Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (25)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Agricola
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Related Collections
Right arrow Cell Biology & Molecular Genetics
Right arrow Crop Genetics
Right arrow Plant Genetic Resources
Crop Science 42:1832-1840 (2002)
© 2002 Crop Science Society of America

CROP BREEDING, GENETICS & CYTOLOGY

Genetic Characterization of CIMMYT Inbred Maize Lines and Open Pollinated Populations Using Large Scale Fingerprinting Methods

Marilyn L. Warburton*,a, Xia Xianchuna, Jose Crossaa, Jorge Francob, Albrecht E. Melchingerc, Matthias Frischc, Martin Bohnc and David Hoisingtona

a CIMMYT, Int, Applied Biotechnology Center, Apdo. Postal 6-641, 06600 Mexico, D.F., Mexico
b Facultad de Agronomia, Universidad de la Republica, Ave. Garzon 780, CP 12900, Montevideo, Uruguay
c Institute of Plant Breeding, Seed Science and Population Genetics, Univ. of Hohenheim, 70593 Stuttgart, Germany

* Corresponding author (mwarburton{at}cgiar.org)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
The International Maize and Wheat Improvement Center (CIMMYT) currently holds about 17 000 samples of maize (Zea mays L.) and teosinte (Z. mays, several subspecies), a wild relative of maize. Seven CIMMYT populations and 57 inbreds were characterized by simple sequence repeat (SSR) markers. SSRs chosen from almost every bin in the maize genetic map were tested for repeatability, ease of automation in allele calling, and discrimination (information content). Eighty-five SSRs were found to be repeatable and easily automated, and were run on all material in the study. Fifty-three of these SSRs were found to be the most discriminatory and will be used in future routine fingerprinting studies. The seven breeding populations clustered as would be predicted on the basis of pedigree and heterotic grouping. Genetic diversity within each population was significantly higher than diversity between populations, indicating that the populations are heterogeneous at the molecular level. The inbreds also showed a high level of genetic diversity, indicating that CIMMYT breeders have successfully incorporated considerable genetic diversity into CIMMYT maize germplasm. Only lines closely related by pedigree clustered together. Population of origin and heterotic grouping were not associated with the clusters formed on the basis of SSR markers, a result consistent with the high level of diversity within source populations of the inbreds. Although this will make it more difficult to assign CIMMYT inbred lines to currently existing heterotic groups by means of markers, the markers may be useful in refining the CIMMYT heterotic groups into additional and more uniform groups.

Abbreviations: CIMMYT, International Maize and Wheat Improvement Center • CML, CIMMYT maize inbred line • SSR, simple sequence repeat • RFLP, restriction fragment length polymorphism • PIC, polymorphic information content


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
GENETIC FINGERPRINTING of maize germplasm is being undertaken at CIMMYT to aid breeders in the placement of breeding lines and populations into the correct heterotic group, to aid in the curation of the CIMMYT genebank collection by refining the core subsets formed from field evaluations, and to understand the evolution of major tropical maize races better. An efficient method for large scale fingerprinting of maize germplasm is necessary, considering the size of the collections of breeding lines, populations, pools, and races produced and maintained by CIMMYT and its partners.

Maize yields have been dramatically boosted by the use of hybrids in many parts of the world. Tropical maize germplasm has not been as fully classified into heterotic groups as has U.S. Corn Belt germplasm, and pedigree information is not available for some of the lines and populations used in tropical maize breeding. Past studies have used restriction fragment length polymorphism (RFLP) markers to place temperate lines into known heterotic groups with considerable success (Dubreuil et al., 1996; Messmer et al., 1992, 1993). RFLPs and microsatellites (SSRs) have also been used to assign tropical Asian maize germplasm to heterotic groups, but these groups are still being verified by traditional field crosses (Yuan et al., 2000).

Genetic diversity in the world's major cereal collections is critical as a resource to find new alleles that will improve yield to fight world hunger. Unadapted and wild relatives contain untapped genetic resources for biotic and abiotic stress resistance (Hoisington et al., 1999). Even unadapted parents with poor phenotypes can contribute favorable alleles to their progeny, when these alleles are placed in an adapted background. Therefore, screening based on phenotype may miss much favorable variation. New allelic variation may be identified by means of markers, following which the contribution of these new alleles can be measured phenotypically. Screening must be done (or complimented) with screening based on genotype (Tanksley and McCouch, 1997).

Much of the CIMMYT tropical maize germplasm from the West Indies and the Americas has been organized into core subsets to aid in characterization and utilization of this material by breeders. Accessions from the CIMMYT Maize gene bank were placed into subsets according to classification on the basis of phenotypic traits (Crossa et al., 1995; Taba et al., 1998). It is possible to refine the composition of these core subsets by molecular markers, or a combination of molecular markers and phenotypic field data (Franco et al., 1998, 2001).

To begin to classify a collection as large as the CIMMYT maize germplasm collection and the hundreds of breeding lines and populations created by CIMMYT breeders, very efficient marker protocols must be in place and tested. Dillman et al (1997) suggested the use of RFLPs, but these markers are slow and expensive to run on such a large scale. Microsatellite markers, or SSRs, have been suggested in other studies, and good correlations have been found between SSR and RFLP diversity and pedigree-based measurements (Pejic et al., 1998; Smith et al., 1997). In most cases, SSRs have the added advantage of having been mapped, so the genome can be uniformly sampled in SSR-based genetic classification. The efficiency of SSRs can be further increased by running multiplexed reactions under automatic electrophoresis conditions, as suggested by Mitchell et al. (1997). Possible impediments to the automatic scoring of SSR markers would be those markers that do not follow a stepwise mutation pattern, i.e., fragment size differences that are not a multiple of the repeat unit. It has been shown in Drosophila (Colson and Goldstein, 1999) that only 7 of 17 SSRs showed stepwise mutation patterns. The majority of the remaining 10 SSRs showed small insertions or deletions (indels) and two showed very large indels. The non-stepwise changes in SSRs will affect identification of alleles and should be examined for use in diversity studies. By means of SSRs, it may be possible to compare diversity studies done in different laboratories with the same SSR markers under standard conditions and by including a few standard genotypes across all laboratories.

The objectives of this study were to: (i) determine the minimum number and identity of the most suitable microsatellites to be used in large-scale tropical maize germplasm characterization projects and (ii) analyze the genetic diversity patterns of seven tropical maize populations and 57 inbred lines by the markers determined in Objective (i).


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Plant Material
DNA was extracted from 42 maize inbreds that have been registered and released as CIMMYT Maize Lines (CMLs), and 15 that have not yet been named and released (Table 1) . The inbreds span a broad range of genetic diversity of CIMMYT tropical breeding lines, including sister lines from the same cross and with the same selection history and lines selected from separate populations unrelated by pedigree. The majority of the lines have been assigned to heterotic groups on the basis of testcross data. DNA was also extracted from individuals representing seven breeding populations (Table 1) that have all been assigned to heterotic groups. Forty-eight individuals were chosen to characterize the diversity present in each population, as this is a convenient number for working in microtiter plates of 96 or 384 wells. Crossa et al. (1993) showed that the sample size (n) required to retain, with probability P at least one copy of each of k allelic classes in each of m loci should be larger than:


View this table:
[in this window]
[in a new window]
  		Table 1. Inbred lines and populations assayed using 85 SSR markers and their respective pedigree information are listed in the first three columns. Populations and pools were assayed by fingerprinting 48 individuals per population.{dagger}

 
With 48 individuals, for m = 5 loci with k = 5 alleles per locus there is a 95% probability of detecting all alleles with a P = 0.05 or greater. With only 24 individuals (also convenient for work in microtiter plates), only alleles of P = 0.12 or greater can be detected at this level of probability. This formula shows that the required sample size is affected more by low allele frequency than by the number of alleles per locus or the number of loci. Therefore, alleles at very low frequency may not be accurately detected without the use of extremely large samples. DNA was extracted from the youngest fully mature leaf of 4- to 6-wk-old seedlings with a sap extractor (MEDU Erich Pollahne; Am Weingarten Germany) according to the Applied Biotechnology Center's Manual of Laboratory Protocols (CIMMYT, 2001) and modified from Clarke et al. (1989).

SSR Primers
Each SSR locus was first amplified separately, and then amplified with other SSR loci under multiplex conditions to find the optimal amplification and electrophoresis conditions. SSR markers were chosen from the MaizeDB database (http://www.agron.missouri.edu/ssr.html, verified June 4, 2002) on the basis of bin location (to maximize genomic coverage) and repeat unit. Dinucleotide repeats were avoided in most cases because of the difficulty in accurately sizing alleles that differ by only two bases. Fluorescent oligonucleotides were bought from commercial companies (Operon Technologies, Inc., Alameda, CA, or GIBCO BRL Life Technologies, Inc., Burlington, ON) and forward primers were labeled at the 5' end with either 6-carboxyfluorescein (6-FAM), tetrachloro-6-carboxyfluorescein (TET), or hexachloro-6-carboxyfluorescein (HEX). Multiplexed PCR reactions were performed in 10-µL volumes containing 2 µL of template DNA (output of the sap extractor diluted 5x with distilled, deionized H2O), 0.4 to 4.0 pmols each of 1 to 4 primers, 1x PCR buffer, 0.25 mM dNTPs, 1.5 to 2.5 mM MgCl2 and 0.75 U Taq polymerase. The reactions were performed with a Peltier Thermal cycler (MJ Research, Inc., Watertown, MA) using the amplification conditions of 94°C for 2 min; followed by 30 cycles of 94°C for 30 s, X°C for 1 min, and 72°C for 1 min; followed by extension at 72°C for 5 min. X°C refers the annealing temperature, which is specified for each primer in Table 2 .


View this table:
[in this window]
[in a new window]
  		Table 2. SSR markers used in the study. Markers determined to be the most discriminatory are shown underlined. Annealing temperature is in °C.

 
Electrophoresis
Samples containing two PCR reactions (0.5 µL of each), 0.3 µL GeneScan 350 or 500 internal lane standard (Applied Biosystems, Foster City, CA) labeled with N, N, N, N-tetramethyl-6-carboxyrhodamine (TAMRA), and 30% (v/v) formamide were heated at 95°C for 5 min, placed on ice, then loaded on 4.5% (w/v) denaturing (6 M urea) acrylamide:bisacrylamide (29:1) gels (36 cm well-to-read). By multiloading two multiplexed PCR reactions, on average, five SSR markers were run in each lane of the gel simultaneously. DNA samples were electrophoresed in 1x TBE buffer (pH 8.3) at constant voltage (3.00 kV) for 2.5 h on an ABI Prism 377 DNA Sequencer (Perkin Elmer, Foster City, CA).

Data Entry
Fragment sizes were automatically calculated with GeneScan 3.1 (Perkin Elmer/Applied Biosystems) using the Local Southern sizing method. The GeneScan data were appended to a table with Genotyper 2.1 (Perkin Elmer/Applied Biosystems), and then exported as an Excel file recording peak size for each individual. Peak sizes were converted to the proper configuration for subsequent analysis (0,1 binary matrix). For population analysis, binary data from the 48 individuals in each population were converted to allele frequencies for each population.

Peak sizes were converted to alleles by creating categories in Genotyper, which combines peak sizes within a predetermined range into the same allele, and thus takes into account error during size calling. Eight primers were chosen for a repeatability test to indicate the error rate in size calling. SSRs with 2, 3, and 4 base repeats were tested by amplifying them singly or under multiplex conditions on 10 inbred lines four separate times (same DNA was amplified and electrophoresed and the alleles called by GeneScan). All primers in this study were tested for repeatability by amplifying them under multiplex conditions at least four separate times on two inbred lines.

Alleles of the SSRs were assumed to increase in size in a stepwise manner in increments of the number of base pairs that reflected the repeat unit. SSRs whose polymorphisms did not follow stepwise mutation patterns and would be consistent with indels of greater than one base pair were eliminated at a very early point in the study because they could not be scored automatically by Genotyper. The remaining SSRs that followed, for the most part, a stepwise polymorphism pattern were scored as though all polymorphisms followed repeat units. Polymorphisms generated by indels were discounted, since an indel of only one or two base pairs would be equal to or smaller than the error associated with allele calling for most of the primers, and so would not preclude the correct scoring of polymorphisms because of differences in number of repeats. Very large indels in the amplified region would create fragments too far out of the expected range of sizes, and would most likely be scored as missing data. This would cause a loss of information, but unless all fragments outside the expected size range were sequenced, it would not be possible to determine whether they were artifacts or contaminants. Indels of numbers of bases equal to the repeat unit (or equal to the repeat unit plus or minus error associated with allele calling) would cause different polymorphisms to be called the same allele, which again would cause a loss of information that could not be avoided without sequencing. However, the error of calling two alleles the same, when they are actually different, is less grave an error than calling two alleles different, when they are actually the same. The error associated with assuming two alleles of the same size are identical by descent when one was generated by an indel and the other by the addition (or deletion of) repeat units should be minimized if sufficient numbers of markers are used.

Selecting the Minimum Subset of SSRs
All individuals in the study were genotyped with a total of 104 SSRs. Basic information from each primer including polymorphism information content (PIC) (Powell et al., 1996) and percent heterozygosity in the inbreds was calculated. A high percent heterozygosity in the inbreds for one or a few primers indicated a problem with the primer, such as uncorrectable stuttering or amplification of a second locus with the same approximate size. Eleven SSRs were removed from the study for these reasons and an additional eight were removed because polymorphisms did not follow the repeat unit and it was not possible to assign these amplification products automatically to the proper allele category by means of Genotyper.

To determine if the number of markers could be reduced in future studies, loci with polymorphisms that followed the repeat unit and with zero (or a very low) heterozygosity in the inbreds were analyzed retrospectively. This was done to determine the minimum number of markers necessary to describe adequately the variation present in the study and to identify the most discriminating (informative) markers (Franco et al. 2001). This analysis was done separately for the inbred lines and for each population to see if there was agreement in the identity and number of markers chosen in each case.

The retrospective analysis classified the individuals using all the alleles of the SSRs used in the study. The first step was to estimate the optimum number of groups (clusters) present in the study by means of the fusion values obtained from the Ward method and the upper-tail rule (Mojena, 1977). The AMOVA procedure (Excoffier et al., 1992) was then used to estimate the "within" and the "among" cluster variance components for different numbers of clusters to determine the number of clusters that produced the largest increment in the among-groups variance (largest reduction in the within-group variance) and/or the maximum average distance among the groups. The generalized linear model (McCullagh and Nelder, 1983) and the GENMOD procedure of SAS (1993) were used to determine the level of significance of each fragment produced by the SSRs in the study. This method ranked the fragments by their ability to discriminate among clusters. Finally, the individuals were again clustered on the basis of a reduced number of fragments, selected by their significance level, to find the minimum number that could be used to generate the same (or very similar) classification as that obtained using all the fragments.

Analysis of Genetic Diversity
For diversity analysis, a matrix of binary or allele frequency data was constructed with columns equal to genotypes (in the case of the inbred lines) or populations, and rows equal to distinct molecular marker fragments (alleles of each primer). For the 57 inbreds, the body of the matrix contains zeros and ones, corresponding to the absence or presence, respectively, of each fragment in each genotype. For the seven populations, the body contains the frequency of each allele in each population. For inbreds, similarity matrices were constructed from the binary data by the Simple Matching similarity coefficient (Kaufman and Rousseeuw, 1990). This is the coefficient that is used in the retrospective analysis to find the minimum subset of markers because it has Euclidian metric properties. Jaccard's and Dice similarity coefficients were also calculated, but were found to give the same results, as correlations between matrices [calculated by the MXCOMP procedure of NTSYSpc 2.01 (Rohlf, 1997)] were significant (rjaccard's vs. simple matching = 0.95**; rjaccard's vs. dice = 0.99**; rdice vs. simple matcing = 0.95**). Therefore, only the Simple Matching coefficient was used to create dendrograms. Matrices of genetic distance between populations were calculated by means of Nei (1972), Roger's, and Modified Roger's coefficients (SAS, 1993, Cary, NC). Standard deviation was calculated for between population distances, and standard error was calculated for within population distances to test the significance of the differences.

Dendrograms of the inbred lines were constructed from the similarity matrix by the UPGMA method (Rohlf, 1997) to visualize the patterns of diversity in the set of lines, and a bootstrap analysis was run by SAS (1993) to calculate the confidence intervals associated with each cluster. For the populations, dendrograms were created with the Ward and UPGMA clustering methods of NTSYS.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Repeatability Test
The eight primers run in the repeatability test showed slightly different amplification patterns when run singly or in the multiplex conditions under which they were optimized. When run singly, additional bands could be seen in many of the primers (data not shown). This indicates that the conditions must be the same in all amplifications, and SSRs run in multiplexes in every case. The PCR reaction is highly competitive, and if fewer primers are competing for resources such as Taq enzyme or dNTPs, a duplicate, but not perfect, primer site may be allowed to amplify in the genome. When multiplex conditions were used, there was some error associated with the four amplifications of each of the individuals in many of the primers. Interestingly, as the repeat unit increased in size, so did the error associated with the alleles; however, this increase was not linear. The error associated with a two base repeat was calculated to be from 0.85 to 1.0 base pairs; for a three base repeat it was on average 1.25; for a four base repeat it was 1.50; for a five base repeat it was 2.00; and for a six base repeat it was 2.50. Therefore, the ambiguity associated with allele calling decreases markedly as the repeat unit increases. It must be noted that if the error for two-base repeat SSRs is close to one, this SSR cannot be easily scored, as many of the bands will fall equidistant to two alleles in size and its true identity cannot be determined. Only two dinucleotide SSRs were successfully used in this study; all other SSRs were trinucleotide or greater.

Allelic Diversity
The 85 SSR loci in this study amplified a total of 416 bands in the inbred lines, with an average number of 4.9 and a range of 2 to 14. In the populations, the 85 SSRs amplified a total of 531 bands, with an average number of 6.3 and a range of 2 to 16. Most previous studies of SSR diversity in maize revealed a similar allelic diversity in inbreds. For example, Lu and Bernardo (2001) report 40 U.S. maize inbreds averaged 4.9 alleles for 83 SSR loci. Senior et al. (1998) reported an average of 5.0 alleles. However, Pejic et al. (1998) reported an average of 6.8 alleles per 27 SSR loci in 33 U.S. maize inbreds, even higher than that reported in the CIMMYT populations. All marker data for inbred lines and populations have been submitted to the Maize-DB database.

Minimum Number of Markers
The retrospective analysis of the inbred lines used to choose the minimum number of markers identified 53 SSR loci having one or more alleles that were highly discriminatory and the most informative (data not shown). These 53 SSRs produced the same clusters as produced from the entire set of 85 in the retrospective analysis. The same analysis was run separately on each of the seven populations. Forty-six of the SSR loci were identified in common in most of the populations and the inbreds; 25 were identified as containing alleles that were discriminatory in only one or a very few populations and 14 of the loci were identified in both the populations and the inbreds as being noninformative (data not shown). This indicates that many of the markers were always informative, and a few of the markers were never ranked as highly informative in any of the analyses. Of course, the composition of the population under study influences which markers will be the most discriminatory.

The most discriminatory SSRs were not always those with the highest PIC. For example, Phi046 identified only two alleles, and had a PIC value of 0.46. However, it was identified as being discriminatory in the inbreds and in five of the seven populations. In contrast, Zcaa391 amplified 14 alleles and had a PIC value of 0.85, but was identified as discriminatory in only one of the populations and not in the inbreds. It is not possible to predict, therefore, which SSRs will be the most discriminatory on the basis of PIC alone. On the basis of the retrospective analysis, and for reasons of cost efficiency, it was decided that the 53 SSRs determined by analysis of the CMLs (which included almost all the 46 that were also identified in the seven populations) will be routinely used in future genetic diversity studies. These 53 markers still provide good coverage of the genome, with at least three and up to seven SSRs per chromosome (Table 2). Fingerprinting for proprietary protection of breeding lines could, however, be run with fewer SSRs, and if a particular data set of 53 SSRs shows poor resolution of patterns of diversity or low confidence of clustering when a bootstrap analysis is applied, up to 85 SSRs (or even more) could still be run.

Populations
Nei's genetic distance calculated on the allele frequencies of each of the seven populations created the dendrogram presented in Fig. 1 by the Ward clustering method. The Ward method combines the two clusters in each step whose fusion leads to the smallest increase in the Euclidean sum of squares within groups, thus leading to a maximized variance between groups and a minimized variance within groups. This was particularly important for the populations, since the variance within populations was already very high. Furthermore, clustering determined by the UPGMA method produced a very similar dendrogram, and is therefore not presented here. The populations clustered as would be predicted on the basis of pedigree and known heterotic group as defined by field evaluations using testers. Cluster 1 contains Populations 21, 22, 29, and Pool 24. Populations 21 and 22 are tropical, late white dent or semident maize types originated from Pool 24, which is from the Tuxpeño race of maize. All these populations belong to heterotic group A. Population 29 is also a tropical, late white dent maize, with both Tuxpeño and Caribe races in its background, and shows heterosis when crossed to both A and B testers, indicating it belongs to neither group. It is the farthest outlier in Cluster 1. Cluster 2 contains only Population 43, which is a mixture of La Posta elite lines, and is also tropical, late, white, and dent. It also belongs to neither heterotic group A nor B. Cluster 3 contains Populations 25 and 32, which are tropical to subtropical, intermediate to late white flints, and belong to heterotic group B.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 1. Ward's dendrogram of the seven populations in the study based on allele frequencies of 85 SSRs. The horizontal axis is expressed in genetic similarities (similarity = 1 - distance) and these were calculated by Nei's (1972) coefficient.

 
Although there were differences in the distances measured by means of the three coefficients, the relative variation within and between populations were the same in each case, and so only the Nei (1972) values and standard errors are presented in Table 3 . In every comparison, the distance is significantly higher within the populations than between them, indicating that the majority of the variation in these populations is partitioned within the populations. Although the populations were originally formed to be uniform for certain phenotypic attributes, at the molecular marker level they are very heterogeneous and are poorly distinguished from one another. Therefore, in the majority of the genome (excluding a few selected loci), these populations can be thought of as highly diverse. This may increase the genetic diversity available for selection to the breeders, but makes classifications based on markers difficult. A test cluster analysis and principal components analysis containing all 48 individuals from all seven populations was run to see if individuals clustered with their populations (data not shown). Not surprisingly, given the above results, there was no pattern discernable on the basis of population of origin of the lines, and two lines from different populations were just as likely to cluster together as were two lines from the same population based on SSR markers.


View this table:
[in this window]
[in a new window]
  		Table 3. Roger's distances between (lower diagonal), and within (diagonal, underlined) populations in this study. Standard deviations for the between populations distances are found in the upper diagonal, and the standard error within populations (used to calculate significance of the differences for within and between population distances) is found in the last row.

 
Inbred Lines
UPGMA was used to produce the dendrogram on the basis of 85 SSRs and 57 inbred lines (Fig. 2) , rather than Ward, since the purpose of the analysis of inbred lines was to study the overall pattern of genetic diversity in this subset of inbreds, and not to maximize distance between the Operational Taxonomic Units (as in the case of the populations). In this sample of inbred lines, certain sets of inbreds can be used as a control, as they are known to be the most highly related lines in the study by pedigree. The lines TS1 - 5 all cluster closely together, and come from the same segregating S1 segregating family. Furthermore, TS4 and CML344 are sister lines sharing the same selection history for seven generations, and the cluster even more closely together (Fig. 1). The LP lines 4 and 5 clustered very closely together with CML341, and these lines share seven generations of the same selection history. LP line 2 clustered together with CML 399, which has the same selection history as LP2 (again for seven generations). Other LP lines originated from the same population, but do not share the same pedigrees beyond the first segregating generation, and this is reflected in the fact that the LP lines formed two distinct clusters. In general, lines clustering together are closely related by pedigree, but in some cases lines that are related by pedigree do not cluster together, and may represent diversity due to divergent selection, random drift, or human error. Lines derived only from the same population and not necessarily the same cross do not cluster together in most cases. As we have seen with the populations, lines from the same population may be more different than lines from different populations, so it will be difficult, if not impossible, to correlate the populations from which CIMMYT inbred lines were selected with marker diversity unless the lines are highly related by pedigree. The markers may be better indicators of relatedness of the lines, in this case.



View larger version (47K):
[in this window]
[in a new window]
  		Fig. 2. UPGMA dendrogram of 57 inbred lines based on 85 SSRs. The horizontal axis is expressed in genetic distances that were calculated by the Nei and Li coefficient. Bootstrap confidence intervals are included at the junctions of each cluster.

 
No good association based on heterotic grouping as assigned by field evaluations using testers can be seen in this figure. Diversity tests using RFLPs in temperate maize have had much more success assigning lines to heterotic groups (for example, Smith et al., 1992; Messmer et al., 1992; Dubreuil et al., 1996). Furthermore, SSR marker comparisons of tropical maize germplasm that came from very diverse breeding programs, and had been selected as divergent heterotic groups, were able to differentiate the heterotic groups (Yuan et al., 2001). It may be that the method used to assign inbreds created by CIMMYT breeders to the proper heterotic group for hybrid breeding is not sufficiently discriminatory. Lines are currently assigned to heterotic group A or B by crossing to one or two tester lines and testing the performance of the hybrids in the field. According to the dendrogram in Fig. 2, it may be suggested that CIMMYT breeding lines fall into considerably more than only two heterotic groups. Furthermore, if the current heterotic groups contain a lot of variation, as can be seen on the basis of the dendrogram, one or two testers cannot be sufficient to represent the diversity present in the heterotic group, and thus may assign lines to the incorrect group. Including more tester lines would be prohibitively time consuming and expensive. It is therefore suggested that the marker data be used to refine the heterotic groupings of CIMMYT breeding lines. More tests, including hybrid field performance of heterotic groups predicted by markers, will be required to see if splitting the two current heterotic groups into subgroups creates more combinations of heterotic patterns that would be expected to show heterosis when crossed.


   ACKNOWLEDGMENTS
 
This study was completed with funds from the German granting agency Bundesministerium für Wirtschaftliche Zusammenarbeit und Entwicklung. Our sincere thanks to the excellent technical assistance of Salvador Ambriz, Leticia Diaz, and Emiliano Villordo.

Received for publication October 22, 2001.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Crop Sci.Home page
M. L. Warburton, J. C. Reif, M. Frisch, M. Bohn, C. Bedoya, X. C. Xia, J. Crossa, J. Franco, D. Hoisington, K. Pixley, et al.
Genetic Diversity in CIMMYT Nontemperate Maize Germplasm: Landraces, Open Pollinated Varieties, and Inbred Lines
Crop Sci., March 19, 2008; 48(2): 617 - 624.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
X. C. Xia, J. C. Reif, A. E. Melchinger, M. Frisch, D. A. Hoisington, D. Beck, K. Pixley, and M. L. Warburton
Genetic Diversity among CIMMYT Maize Inbred Lines Investigated with SSR Markers: II. Subtropical, Tropical Midaltitude, and Highland Maize Inbred Lines and their Relationships with Elite U.S. and European Maize
Crop Sci., October 27, 2005; 45(6): 2573 - 2582.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
J. Zhang, Y. Lu, R. G. Cantrell, and E. Hughs
Molecular Marker Diversity and Field Performance in Commercial Cotton Cultivars Evaluated in the Southwestern USA
Crop Sci., June 24, 2005; 45(4): 1483 - 1490.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
X. C. Xia, J. C. Reif, D. A. Hoisington, A. E. Melchinger, M. Frisch, and M. L. Warburton
Genetic Diversity among CIMMYT Maize Inbred Lines Investigated with SSR Markers: I. Lowland Tropical Maize
Crop Sci., November 1, 2004; 44(6): 2230 - 2237.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
J. C. Reif, X. C. Xia, A. E. Melchinger, M. L. Warburton, D. A. Hoisington, D. Beck, M. Bohn, and M. Frisch
Genetic Diversity Determined within and among CIMMYT Maize Populations of Tropical, Subtropical, and Temperate Germplasm by SSR Markers
Crop Sci., January 1, 2004; 44(1): 326 - 334.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
S. A. Mohammadi and B. M. Prasanna
Analysis of Genetic Diversity in Crop Plants--Salient Statistical Tools and Considerations
Crop Sci., July 1, 2003; 43(4): 1235 - 1248.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (25)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Agricola
Right arrow Articles by Warburton, M. L.
Right arrow Articles by Hoisington, D.
Related Collections
Right arrow Cell Biology & Molecular Genetics
Right arrow Crop Genetics
Right arrow Plant Genetic Resources

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Plant Registrations Soil Science Society of America Journal
Journal of Natural Resources
and Life Sciences Education		 Journal of
Environmental Quality