Index INTRODUCTION MATERIALS AND METHODS Plant materials CtSSR and AFLP analyses Data analysis Sequencing of ctSSR fragments RESULTS CtSSR analysis AFLP analysis DISCUSSION References

INTRODUCTION

The wheat genus Triticum L. consists of diploid einkorn, tetraploid emmer and timopheevi, and hexaploid common or dinkel wheat (for review, see Lilienfeld 1951). Common wheat (Triticum aestivum L., 2n=42, genome constitution of AABBDD) is classified into six subspecies, i.e., ssp. aestivum (L.) Thell., ssp. compactum (Host.) Mac Key, ssp. sphaerococcum (Perc.) Mac Key, ssp. macha (Dekapr. & Menabde) Mac Key, ssp. spelta (L.) Thell. and ssp. vavilovii (Jakubz.) A. Love, all of which are cultivated forms (Mac Key 1966). Among them, ssp. aestivum represents the most common and widely cultivated bread wheat. Two subspecies, compactum and sphaerococcum, are other free-threshing (non-hulled or naked) forms of common wheat. Three subspecies, macha, spelta and vavilovii, are all non-free-threshing (hulled) and thus seemingly primitive types of common wheat. Current cultivation of spelta can be seen locally in central Europe and Spain, while macha is endemic in Transcaucasia. Cultivation of spelta in Asia also has been reported (Kuckuck and Schieman 1957).

Genetic studies on the origin and evolution of common wheat have a long history. Kihara (1944) proposed from results of a comparative morphological analysis the origin of common wheat through amphidiploidization of a hybrid between emmer wheat (T. turgidum L., AABB) and the wild diploid species, Aegilops tauschii Coss. (DD, synony-mous with Ae. squarrosa L.). McFadden and Sears (1944, 1946) independently reported that hybrids between emmer wheat and Ae. tauschii morphologically resembled the non-free-threshing hexaploid subspecies, spelta, and colchicine-induced amphidiploids of this hybrid gave rise to meiotically stable and fully fertile offspring. Thus, they considered that the first primitive common wheat (spelt wheat) originated through natural hybridization followed by amphidiploidization between emmer wheat and Ae. tauschii. Kerber and Rowland (1974) reported that amphiploids derived from all crosses between tetraploid wheat (both non-free-threshing and free-threshing) and Ae. tauschii result in the non-free-threshing spelt type. This finding suggested that the original hybridization between emmer wheat and Ae. tauschii that gave rise to spelta wheat occurred before the evolution of free-threshing hexaploid wheat. Although these proposals have been widely accepted, the origin of spelta wheat still remains controversial. Recent archaeological studies suggested that spelta wheat appeared in central Europe in the Early Bronze Age (2200-1500 calibrated BC) (for review, see Nesbitt 2001). Carbonated seeds of free-threshing common wheat, however, were found in the region in the earliest Neolithic period (5400-4900 calibrated BC) (Nesbitt 2001). These results suggested that free-threshing common wheat appeared earlier than spelta wheat in Europe. Schiemann (1951) and Mac Key (1966) proposed that ssp. spelta was derived from hybrids between ssp. compactum and European emmer wheat (T. turgidum ssp. dicoccum). Ohtsuka (1998) supported this proposal by studying hybrids from the same cross combination as that used by Mac Key (1966). Based on comparative analysis of hybrid necrosis and chlorosis genes, Tsunewaki (1968, 1971) and Tsunewaki and Nakai (1973) suggested that spelta found in Europe and another non-free-threshing subspecies, macha, in Transcaucasia, originated separately from different pentaploid hybrids between tetraploid dicoccum and hexaploid aestivum. This hypothesis was partly supported by the observation that restriction fragment length polymorphisms (RFLPs) between spelta and aestivum were mainly ascribed to A and B genome differences (Liu and Tsunewaki 1991). Blatter et al. (2004) studied a part of the promoter and the coding region of the high-molecular-weight (HMW) glutenin B1-1 and A1-2 in ssp. spelta and ssp. aestivum. They found that allele frequency at Glu-B1-1 and Glu-A1-2 in European spelta was different from that in Asian spelta and also suggested the introgression of tetraploid wheat into free-threshing hexaploid wheat as the origin of European spelta. According to the study of molecular variation in Ae. tauschii, Dvorák et al. (1998) suggested that both free- and non-free-threshing hexaploid wheat share a common D-genome gene pool and thus at least the D-genome of all forms of common wheat originated monophyletically, in spite of the involvement of several Ae. tauschii parents in the evolution of common wheat. Further study of RFLPs in common wheat, however, revealed that European spelta showed the closest genetic distance with free-threshing common wheat (especially ssp. compactum), while Asian spelta wheat showed a distant relationship (Dvorák and Luo 2001). Dvorák and Luo (2001) thus reached a similar hypothesis that European spelta originated by introgression between emmer wheat and free-threshing common wheat.

In the study of genetic diversity and phylogenetic relationships in wheat and its related Triticum and Aegilops species, various DNA marker systems have been employed, including random amplified polymorphic DNA (He et al. 1992; Dweikat et al. 1993; Cao et al. 1998), RFLPs (Liu et al. 1990; Lubbers et al. 1991; Vaccino et al. 1993; Takumi et al. 1993; Siedler et al. 1994; Tsunewaki and Nakamura 1994; Mori et al. 1995; Sasanuma et al. 1996; Dvorák et al. 1998) and amplified fragment length polymorphisms (AFLPs) (Donini et al. 1997, Özkan et al 2002). Simple sequence repeats (SSRs) or microsatellite polymorphisms (Litt and Luty 1989; Tautz 1989) have recently been exploited for the study of ctDNA in plants such as soybean (Powell et al. 1995, 1996), rice (Provan et al. 1996, 1997; Ishii and McCouch 2000) and maize (Provan et al. 1999). In wheat, 24 ctSSR loci have been identified and used to analyze the allelic diversity in common wheat and its ancestral species (Ishii et al. 2001).

We herein report our study on genetic diversity and phylogeny in common wheat by ctSSR and AFLP analyses. Based on the combined data on both ct and nuclear DNA variations, the origin and evolution of common wheat is discussed.

MATERIALS AND METHODS

Plant materials

Seventy-two accessions of five common wheat (Triticum aestivum L.) subspecies (aestivum, spelta, macha, compactum and sphaerococcum) and one accession of Aegilops tauschii (KU2080) were used (Table 1). Fourteen accessions of emmer wheat (T. turgidum ssp. dicoccum (Schrank) Thell.) and a single accession each of T. turgidum ssp. carthlicum (Nevski) Mac Key, T. turgidum ssp. turgidum conv. durum Desf. Mac Key and T. aestivum ssp. aestivum cv. Chinese Spring (abbreviated as CS) were used as references. An accession of T. urartu Thum. (KU199-1) and an additional accession of T. turgidum ssp. dicoccum (KU9763) were included as refer-ence samples in a sequencing analysis of one ctSSR locus (see RESULTS). Seeds were provided from the Plant Germplasm Institute, Graduate School of Agricultural Science, Kyoto University, Japan (accessions with KU numbers), and the US Department of Agriculture, Bethesda, MD (accessions with PI or Citr numbers).

View Details

Table 1. Wheat accessions used in this study

CtSSR and AFLP analyses

Total DNA was extracted from leaves of glasshouse-grown plants according to Liu et al. (1990). DNA concentration was determined using a DyNA Qant^TM200 fluorometer (Hoefer Pharmacia Biotech, USA).

Using the DNA samples as templates, 24 ctSSR loci were amplified and analyzed according to Ishii et al. (2001). Since the two ctSSR loci, WCt20 and 21, are located only eight nucleotides apart, a pair of primers was designed to amplify the fragment containing both SSRs. Amplification was performed with Taq DNA polymerase (TOYOBO, Japan) using Gene Amp PCR System 9700 (Applied Biosystems, USA) under the following conditions: after 5 min preincubation at 94°C, DNA was amplified for 35 cycles at 94°C, 55°C and 72°C, all for 1 min, followed by post-extension for 7 min at 72°C. Amplified fragments were resolved by 6% denaturing polyacrylamide gel electrophoresis and visualized by staining with SILVER SEQUENCE^TM DNA Staining Reagents (Pro-mega, USA). Sizes of amplified fragments were determined by comparison with the standard SSR marker fragments of CS (Ishii et al. 2001) and AFLP^TM DNA Ladder size markers (GIBCO BRL, USA).

AFLP analysis was conducted basically according to Vos et al. (1995) with the modification of adding four, instead of three, selective nucleotides in the EcoRI primer. Briefly, template DNA was digested with EcoRI and MseI, and EcoRI and MseI adaptors were ligated using T4 DNA ligase. Preamplification was for 20 cycles of 30 sec at 94°C, 1 min at 56°C and 1 min at 72°C. Selective amplification using 10 primer sets was performed for 36 cycles of 30 sec at 94°C and 30 sec at different annealing temperatures (stepwise decrease from 65°C to 56°C for the first 13 cycles and then held constant at 56°C for 23 cycles) followed by 1 min at 72°C. After electrophoresis in an 8% denatured polyacrylamide gel, DNA fragments were stained by SILVER SEQUENCE^TM DNA Staining Reagents (Promega, USA).

Data analysis

The allelic diversities at the ctSSR loci were calculated according to the gene diversity value (H) described by Nei (1987) as follows: H=



, where x_ij represents the allelic frequency of the jth allele for marker i and summation extends over n alleles. Phylogenetic relationships among ct plastotypes were studied based on the dissimilarity of ctSSR allele sizes. The mean number of different alleles per locus was used as a genetic distance index. Based on the genetic distance, a tree showing the genetic relationship among the ct genomes was constructed by the neighbor joining method (Saitou and Nei 1987). The reliability of groupings was tested by the bootstrapping method (Felsenstein 1985) with 1000 subsamplings. For the AFLP data, a pairwise genetic distance (d), defined as the number of nucleotide differences per site, was calculated according to Innan et al. (1999). Based on these d values, a phylogenetic tree was constructed according to UPGMA (Sneath and Sokal 1973). The genetic diversity (π), which is defined as the average number of pairwise nucleotide changes per site (Nei and Li 1979), was calculated within subspecies as the mean genetic distance among all accessions of respective subspecies, while genetic distances between subspecies were calculated as the average pairwise genetic distance between accessions in the respective subspecies.

Sequencing of ctSSR fragments

Uniquely polymorphic ctSSR fragments that were detected in T. aestivum ssp. sphaerococcum (Spl 1, 2 and 3, see Table 1), T. turgidum ssp. dicoccum (KU9763) and T. urartu (KU199-1) were cut out from a polyacrylamide gel and extracted with TE buffer. PCR was performed using Taq DNA polymerase (Toyobo, Japan) and the extracted DNA frag-ments as templates. The amplified fragments were cloned into the pGEM-T vector (Promega, USA). Sequ-encing was done by an ABI PRISM^TM 310 Genetic Analyzer (Applied Biosystems,USA).

RESULTS

CtSSR analysis

DNA polymorphisms at 24 ctSSR loci were surveyed. These loci were designated WCt according to Ishii et al. (2001). Among 89 accessions of eight subspecies of polyploid wheat species, polymorphic band patterns were observed at 19 loci. The number of alleles at these polymorphic loci varied from 1 to 4 with an average of 2.1 (Table 2). The average number of alleles in common wheat was 1.3 in 73 accessions (Table 3), which was nearly one-third of the value of 3.6 previously estimated for two subspecies of tetraploid wheat [T. turgidum ssp. dicoccoides (Körn) Thell. and T. timopheevi ssp. araraticum (Jakubz.) Mac Key] (Ishii et al. 2001). The diversity index (H) among the polyploid wheats varied from 0.000 to 0.327 with an average of 0.124 (Table 2). The average H value in common wheat was 0.040 (Table 3), which was about 14-fold smaller than the value (H = 0.570) previously estimated in wild tetraploid wheat (Ishii et al. 2001). These results, as expected, clearly show that genetic diversity in ctDNA among common wheat accessions is much smaller than that among tetraploid wheat accessions.

View Details

Table 2. Number of alleles and diversity index (H) at 24 ctSSR loci in 89 accessions of eight polyploid wheat subspecies.

View Details

Table 3. Genetic diversity estimated within five subspecies of common wheat and one subspecies of emmer wheat using nuclear and ctDNA markers

Based on ctSSR data, 16 and eight plastotypes, respectively, were identified in common and emmer wheat (Table 4). The plastotypes designated H and E represent ones found in common and emmer wheat, respectively. Among these plastotypes, H10 and E10 are identical to plastotype 10 reported by Ishii et al. (2001). Fig. 1 shows a neighbor-joining tree of the 24 plastotypes found in this study. These plastotypes were classified into two distinct plastogroups, I and II. Two clades representing plastogroups I and II in the tree were well supported by a bootstrap test with 1000 resamplings. The bootstrap values for plastogoups I and II were 87% and 95%, respectively. Between these two plastogroups, at least seven loci (WCt 2, 12, 13, 15, 19, 23 and 24) showed allelic differences (Table 4). The results clearly show that there are two distinct maternal lineages associated with plastogroups I and II in common wheat. There were 16 plastotypes in common wheat, and about 90% of the accessions belonged to plastogroup I, while the remaining 10% belonged to plastogroup II (Fig. 1). All eight plastotypes found in emmer wheat belonged to plastogroup I (Table 4). In plastogroup I, 59% of the common wheat accessions (43 accessions) shared the identical plastotype H10 (Table 4). In plastogroup II, eight accessions of ssp. spelta (Spl16, Spl22, Spl26, Spl18, Spl19, Spl22, Spl15 and Spl27) and one accession of ssp. aestivum (Ast1) were found (Table 4). All accessions of ssp. macha possessed either plastotype H6 or H7 belonging to plastogroup I (Table 4).

View Details

Table 4. Plastotypes found in this study

View Details

Fig. 1. Neighbor-joining tree of 24 plastotypes identified among six subspecies of polyploid wheat. Circled areas are drawn in proportion to the frequency of respective ct plastotypes. Lengths of lines are proportional to numbers of the WCt loci that differed between ct plastotypes. An open triangle represents the 18 bp-duplication.

Accessions of ssp. sphaerococcum had two plastotypes, H11 and H12, in plastogroup I and possessed unique alleles with an 18-bp insertion at the WCt20/21 locus (Table 4). To study in detail the nature of this insertion, the polymorphic fragment was cloned and sequenced. The analysis showed that the insertion occurred in the vicinity of the targeted SSR loci (WCt20 and WCt21) and was due to a tandem duplication (Fig. 2). In a separate study, a similar allele with a 19-bp insertion was detected in one accession of T. turgidum ssp. dicoccum (KU9763) collected from Ethiopia, and another allele with a 17-bp deletion was detected in one accession (KU199-1) of a wild diploid A genome species, T. urartu. Sequence analysis of these alleles showed that the 19-bp insertion in dicoccum was due to a tandem duplication similar to that found in sphaerococcum. The 17-bp deletion in T. urartu occurred within the targeted SSR locus. The significance of these deletions in relation to wheat phylogeny is now under investigation.

View Details

Fig. 2. Sequence alignment of unique alleles at the WCt20/21 locus. (a) T. aestivum ssp. aestivum cv. CS, (b) ssp. sphaerococcum (Sph1 ~ 3, in Table 1), (c) T. turgidum ssp. dicoccum (KU9763) and (d) T. urartu (KU199-1). Nucleotides in bold letters represent the coding region of infA gene at the locus. Boxed with thin lines: Microsatellite loci, WCt20 and WCt21. Boxed with thick lines: Short tandem repeats.

AFLP analysis

Using 10 selective primer sets, 487 discernible DNA fragments were obtained. Table 3 shows the genetic diversity index (π) estimated by AFLP analysis within the five common wheat and one emmer wheat subspecies. Among the common wheat subspecies, ssp. compactum showed the largest π value (0.00491), followed by ssp. aestivum (0.00488). The smallest π value was found in ssp. macha (0.00202) followed by that in ssp. spelta (0.00269). The π value within ssp. dicoccum (0.00766) was much larger than those of all common wheat subspecies.

Genetic distances (d) were calculated among the polyploid wheat subspecies (Table 5). Among all pairs of common wheat subspecies, the average genetic distance between macha and aestivum was the smallest (0.00485), while that between compactum and spelta was the largest (0.00603). The genetic distances between emmer wheat (ssp. dicoccum) and all common wheat subspecies were nearly three times larger than those among common wheat subspecies. This probably is due to the absence of D genome markers in common wheat. In a phylogenetic tree constructed by UPGMA based on the calculated genetic distances (tree not shown), aestivum and macha showed the closest relationship, while spelta was most distantly related to the other common wheat subspecies.

View Details

Table 5. Genetic distances (d × 100) estimated among all pairs of six wheat subspecies1) based on AFLP data.

A phylogenetic tree was constructed using the AFLP data obtained from all 89 accessions of polyploid wheat with one accession of Ae. tauschii (sqr1) as an outgroup (Fig. 3). In this tree, plastotypes of all accessions were added as references. The tree contained two major clusters, I and II. Cluster I consisted of all common wheat accessions and one tetraploid wheat accession (Crt1), whereas cluster II consisted of the remaining emmer wheat accessions. The major part of cluster I was further divided into two subclusters, A and B. Subcluster I-A consisted of all common wheat subspecies other than spelta (except for three spelta accessions, Spl11, Spl13 and Spl18). All accessions in this subcluster except one for spelta (Spl18) and aestivum (Ast1) had a genome belonging to plastogroup I. Subcluster I-A-1 consisted mainly of macha and aestivum accessions collected from countries mostly west of Iraq. This subcluster was divided into four minor clusters, I-A-1-a to I-A-1-d. Subcluster I-A-2 consisted mainly of aestivum and compactum accessions collected from countries east of Iran. A single spelta accession (Spl13) from Iran was included in this subcluster. Subcluster I-A-3 consisted of three sphaerococcum accessions. Subcluster I-B consisted only of spelta accessions, and was further divided into two subclusters. Subcluster I-B-1 consisted of 10 Spanish accessions and two German accessions, all of which belonged to plastogroup I. Subcluster I-B-2 consisted only of German accessions of spelta, among which four belonged to plastogroup I and seven belonged to plastogroup II. All accessions of dicoccum were grouped into the second major cluster, II, which was divided into two subclusters, II-A and II-B. Subcluster II-A contained accessions from India, Afghanistan and Ethiopia, while II-B contained accessions from Near and Middle Eastern countries. All accessions belonging to subcluster II-A had either the E26 or E30 plastotype, both of which commonly possessed two 1-bp deletions at the loci WCt16 and Wct23. Five plastotypes (E10, E34, E33, E29 and E68) found in accessions belonging to subcluster II-B did not possess such deletions, and thus plastotypes E26 and E30 were considered to be differentiated from the others.

View Details

Fig. 3. Phylogenetic tree constructed by UPGMA showing relationships among 90 wheat accessions. Afg: Afghanistan, Arm: Armenia, Eth: Ethiopia, Grg: Georgia, Grm: Germany, Ind: India, Irn: Iran, Irq: Iraq, Jpn: Japan, Jrd: Jordan, Lbn: Lebanon, Pks: Pakistan, Pls: Palestine, Spn: Spain, Syr: Syria, Trk: Turkey, Usr: USSR. ct DNA type I and II represent plastogroup I and II, respectively.

DISCUSSION

The origin and evolution of common wheat has long been debated. A recent study of ctSSR variation has demonstrated the existence of two discrete ct genome types in common wheat, and further suggested that they might have independently originated from corresponding wild and cultivated emmer wheat (Ishii et al. 2001). To clarify the origin and evolution of common wheat, we employed ctSSR and AFLP analyses. Our data, based on ctDNA variation, confirmed the existence of two clearly differentiated maternal lineages in common wheat, especially in spelta (Table 4 and Fig. 1). Comparisons between the results of AFLP analysis of total DNA and ctSSR analysis also suggested the occurrence of nuclear introgressions among common wheat during and/or after their evolution.

According to Tsunewaki (1968, 1971) and Tsunewaki and Nakai (1973), spelta originated independently in Asia and Europe. They proposed that Asian spelta (found in Iran) is probably the progenitor of other common wheats as proposed by Kuckuck (1959) (cited by Tsunewaki 1971), and that European spelta independently originated through introgressive hybridization between hulled emmer wheat and free-threshing common wheat. Dvorák and Luo (2001) and Blatter et al. (2004) reached a similar conclusion.

In ctSSR analysis, we found that 19 accessions of spelta possessed one of four plastotypes (H10, H2, H8 and H7) belonging to plastogroup I, whereas eight other accessions possessed one of four plastotypes belonging to plas-togroup II (H13 to H16) (Table 4). These two plastogroups are clearly differentiated (Fig. 1). All 12 spelta accessions from Spain belonged to plastogroup I, while both plastogroups I and II were found in spelta accessions from Germany (six belonged to plastogroup I and eight belonged to plastogroup II) (Table 4). Assuming that most of the variations detected by AFLP analysis are due to nuclear DNA variation, and thus the tree constructed by AFLP analysis reflects nuclear genome differentiation, we can speculate on the origin and evolution of spelta by comparing the magnitudes of differentiation in ct and nuclear DNAs. In an AFLP data-based phylogenetic tree, a major cluster, I-B, consisted only of spelta accessions, within which all German accessions were grouped into subcluster I-B-2 and all Spanish accessions were grouped into subcluster I-B-1 (Fig. 3). The spelta accessions in Germany are thus closely related to each other according to the variation in the nuclear genome, while they contain two well differentiated plastotypes (Table 4 and Fig. 3). As to the maternal origin of European spelta, the following three scenarios could be con-sidered. 1) The two groups of spelta possessing either plastogroup I or II originated independently by crosses between emmer wheat (T. turgidun ssp. dicoccum) and Ae. tauschii and they migrated to Europe. 2) spelta originated twice independently, i.e., once in Asia and once in Europe, and the origin of European spelta was through introgression of the spelt character from emmer wheat to European free-threshing common wheat. 3) spelta originated once in Asia (either with group I or II ctDNA) and thereafter mutations occurred at multiple ctDNA microsatellite loci, converting the plastotype either from I to II or from II to I, and thus establishing both European and Asian types of spelta. The third scenario is unlikely because at least seven ctSSR loci showed allelic differences between the two plastogroups (Table 4), and the probability of converting one plastotype to the other in a single event should be extremely low. Although only one accession of spelta from Asia (Spl13 from Iran) was used in this study, this Asian spelta showed a distant relationship to the European spelta in the AFLP tree (see subcluster I-A-2 in Fig. 3). This result suggests that Asian spelta and European spelta are differentiated in their nuclear genomes. In addition, spelta showed a five times larger H value than the average value in other common wheat subspecies for the ct genome, while the relative diversity in the nuclear genome was comparable to those in aestivum and compactum (Table 3). This result could be explained if we assume introgression of cytoplasm, e.g., plastogroup II ctDNA into common wheat. In the present study, all emmer wheat accessions had the plastogroup I plastotype (Table 4). However, in our recent survey, we found both plastogroups I and II in domesticated emmer wheat (T. turgidum ssp. dicoccum), including those in central Europe (Mori et al. in preparation). Taken together, our results support scenario 2), which has been proposed earlier (Mac Key 1966; Tsunewaki 1968, 1971; Tsunewaki and Nakai 1973; Ohtuka 1998; Dvorák and Luo 2001, Blatter et al. 2004). Further study of southwestern Asian accessions of spelta should help clarify the origin and phylogenetic relationship of this non-free-threshing common wheat.

Tsunewaki (1968, 1971) and Tsunewaki and Nakai (1973) reported that macha, another non-free-threshing common wheat subspecies endemic in Transcaucasia, possessed the Ch1ch2 genotype (two complementary loci that cause hybrid chlorosis) at a very high frequency (85%), while all other common wheat subspecies frequently had the complementary genotype, ch1Ch2. Considering the distribution of hybrid chlorosis genotypes in other wheat species, including Ae. tauschii, Tsunewaki (1971) proposed that macha originated through introgression of the aestivum D genome into dicoccum. We could not test this hypothesis because of the lack of useful ctDNA and/or nuclear DNA variations between macha and other common wheat subspecies. All accessions of macha possessed either the H6 or H7 plastotypes, which are unique to this subspecies but very closely related to the major plastotype, H10 (Table 4). Furthermore, genetic diversity within macha estimated based on AFLP data was the smallest among all common wheat subspecies (Table 3), and its genetic distance with aestivum was the smallest (Table 5). All 10 accessions of macha belonged to one minor cluster, I-A-1-a (Fig. 3). All these results suggest that this subspecies is closely related to aestivum, although it is somewhat differentiated with respect to both ct and nuclear DNA variations.

All sphaerococcum accessions formed one subcluster, I-A-3, and possessed either the H11 or H12 plastotype in plastogroup I. Genetic diversity (π) within this subspecies, estimated based on AFLP data, was smaller than those in spelta, compactum and aestivum (Table 3). These results suggest that sphaerococcum might have experienced either a strong bottleneck or strong selection. This subspecies is distinguished from aestivum by its compact and square-headed spike, which is regulated by the S gene, and is found only in India and Pakistan (Hosono 1954). Both plastotypes H11 and H12 found in sphaerococcum possessed an 18-bp tandem repeat near the WCt20/21 locus (Fig. 2). We therefore studied additional accessions of sphaerococcum and aestivum collected from India and Pakistan and found that the two groups with and without this insertion coexist in this area (Mori et al., unpublished). These results suggest that genetic introgression likely took place between them and the unique plastotype might have been introduced from sphaerococcum to aestivum in India and Pakistan. Interestingly, a highly homologous tandem repeat was also found in one Ethiopian dicoccum accession (KU9763) (Fig. 2). Therefore, it is also possible that there was a genetic interchange between Ethiopian emmer and sphaerococcum or aestivum in India and Pakistan, although further study has to be done to test this possibility.

Another free-threshing common wheat subspecies, ssp. compactum, which is distinguished from aestivum by its compact spikes, possessed three different plastotypes in plastogroup I (Table 4) but did not form a distinct cluster in an AFLP data-based phylogenetic tree (Fig. 3). It showed the largest genetic diversity in the nuclear genome but the diversity in the ct genome was the second lowest among common wheat subspecies (Table 3). This situation is in clear contrast to that of spelta, although the reason for this phenomenon is unknown.

All dicoccum accessions formed one major cluster, II, which was further divided into three subclusters II-A, II-B and II-C (Fig. 3). Subcluster II-A consisted of accessions from India, Afghanistan and Ethiopia and possessed either the E26 or E30 plastotype, which had two 1-bp deletions, commonly in the WCt16 and Wct23 loci. In contrast, subcluster II-B consisted of accessions with five different plastotypes (E10, E34, E33, E29 and E68). The genetic distance estimated based on AFLP data between accessions belonging to these two subclusters (0.00887) was much larger than the mean genetic distance between all pairs of common wheat subspecies (0.00542). This result suggests that nuclear genome of dicoccum found in Afganistan, Ethiopia and India is differentiated from others. A tetraploid subspecies, carthlicum (Nevski) Mac Key, which is endemic in Transcaucasia, was discovered by Vavilov (1919) (cited by Ohtsuka 1991). Vavilov (1926) (cited by Ohtsuka 1991) considered that this tetraploid wheat was a secondary species derived from pentaploid hybrids between emmer and common wheat. In our AFLP study, an accession of carthlicum was located in the phylogenetic tree close to several common wheat accessions, despite the fact that it does not possess the D genome (Fig. 3). Our result thus supports the secondary origin of this tetraploid wheat.

In conclusion, our data regarding ct and nuclear DNA variations indicate the existence of at least two discrete maternal lineages in the evolution of common wheat, and also suggest nuclear introgressions among polyploid wheats during and/or after their evolution. The origin of European spelta wheat was most probably a secondary event that took place through hybridization between non-free-threshing emmer wheat and common wheat.

We thank the National Small Grain Facility (USDA-ARS, USA) for providing us with the seed stocks used in this investigation. We acknowledge support by a grant-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 16201047). Contribution No. 167 from the Laboratory of Plant Genetics, Faculty of Agriculture, Kobe University.