients of RAPD marker data with those obtained using other marker systems were lower than those among similarity estimates based on AFLPs, RFLPs, and SSRs. The extent to which similarity values were correlated varied considerably across the whole data set. When the whole set of pairwise data (528) was divided in two groups (according to the arithmetic mean of the observed GS range based on RFLP data), "more similar" lines (GS>0.37) and "less similar" lines (GS<0.37), it became apparent that genetic similarity estimated by different marker systems was mainly correlated only among similar lines, while the relationships among dissimilar lines were low and not significant. The GS values plotted against the estimate of coancestry value based on pedigree data followed the same pattern.

The cophenetic correlation coefficients provided for each marker system indicate the extent to which the clustering of genotypes depicted in the trees accurately represents the estimates of genetic similarity between inbreds obtained with that marker system. Overall the cophenetic coefficients were medium to high, with the RFLP (0.84), and AFLP (0.83) data resulting in the highest correlations and the RAPD (0.72) assay producing the lowest correlation.

All methods could clearly distinguish all 33 inbred lines, although the SSR data provided the highest level of discrimination between any pair of inbreds. In general, the grouping agreed with pedigree information of the lines, although some discrepancies were observed. In particular genetic similarities based on AFLP data had the highest correlation with pedigree data while those based on RAPDs had the lowest one.

In order to determine the sampling variance of genetic similarities calculated from different molecular marker data sets, bootstrap analysis with declining number of bands was performed. The relationships between number of bands and sampling variance of genetic similarity among all pairs of inbred lines for the four molecular techniques indicated that the standard deviation of the estimate was no longer significantly reduced when more than 150 bands were analysed because of a decreasing slope of the curve.

In conclusion the results of this study within a set of maize inbred lines, and the comparison between the methods employed, indicated that, with the exception of RAPDs, the other DNA markers provide consistent information for germplasm identification and pedigree validation. We have shown that SSR and AFLP profiling technologies can be good candidates to replace RFLP markers in genetic similarity estimates and variety description, and that they have comparable accuracy in grouping inbred lines selected by pedigree. They are generally much simpler to apply and more sensitive than the traditional morphological and biochemical methods or the RFLP-based fingerprinting techniques because they are more efficient in detecting polymorphism; yet they are generally correlated with RFLP analysis. A major advantage of the SSR and AFLP methods is that they can be automated and so have great potential in large-scale population genetics and plant breeding. While SSRs, thanks to their multiallelism and codominance, appear to be especially suited for the analysis of outcrossing heterozygous individuals, AFLPs, thanks to their high multiplex ratio, offer a distinctive advantage when genome coverage is a major issue due to the presence of linkage disequilibrium such as in inbred lines and breeding materials.

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