| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USDA-ARS, U.S. Vegetable Laboratory, Charleston, SC 29414
b USDA-ARS Grain Legume Genetics and Physiology Unit, Pullman, WA 99164. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U. S. Department of Agriculture
* Corresponding author (gvandemark{at}wsu.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: FDC, fall dormancy class • NAAIC, North American Alfalfa Improvement Conference • NAMLVRB, National Alfalfa and Miscellaneous Legume Variety Review Board • PCA, principal component analysis • PCR, polymerase chain reaction • PI, plant introduction • SRAP, sequence-related amplified polymorphism
a USDA-ARS, U.S. Vegetable Laboratory, Charleston, SC 29414
b USDA-ARS Grain Legume Genetics and Physiology Unit, Pullman, WA 99164. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U. S. Department of Agriculture
* Corresponding author (gvandemark{at}wsu.edu).
Synthetic populations of alfalfa (Medicago sativa L.) consist of genetically heterogeneous mixtures of plants, which may complicate use of molecular markers for examining genetic relationships among populations. Molecular marker techniques combined with bulked DNA from multiple plants provides perhaps the best combination of accuracy and high throughput for characterization and differentiation of alfalfa populations. Using 20-plant bulked DNA templates we assessed how sequence-related amplified polymorphisms (SRAPs) would estimate genetic similarity among 29 nondormant and semidormant alfalfa populations representing diverse genetic backgrounds and the nine historically recognized U.S. germplasm sources. Cluster analysis based on estimates of genetic similarity showed separation of populations in fall dormancy classes 6 to 8 (semidormant) from fall dormancy classes 9 to 11 (nondormant) and in many examples, DNA bulks of the same population clustered together. A single SRAP marker was identified that was present only in bulks of fall dormancy classes 6 and 7, and another marker was identified that was detected only in bulks of fall dormancy class 8. This study demonstrates that the amplification of SRAP markers from DNA extracted from bulked plant samples is an efficient method for estimating genetic similarity both within and between a large numbers of populations. The SRAPs may also be useful for indirectly classifying relative fall dormancies of uncharacterized populations.
Abbreviations: FDC, fall dormancy class • NAAIC, North American Alfalfa Improvement Conference • NAMLVRB, National Alfalfa and Miscellaneous Legume Variety Review Board • PCA, principal component analysis • PCR, polymerase chain reaction • PI, plant introduction • SRAP, sequence-related amplified polymorphism
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Barnes et al. (1977) suggested that North American alfalfa cultivars are predominately derived from nine distinct sources of germplasm. These germplasm sources are M. falcata, M. sativa ‘Ladak’, M. varia, a naturally occurring hybrid between M. sativa and M. falcata, M. sativa ‘Chilean’, M. sativa ‘Flemish’, M. sativa ‘Turkistan’, M. sativa ‘Peruvian’, M. sativa ‘Indian’, and M. sativa ‘African’. These germplasm sources were classified primarily based on geographic origin, flower color, and fall dormancy characteristics (Barnes et al., 1977). Medicago falcata, M. varia, Ladak, and Turkistan are considered to be sources of fall dormancy in commercial cultivars (Barnes et al., 1977). The Indian, African, Chilean, and Peruvian germplasm sources are considered to primarily contribute to nonfall dormant cultivars (Barnes et al., 1977; Lehman et al., 1987).
Several studies have demonstrated correlations between reduced fall dormancy and winter survival (Smith, 1958, 1961; Stout, 1985). Although genetic analysis suggests that these two traits may be inherited independently (Brouwer et al., 2000; Brummer et al., 2000), nondormant alfalfa cultivars have been observed to have significantly lower winter survival than more dormant cultivars when grown under identical methods of harvest management in locations having severe winters (Sheaffer et al., 1992). In addition, selection for increased fall dormancy in the nondormant cultivar CUF101 (Lehman et al., 1983) resulted in improved winter survival (Cunningham et al., 1998).
Alfalfa cultivars are placed in a FDC based on a protocol established by the North American Alfalfa Improvement Conference (NAAIC), which relies on regrowth measurements in fall, multiple years of data, and comparison to cultivars of known FDC to determine placement of a cultivar into 1 of 11 classes (Teuber et al., 1998). The protocol, while widely accepted and successful in FDC determination, is time and labor intensive. The identification of molecular markers that are tightly linked to traits that are difficult to accurately evaluate without multiple years of data, such as fall dormancy, could accelerate the process of alfalfa cultivar development by allowing for the direct selection of seedlings with desired genotypes. However, few examples exist in the literature describing the identification of molecular markers linked specific phenotypes in tetraploid alfalfa (Brouwer et al., 2000; Musial et al., 2005; Robins et al., 2007a,b). The paucity of useful markers for alfalfa breeding is likely due to difficulties associated with identifying marker linkages in tetraploid genomes and the inability to produce inbred lines because of inbreeding depression.
Most studies using molecular markers in alfalfa have focused on examining genetic relationships between different populations. Commercial alfalfa cultivars are predominately synthetic populations resulting from multiple cycles of random mating among selected parental plants and their progeny (Busbice et al., 1972). As a consequence of repeated cycles of random mating, cultivars consist of genetically heterogeneous mixtures of genotypes. The genetic diversity present between plants in a given cultivar creates challenges for developing marker experiments that adequately examine a sufficient number of plants in a cultivar to sample the genetic variation present within the cultivar, while permitting the examination of a large number of different cultivars.
Several efforts have attempted to examine genetic relationships between alfalfa populations by using molecular markers to detect polymorphisms in DNA extracted from bulked plant samples. Kidwell et al. (1994) separated M. falcata (Bingham, 1993) and Peruvian germplasm from the other seven sources of M. sativa germplasm by detecting restriction fragment length polymorphisms (RFLPs) (Beckmann and Soller, 1983) between four and six plant bulk DNA extracts. Yu and Pauls (1993) demonstrated that randomly amplified polymorphic DNA (RAPD) (Welsh and McClelland, 1990) could distinguish alfalfa populations and cultivars based on the amplification of bulk DNA collected from 10 plants. Segovia-Lerma et al. (2003) used amplified fragment length polymorphisms (AFLPs) (Vos et al., 1995) and bulk DNA to separate the nine germplasm sources into two main clusters, one containing M. falcata and another containing the other eight germplasm sources. Within the cluster containing all sources of germplasm except for M. falcata was another subcluster supported by a bootstrap value >50 that contained Ladak, Flemish, and M. varia germplasm. Genetic relationships between populations have been estimated using molecular markers and bulked plant DNA for several other crop species, including Kentucky bluegrass (Poa pratensis L.) (Johnson et al., 2002), safflower (Carthamus tinctorius L.) (Sehgal and Raina, 2005), and crested wheatgrass (Agropyron cristatum Gaertn.) (Mellish et al., 2002).
Molecular markers vary in their ability to detect genetic polymorphisms in a timely and cost-effective manner. Sequence related amplified polymorphisms (SRAPs) (Li and Quiros, 2001) are DNA markers amplified by PCR in a single tube using forward and reverse primers with the following sequence motifs: (i) the first 10 to 11 nucleotides at the 5' end of each primer is a random sequence of nucleotides, similar to primers used to amplify RAPDs; (ii) the sequence CCGG in the forward primer and AATT in the reverse primer, and (iii) three selective nucleotides at the 3' end of each primer, which is reminiscent of primers used in the secondary PCR to generate AFLPs. The CCGG sequence in the forward primer results in this primer preferentially annealing to exonic sequences, while the AATT domain in the reverse primer causes preferential annealing to introns. Li and Quiros (2001) observed that 60% of the SRAPs amplified and sequenced from Brassica oleracea L. shared significant homology to reported gene sequences in GENBANK, suggesting that SRAPs preferentially amplify sequences in open reading frames (ORFs). The SRAPs are evenly distributed among linkage groups and highly reproducible (Li and Quiros, 2001). The SRAPs have also been shown to provide good distinction of populations of buffalograss (Buchloe spp.) (Budak et al., 2004), wheat (Triticum aestiva L.) (Fufa et al., 2005), squash (Cucurbita moschata L.) (Ferriol et al., 2004) and oilseed rape (Brassica napus L.) (Riaz et al., 2001). Vandemark et al. (2006) demonstrated that SRAPs amplified from bulked DNA could separate a small set of public alfalfa cultivars from the nine historical sources of alfalfa germplasm.
The ability to determine genetic relationships between alfalfa populations can confer several benefits to breeding programs, including minimizing redundancies among populations and clones maintained as parental materials. An improved understanding of the genetic structure within specific populations will also provide essential knowledge of the effects of multiple cycles of random mating on the genetic heterogeneity of synthetic cultivars. Ultimately identifying markers closely associated with traits that require laborious tests and multiple years of data, such as fall dormancy, would greatly simplify selection of plants with desirable genotypes. Previous efforts to characterize North American alfalfa populations with molecular markers have collectively only examined the nine original sources of germplasm and a small subset of cultivars (Bauchan et al., 2003; Kidwell et al., 1994; Segovia-Lerma et al., 2003; Yu and Pauls, 1993; Vandemark et al., 2006). The objectives of this study were (i) to use SRAPs and bulked DNA to estimate genetic similarities among the nine populations representing the original sources of Medicago germplasm in North America and 29 semidormant and nondormant alfalfa cultivars, (ii) to survey genetic variation within these alfalfa populations using SRAPs, and (iii) to examine the ability of SRAPs to differentiate among cultivars of different FDCs.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Sequence-Related Amplified Polymorphism Reactions
Seven SRAP primer pair combinations (forward primer plus reverse primer) were selected for the study based on previously observed ability to generate polymorphisms in alfalfa populations. These SRAP primer pairs were F7/em5, F11/R15, F12/R9, me2/R9, me4/R14, me4/em2, and F9/R9 (Table 2
). The SRAPs were amplified from 25 µL reaction volumes using conditions and thermocycling profiles previously described (Vandemark et al., 2006) for detecting SRAPs from bulked alfalfa DNA. All PCR reactions were completed in a GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA). Fifteen µL of PCR product per reaction were resolved by gel electrophoresis at 100 V for 5 h using Criterion precast 15% PAGE-TBE gels (Bio-Rad, Hercules, CA). Gels were stained in 0.01% (w/v) ethidium bromide solution for 30 min and amplicons were visualized with UV light. The reproducibility of SRAP reactions was evaluated by taking a subset of bulk DNA samples (Indian bulks 1–3, M. varia bulks 1,2, and 5, Beacon bulks 1–3, and Ameristand 801S bulks 1–3) and performing replicate SRAP reactions with the primer pairs F7/em5, F9/R9, and F11/R15 using two different GeneAmp PCR System 9700 thermocyclers. Both replicates of each sample/primer pair combination were resolved in adjacent wells of 15% PAGE-TBE gels as described above.
|
An unweighted pair-group method using arithmetic averages (UPGMA) cluster analysis was conducted to generate a dendrogram representing genetic similarity based on Gs values of 20-plant bulk DNA samples. Calculations of Gs, cluster analysis and representation were conducted in NTSyspc version 2.20e (Rohlf, 2002). Bootstrap support for clusters was conducted in FreeTree (http://www.natur.cuni.cz/
flegr/freetree.htm) using 1000 permuted datasets (Pavlicek et al., 1999). Principal component analysis (PCA) was conducted using the PRINCOMP procedure in SAS (SAS Institute, Cary, NC) and the broken stick model (MacArthur, 1957) was used to determine the significance of principle components.
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
10% and 
90% frequency classes (Fig. 1). No cultivar or population-specific amplicons (i.e., present only in all bulks of a single population) were detected. A 105 bp SRAP marker amplified by primer pair me 4/em 2 was detected that was present only in bulked DNAs of FDC 6 and 7. Similarly, a 290 bp SRAP marker amplified by primer pair F 11/R 15 was identified that was present in only in FDC 8 bulks.
|
|
Cluster analysis based on Gs revealed that the SRAP markers detected in bulked plant DNA samples could partially separate the populations examined in this study (Fig. 2 : http://www.ars.usda.gov/sp2UserFiles/person/5791/PDF/alfalfadiversitydendro.pdf, verified 21 Aug. 2007). For the sake of narrative, some of the clusters have been numbered on the dendrogram and relevant subclusters have been labeled with letters. While there was a general lack of robust bootstrap values ( >50%) in the base clusters, high support values were obtained for several subclusters joining different populations or bulk DNA samples of the same population.
|
Subclusters on the dendrogram (Fig. 2) reveal additional grouping of cultivars that reflect similar fall dormancy classes (Table 1). Cluster 1 only contains bulks of cultivars in fall dormancy classes 7 and 8, Cluster 2A contains only bulks of fall dormancy class 6 cultivars, Cluster 2B only contains bulks of fall dormancy class 8 cultivars, Cluster 3 contains only bulks of fall dormancy class 7 cultivars, and Cluster 5 contains all the bulks of all the cultivars in fall dormancy classes 9 to 11 except for a single bulk of Beacon.
Several subclusters can be observed that are supported by robust bootstrap support values (>50) and only contain all the bulks of a single cultivar (Fig. 2). A subcluster containing all five bulks of 13R Supreme is observed in Cluster 1, while Cluster 4 is composed entirely of the five bulks of Wilson. All five bulks of cultivars 5929, UC 1465, UC 1887, Salado, and CUF 101 are located in Cluster 5 on individual subclusters supported by robust bootstrap support values.
Cluster 7 is composed of all five bulks of M. falcata WISFAL, while the bulks of the other historical sources of Medicago germplasm are primarily found in Clusters 5 and 6 (Fig. 2). Cluster 5 contains a subcluster supported by a robust bootstrap value that contains all five Chilean bulks. Between Clusters 5 and 6 is a well-supported cluster containing all five bulks of M. varia. Subcluster 6A contains all the Flemish and African bulks. Subcluster 6B separates into two additional subclusters, one of which contains all of the Peruvian bulks as well as four Indian bulks, and another, supported by high bootstrap values, containing all five Turkistan bulks.
To assess the importance of marker prevalence, cluster analyses were conducted using subsets of the marker data. Analysis of markers present in 90%, 60%, and 10 to 90% of the bulked DNAs did not provide better resolution of populations nor were better bootstrap support values obtained than what resulted from using all 188 markers (data not shown). The PCA showed the first three components to account for 11.9, 6.0, and 5.4% of variation within cultivars, and while additional components also accounted for significant variation based on the broken stick model (roughly 65% in total), they provided no further separation of populations.
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A comparison of the dendrogram produced using SRAP data (Fig. 2) with that based on AFLP data (Segovia-Lerma et al., 2003) suggests that AFLPs were more effective than SRAPs at separated the nine germplasm sources into clusters that agree with expected relationships based on geographic origins (Segovia-Lerma et al., 2003). The AFLPs closely grouped the Chilean and Peruvian populations (Segovia-Lerma et al., 2003), whereas SRAPs grouped Chilean closer to M. varia than to Peruvian (Fig. 2). Segovia-Lerma et al. (2003) based their results on 3754 AFLPs, while our results are based on 188 SRAPs, and it is possible that an examination based on a greater number of SRAPs might produce results more similar to those reported by Segovia-Lerma et al. (2003). Different results based on SRAPs and AFLPs may also reflect differences in which populations were used to represent the original sources of germplasm. In this study we used the populations developed by Melton et al. (1990), whereas Segovia-Lerma et al. (2003) used older populations, which were later combined along with other populations by Melton et al. (1990). The populations developed by Melton et al. (1990) have a broader base of parental materials and are likely to be more genetically heterogeneous. In addition, we examined five bulks of 20 plants per bulk for each population, whereas Segovia-Lerma et al. (2003) examined a single bulk of 30 plants for each population, with each bulk being composed of 15 plants of each of two relic populations. Due to the reductive nature of cluster analysis, it is not surprising that a similar number of genotypes, bulked into a single categorical (i.e., population) sample, would produce tighter clusters on a dendrogram as compared to our design using five samples of genotype bulks per population category.
Examining multiple bulks for each population allowed for comparisons between populations in the mean Gs among bulks within a population. The within-population mean Gs of the Chilean and M. falcata WISFAL populations were the highest among the nine historical source populations, suggesting these two populations have the least genetic diversity. Kidwell et al. (1994) similarly observed that M. falcata WISFAL had the lowest within-population marker diversity among the nine germplasm sources based on RFLPs. The relatively high within-population mean Gs observed for the Chilean population may reflect a relatively narrow genetic base in the materials used to develop this population. The Chilean population was developed from intermating six populations, ‘Chilean Common’, ‘Chilean 21-5’, ‘Chilean 21-5-5’, ‘California Common’, ‘California Common 49’, and ‘Caliverde’ (Melton et al., 1990). All of these populations are estimated to be composed entirely of original introductions of Chilean germplasm into North America, except for Caliverde, which is estimated to contain 90% Chilean and 10% Turkistan germplasm (Barnes et al., 1977).
In contrast, the within-population mean Gs of the Indian and M. varia populations were the lowest among the nine historical source populations (Table 3). The detection of a relatively high amount of within-population marker diversity in M. varia is not surprising, since it was developed from crosses between PI 255178, which is an introduction from Poland, and populations derived from selections among hybrids between M. falcata and M. sativa that were independently introduced into North America from Germany and Russia (Barnes et al., 1977; Melton et al., 1990). The relatively high within-population marker diversity detected in Indian germplasm was not expected based on its pedigree (Melton et al., 1990), which indicates that it was developed from Arizona Indian, Sirsa No. 9 and Mesa-Sirsa, all of which are considered to be composed entirely from the original introduction of Indian germplasm into California. However, Warburton and Smith (1993) examined agronomic and morphological traits in 21 different PI accessions from northern and western India and found that cluster analysis separated the accessions into four clusters. The relatively low within-population mean Gs of the Indian population may be partially due to the effects of sampling a finite number of markers, but may also reflect greater genetic variation in this population than would be expected based on the classification system proposed by Barnes et al. (1977).
Although in general, all five bulks of a population tended to cluster closely together on the dendrogram (Fig. 2), there were several individual bulks, most notably Indian 1, M. varia 5, Beacon 3, and AmeriStand 3 that were distant on the dendrogram from the four other bulks of each respective population. The mean within-population Gs of Indian 1, M. varia 5, Beacon 3, and AmeriStand 3 were 0.75, 0.76, 0.79, and 0.78, respectively, while the average mean within-population Gs of the other four bulks of each population were 0.80, 0.81, 0.87, and 0.82. This indicates that each distant bulk had a mean within-population Gs that was at least 90% of the average mean within-population Gs of the other four bulks of its population. The magnitude of these differences would likely be greater if they were the result of gross technical errors in PCR, such as a lack of amplification in these distant bulks, or the spurious amplification of a large set of unique SRAPs in these bulks. In addition, the amplification profiles generated using these four distant bulks and several different primer pairs were unanimously reproduced using two different thermocyclers. These observations suggest that the positioning of these distant bulks on the dendrogram may be reflecting differences in allele frequencies between these bulks and other bulks of the same populations.
Cultivars in FDC 9 to 11 were more closely grouped with the populations representing the original sources of Medicago germplasm than were the cultivars in FDC 6 to 8 (Fig. 2). This suggests that the very nondormant cultivars (FDC 9–11) examined in this study are more similar genetically to the original sources of alfalfa germplasm than are cultivars in FDC 6 to 8. Cultivars in FDC 6 to 8 typically have higher levels of resistance than cultivars in FDC 9 to 11 to several diseases, including bacterial wilt, anthracnose, and stem nematodes (Alfalfa Council, 2005). The closer association of the very nondormant cultivars to the original Medicago germplasm sources may reflect greater genetic divergence of the FDC 6 to 8 cultivars from the original source populations as a consequence of selection for traits such a resistance to multiple diseases. The distant grouping of the majority of cultivars from the original sources of Medicago germplasm (Fig. 2) suggests that the original sources of alfalfa germplasm may only have limited utility for improving alfalfa cultivars.
Several alfalfa cultivars, including Wilson and 13R Supreme, could be distinguished based on cluster analysis of SRAP marker data (Fig. 2). An averaging of the mean Gs across all entries (Table 3) indicated that Wilson was the most genetically distinct cultivar. The cultivar description for Wilson suggests that it is unique among the other cultivars examined in this report in that it is constituted primarily of Turkistan (72%) and Chilean (22%) parentage. 13R Supreme also has a rather unique combination of contributions from the historical sources of germplasm among the cultivars examined, in that it is estimated to be composed of 70% African and 30% Turkistan germplasm.
Some relationships were observed between cultivars based on cluster analysis that agreed with similarities in reported pedigrees (Fig. 2). For example, 5929 was found in close proximity to CUF 101, which was one of the parental populations used in the development of 5929 (Lehman et al., 1983; Woodward et al., 1988). Very close clustering was also observed between Malone and Dona Ana, which was one of two parental populations used in the development of Malone (Melton et al., 1989). These results are encouraging with respect to the ability of SRAP markers to suggest genetic relationships that reflect pedigree information. However, there was no close grouping of Ameristand 801S bulks with bulks of Salado, which is the cultivar from which Ameristand 801S was selected directly. Similarly, Arriba was largely selected from Lobo, but these two cultivars did not closely group together. Several factors may have contributed to discrepancies between clustering based on SRAP markers and expected relationships based on pedigree information. These include the number of SRAP markers detected, possible changes in gene frequencies between newer cultivars and earlier cultivars due to random mating among selected parental clones and progeny, genetic shift during seed production, and directed selection for desirable traits in the newer cultivars.
In a preliminary analysis of alfalfa cultivars with SRAPs, clustering among cultivars was observed that reflected shared fall dormancy ratings, with the dormant cultivars Oneida (FDC = 2) and Vernal (FDC = 3) clustering together while the less dormant cultivars Malone (FDC = 7) and CUF101 (FDC = 9) also formed a distinct cluster (Vandemark et al., 2006). Similar results were observed in this study (Fig. 2). In addition, a SRAP marker was identified that was present only in bulks of FDCs 6 and 7, and another marker was identified that was detected only in bulks of FDC 8. The process of evaluating plants and populations for fall dormancy would be greatly accelerated if a molecular marker could be identified that was associated with the this trait. It will be a considerable challenge to develop a set of markers that can discriminate among all FDCs. Nonetheless, the overall separation of cultivars based on cluster analysis and the identification of SRAPs associated with specific fall dormancy classes indicates that these markers have potential for accurately discriminating alfalfa cultivars based on fall dormancy.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Received for publication December 11, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
