Part 5. Historical Aspects and Crop Evolution

Genetic Evidence on the Origin of Triticum aestivum L. - J. Dvorák, M.-C. Luo and Z.-L. Yang
Introgression of Durum into Wild Emmer and the Agricultural Origin Question - M.A. Blumler
The Variation of Grain Characters in Diploid and Tetraploid Hulled Wheats and its Relevance for the Archaeological Record - K. Hammer and C.-E. Specht
Utilization of Ancient Tetraploid Wheat Species for Drought Tolerance in Durum Wheat (Triticum durum Desf.) - A. Al Hakimi and P. Monneveux
Archaeobotanical Evidence for Evolution of Cultivated Wheat and Barley in Armenia - P.A. Gandilian
Extinction Threat of Wild African Gossypium species in their Center of Diversity - V. Holubec

Genetic Evidence on the Origin of Triticum aestivum L. - J. Dvorák, M.-C. Luo and Z.-L. Yang

Introduction

Because of the great economic significance of wheat, its evolution has received a great deal of attention over the past 80 years. It was recognized early that wheat species fall into three natural groups according to their ploidy (Kihara 1924). Six biological species are recognized today, two at each ploidy level (Fig. 1). Of the six species, the most important economically is Triticum aestivum L., because it includes bread wheat. Triticum aestivum comprises a number of subspecies or other taxa which are interfertile and which differ from each other by a single or a few major genes (Mac Key 1966). In some of these subspecies, such as subsp. aestivum (bread wheat), subsp. compactum (Host) Thell. (club wheat), subsp. sphaerococcum (shot wheat) (Perc.) Mac Key, T. petropavlovskyi Udach. & Migush. (rice wheat) and subsp. tibetanum Shao (Tibetan wheat), glumes do not adhere to kernels and they are free-threshing. In contrast, subsp. spelta (L.) Thell. (spelt), subsp. macha (Dekapr. & Menabde) Mac Key, subsp. vavilovii (Jakubz.) A. Love and subsp. yunnanense King are hulled.

The fact that T. aestivum originated from hybridization of tetraploid T. turgidum L. (Fig. 1) with diploid Aegilops tauschii Coss. (Kihara 1944; McFadden and Sears 1946) is well documented by genetic evidence. Hexaploid wheat has been resynthesized by hybridization of T. turgidum with Ae. tauschii (McFadden and Sears 1946). The synthetic wheats resemble spelt and are invariably hulled (McFadden and Sears 1946; Kihara et al. 1965; Kerber and Rowland 1974). On the basis of this and other lines of evidence, it has been argued that spelt is ancestral to the free-threshing hexaploid wheat (McFadden and Sears 1946; Kuckuck 1959; Andrews 1964). Yet, some genetic and archaeological findings suggest that spelt may have been actually derived by introgressive hybridization between free-threshing T. aestivum and hulled T. turgidum subsp. dicoccon (Schrank.) Thell. (Tsunewaki 1968; Liu and Tsunewaki 1991; Nesbitt and Samuel 1996).

Whether the various forms of T. aestivum are monophyletic or polyphyletic is also open to debate. This uncertainty is true not only for subspecies but also for varieties within subspecies, such as the various types of spelt. Originally, spelt was known only from Europe which made it a poor candidate as an ancestral form of hexaploid wheat because Ae. tauschii does not grow anywhere nearby (Schiemann 1951). Later, spelt was discovered in Iran (Kuckuck and Schiemann 1957; Kuckuck 1959) and other locations in Asia. Jaaska (1978) concluded from isozyme variation in the A genome that the European spelt originated independently of the Asian spelt. The possibility that some of the subspecies and types of the Asian hulled wheats may be polyphyletic and may or may not be directly related to the evolution of free-threshing wheat must be seriously considered (Swaminathan 1966; Tsunewaki 1968; Johnson 1972).

A question related to the above dilemma is the exact source of the wheat D genome and the geographic place of the origin of T. aestivum. Aegilops tauschii encompasses several morphological varieties which are broadly grouped into two subspecies, tauschii and strangulata (Eig) Tzvelev. The former is distributed from eastern Turkey to China and Pakistan, whereas the latter occurs in two disjoined regions, southeastern Caspian Iran and Transcaucasia (Kihara et al. 1965; Jaaska 1995). Jaaska (1978, 1981) identified polymorphisms at the aspartate aminotransferase and aromatic alcohol dehydrogenase loci between the Ae. tauschii subspecies, and in both cases the subsp. strangulata allele corresponded to that in the T. aestivum D genome. He also pointed out that the profile of esterase alleles characterizing the D genome of T. aestivum fits subsp. strangulata better than subsp. tauschii (Jaaska 1980). The spectrum of alleles at the a-amylase loci in T. aestivum also corresponds better to the allele spectrum in subsp. strangulata than to that in subsp. tauschii (Nishikawa 1974; Nishikawa et al. 1980).

Fig. 1. Phylogeny of polyploid species of Triticum inferred from variation in repeated nucleotide sequences (Dvorák et al. 1988, 1993; Dvorák and Zhang 1990, 1992). Genomes of each species are indicated by capital letters.

Variation in the lengths of restriction fragments (RFLP) at a large number of single-copy loci and the rRNA gene nontranscribed spacer at the Nor3 locus was used to investigate genetic relationships among accessions of Ae. tauschii grouped according to their geographic origin and subspecies (Dvorák et al. 1998). Accessions from Turkey and western Iran were more closely related to those from Afghanistan, Turkmenistan and China than to those from the neighboring north-central Iran and southwestern Caspian Iran (Dvorák et al. 1998). A similar observation was made earlier by Lubbers et al. (1991). Dvorák et al. (1998) suggested that these findings reflect gene migration between subsp. strangulata and subsp. tauschii in Iran and proposed that Ae. tauschii is composed of two genepools, the strangulata and tauschii genepools, which do not correspond to the morphological delimitation of the subspecies. The strangulata genepool is wider than appears on the basis of morphology and comprises subsp. tauschii in north-central Iran and Caspian Iran. This study illustrated that morphology is of dubious value in assessing intraspecific genetic relationships.

In Transcaucasia, where the strangulata and tauschii genepools overlap (Jaaska 1995), genetic distances of individual accessions to subsp. strangulata and subsp. tauschii outgroups revealed that numerous accessions were allocated to an erroneous genepool on the basis of morphology (Dvorák et al. 1998). There is less introgression between subsp. strangulata and subsp. tauschii in Transcaucasia than in Iran. Only about 5% of all Transcaucasian accessions showed intermediate genetic distances to the outgroups; the rest of the accessions showed genetic relationships characteristic of either subsp. strangulata or subsp. tauschii. This contrasts with Caspian and north-central Iran where no accessions showing genetic relationships characteristic of true subsp. tauschii were found. Reallocation of Transcaucasian accessions to the respective genepools according to genetic relationships provided strong evidence that the T. aestivum D genome was contributed by the strangulata genepool (Dvorák et al. 1980).

Jaaska (1980) considered Transcaucasia to be the center of the distribution of subsp. strangulata and hence placed the origin of T. aestivum to Transcaucasia. Nishikawa et al. (1980) found that the a-amylase isozyme profile present in T. aestivum is most prevalent in southeastern Caspian Iran, not in Transcaucasia, and therefore suggested that the most likely birthplace of T. aestivum is southeastern Caspian Iran. Tsunewaki (1966) considered mountainous Azerbaijan as the place of the origin of T. aestivum because of the distribution of the waxy-bloom allele. Nakai (1979) placed the birthplace of T. aestivum in southwestern Caspian Iran and Transcaucasia on the basis of the distribution of esterase alleles.

In spite of the existence of polymorphism at isozyme loci in Ae. tauschii, no polymorphism has been found in the T. aestivum D genome that is shared with Ae. tauschii. Even at the highly polymorphic nontranscribed spacer at the Nor3 locus encoding the 18S-5.8S-26S rRNA species (rDNA NTS), only a single Ae. tauschii allele has been detected in wheat (Clarke et al. 1989; Lagudah et al. 1991; Dvorák et al. 1998). The only potential exception to this may be the high-molecular-weight (HMW) glutenin locus Glu1. The locus is composed of two genes: x, encoding HMW-weight glutenin subunits with a slower electrophoretic mobility, and y, encoding HMW-weight glutenin subunits with a faster electrophoretic mobility. In the wheat D genome, the Glu1a haplotype encodes HMW-glutenin subunits 2 (x subunit) + 12 (y subunit). This haplotype has been detected in Ae. tauschii (Lagudah and Halloran 1988). Another, less common, haplotype in the wheat D genome is Glu1d which encodes subunits 5 (x subunit) + 10 (y subunit). A haplotype encoding similar subunits exists in Ae. tauschii (Lagudah and Halloran 1988). Two chromosomes ID carrying this haplotype have been substituted to T. aestivum, one from subsp. strangulata and one from subsp. tauschii (Jones et al. 1990, 1991). A close examination of the electrophoretic mobility of the subunits encoded by the Ae. tauschii haplotype revealed that the mobility of the HMW-glutein subunit 5 is identical in wheat and Ae. tauschii. The mobility of the wheat subunit 10 was found to be slightly slower under some electrophoretic conditions (Jones 1991). Lagudah and Halloran (1988) called this Ae. tauschii subunit 10 whereas Jones (1991) called it T5. Although Lagudah and Halloran (1988) could not distinguish the two forms of subunit 10 by electrophoretic mobility, they found that they differ in the isoelectric point. While the Glu1 locus constitutes the only case in which more than a single Ae. tauschii allele at a locus may have been detected in wheat, because of the differences between subunits 10 and T5, the evidence is hardly conclusive, as pointed out by Lagudah and Halloran (1988).

The question of whether wheat originated by single or multiple hybridization events has ramifications for all of the problems discussed earlier. It is evident that if modern T. aestivum originated by multiple hybridization events, as speculated, e.g. by Jakubziner (1958), Dekaprelevich (1961), Kuckuck (1964), Morris and Sears (1967) and Yen et al. (1983), then the source of the T. aestivum D genome and the geographic place of its origin would be intrinsically uncertain. If wheat originated by a single hybridization event, as has been concluded by Jaaska (1980) or tacitly assumed by other authors (Liu and Tsunewaki 1991; Lubbers et al. 1991), all variation in the T. aestivum D genome would have originated since the origin of T. aestivum, which archaeologists place in the 7th millenium BC (for recent review, see Nesbitt and Samuel 1996). If, however, T. aestivum originated recurrently, some of the variation of Ae. tauschii would have been introgressed into wheat. These alternatives lead to different views of the wheat genetic system and wheat evolution.

To gain insight into these questions, polymorphisms of restriction fragment lengths (RFLP) at 53 single-copy loci, rDNA NTS and the HMW-glutenin locus Glu1 were investigated with the objective of assessing parallel polymorphisms between T. aestivum and Ae. tauschii and the evolution of T. aestivum. Additionally, the electrophoretic mobility of HMW-glutenin subunits was investigated in Ae. tauschii and the T. aestivum D genome with the objective of determining whether the widespread D-genome haplotypes Glu1a and Glu1d were contributed by Ae. tauschii.

Material and methods

Plant material

RFLP was investigated in DNAs of 172 accessions of Ae. tauschii, 178 accessions of subsp. aestivum, subsp. compactum and Chinese endemic wheats (Tibetan wheat, Yunnan wheat and rice wheat), 64 accessions of spelt, 10 accessions of T. aestivum subsp. macha and 2 accessions of T. aestivum subsp. vavilovii. The spelt accessions consisted of 52 accessions of the European spelt, 1 accession of spelt from Azerbaijan, 5 accessions of spelt from Iran, 2 accessions of spelt from Tadjikistan and 4 accessions of spelt from Afghanistan (the term spelt will be used only in reference to subsp. spelta, not to all hulled hexaploid wheats). The accessions and their sources have been described earlier (Dvorák et al. 1998). Triticum aestivum accessions were grouped according to subspecies and geographic region of origin (western and eastern). Ae. tauschii accessions were grouped by subspecies and the geographic region of the origin into 11 populations (Fig. 2).

RFLP

Nuclear DNA was isolated from leaves of a single plant per accession (Dvorák et al. 1988) and digested with DraI or XbaI. Restriction endonuclease-digestion, blotting, hybridization and DNA probes have been described previously (Dvorák et al. 1998). The position of each locus and allele in the wheat genomes was confirmed by synteny mapping using nullisomic-tetrasomic stocks (Sears 1966) and in some cases disomic substitution lines harboring single chromosome pairs of Lophopyrum elongatum (Host) A. Love substituted individually for wheat homoeologous chromosome pairs (Dvorák 1980; Dvorák and Chen 1984).

Variation at 55 loci was investigated (Dvorák et al. 1998). In Ae. tauschii, RFLPs at 53 loci were investigated in DraI-digested DNAs. For the Nor3 locus, TaqI digests were employed, and for the XGlu1 locus, XbaI digests were used. For 15 of the 53 single-copy loci, RFLP was also investigated in XbaI-digested DNAs. Thus, a total of 70 enzyme x probe combinations were used for Ae. tauschii. RFLP at the same 70 enzyme x probe combinations was also investigated in all accessions of T. aestivum subsp. macha and subsp. vavilovii, a limited sample of European and Asian spelta, and reference cv. Chinese Spring. Since some loci were monomorphic in Ae. tauschii or not informative in wheat, a subpopulation of 27 informative loci was selected for the investigation of the relationships between Ae. tauschii populations and those of T. aestivum. RFLP at 21 loci was investigated with DraI, two loci with XbaI, two loci with both DraI and XbaI, one locus with TaqI, and one locus with EcoRV. Thus, a total of 29 enzyme x probe combinations were used for the 27 loci.

Fig. 2. Approximate geographic location of Ae. tauschii populations used in this study: (1) Turkey, (2) Transcaucasia subsp. strangulata, (3) Transcaucasia subsp. tauschii, (4) western Iran, (5) southwestern Caspian Iran, (6) north-central Iran, (7) southeastern Caspian Iran subsp. strangulata, (8) southeastern Caspian Iran subsp. tauschii, (9) Turkmenistan, (10) Afghanistan and (11) China.

Shared polymorphism between Ae. tauschii and T. aestivum

Dvorák et al. (1998) reported polymorphisms in the T. aestivum D genome shared with Ae. tauschii at five loci. To scrutinize further the correspondence of these shared polymorphisms, several accessions of Ae. tauschii and T. aestivum representing each shared DraI or XbaI haplotype were selected and digested with either all or a subset of the following restriction endonucleases; ApaI, BglII, EcoRV, KpnI, SstI and XbaI. Southern blots were hybridized with clones detecting DNA fragments from these five loci. DNAs of relevant Chinese Spring nullisomic-tetrasomics and L. elongatum disomic substitution lines in the Chinese Spring genetic background were included in the blots. Restriction fragments belonging to the D-genome were identified in each restriction digest by comparing the profile of Chinese Spring with those of nulli-tetrasomics and disomic substitution lines.

HMW-glutenins

The endosperms of 222 accessions of Ae. tauschii and 25 accessions of T. aestivum were crushed, proteins were extracted and electrophoretically fractionated in denaturing polyacrylamide gels as described by Cole et al. (1981) and stained in 0.025% Coomassie brilliant blue R250 in 12% trichloracetic acid (TCA) for 24 hours and destained in 12% TCA for 2 days.

Results

Polymorphism sharing between wheat and Ae. Tauschii

Polymorphisms were encountered in the D genome of T. aestivum at 14 of the 70 enzyme x probe combinations representing 55 loci (Table 1). At five loci - Xpsr899, Xpsr928, Xpsr666, Xpsr901 and Glu1 - two Ae. tauschii haplotypes were found (Table 1). At 10 loci, haplotypes were found in T. aestivum which were not found in Ae. tauschii; a total of 18 haplotypes not encountered in Ae. tauschii were found in T. aestivum (Table 1). The possibility that some of these 18 T. aestivum-specific haplotypes are actually shared with Ae. tauschii but were not present among the Ae. tauschii accession investigated here cannot be discounted.

The Xpsr899 locus was investigated with four restriction endonucleses. DNAs of T. aestivum and Ae. tauschii accessions having DraI haplotype a shared restriction fragments in all digests (Fig. 3). The same was true for accessions sharing haplotype b (Fig. 3). In all four digests, the fragments of the a haplotypes differed from those of the b haplotypes (Fig. 3), indicating that haplotype a differs from haplotype b in a number of restriction sites in both species. By these criteria, the T. aestivum haplotype a appears to be identical to haplotype a in Ae. tauschii, and T. aestivum haplotype b appears to be identical to haplotype b in Ae. tauschii.

The Xpsr928 locus was investigated with six restriction endonucleases. Iranian and Azerbijan spelt and two accessions of Ae. tauschii represented the DraI haplotype b and Chinese Spring and two accessions of Ae. tauschii represented the DraI haplotype a (Fig. 4). Identical restriction fragments were shared by DNAs of the T. aestivum and Ae. tauschii accessions representing the DraI haplotype b in all six digests. However, DNAs of the T. aestivum and Ae. tauschii accessions representing haplotype a shared the same fragments only in the KpnI and DraI digests (Fig. 4) suggesting that T. aestivum haplotype a is related but not identical to haplotype a in Ae. tauschii accessions KU2103 and KU2824. Actually, the two Ae. tauschii accessions also did not have identical haplotypes, as evidenced by polymorphism between them in the BglII and KpnI profiles (Fig. 4).

The Xpsr666 locus was investigated with a total of seven restriction endonuclases. Chinese Spring and Ae. tauschii accessions KU2103 and KU2824 represented the XbaI haplotype a and Tadjikistan spelt VIR56569 and Ae. tauschii accessions KU2103 and KU2824 represented the XbaI haplotype b. No fragment belonging to the D genome was seen in the SstI digest and no polymorphism between haplotypes a and b was observed in the BglII and EcoRV digests. In the remaining four digests, haplotype a differed from haplotype b and the differences were the same in both species.

The Xpsr901 locus was investigated with six restriction endonucleases. The D-genome restriction profiles in Chinese Spring and spelt PI367199 were identical to those in Ae. tauschii accessions KU2104 and KU2103 representing DraI haplotype a in all digests. The D-genome profiles in spelt PI90962 and PI347926 and Ae. tauschii accessions KU2112 and KU2151 representing DraI haplotype b shared the same DNA fragments in the DraI and SstI digests but not in the ApaI and BglII digests. In four digests - ApaI, BglII, DraI and SstI - T. aestivum a and b haplotypes differed/They were identical in the EcoRV and KpnI digests but the Ae. tauschii accessions representing haplotypes a and b differed. Because of the incomplete agreement between the DraI b haplotypes in wheat and Ae. tauschii accessions KU2112 and KU2151, all 85 Ae. tauschii having DraI haplotype b were digested with ApaI and BglII and Southern blots were hybridized with PSR901. Of the 85 Ae. tauschii accessions, only two, AL10/80-1 and AL10/80-2 collected in Nakhichewan by V. Jaaska (pers. comm.), had the same haplotype as spelt accessions PI90962 and PI347926. Both accessions belong to the tauschii genepool (Dvorák et al. 1998). Thus, there is a complete agreement between the DraI a haplotypes in T. aestivum and Ae. tauschii and between the DraI b haplotypes in T. aestivum and Ae. tauschii in all investigated restriction sites.

Table 1. Multiple allelisms in the D-genome of T. aestivum and the genome of Ae. tauschii and polymorphism sharing between T. aestivum^† and Ae. tauschii^‡ in DNAs digested with DraI or a restriction endonuclease indicated in parentheses.

Groups of accessions

Xmwg2031

Xcdo 1400

Xpsr371-6D

XGlu1 (Xbal)

Xpsr899

Xbcd 1302

XEsi3

Xpsr666 (Xbal)

Xpsr102

Xpsr 901-2D

Xpsr 928

Xwg 644

XGsp

Xcdo 749

Ae. tauschii

a,b

b,c,d,e

a,b,c

a,b,c,d

a,b,c,d,e,f,g

b,c,d,e,f

b,c,d

a,b,c

a,b

a,b

a,b

a,b,c

b,c,e,f

a,b,c,d

wheat west reg.

a

a,c

a

a,b

a,f,h,i

a,g

d

a

a,c

a,b

a,b

a

a

a

wheat east reg.

a

a,c

a,d,e

-

a,f

a,c

a,d

a,b

a,d

a,b

a,b

a

a

a

subsp. macha

a

c

a

a,b

a,f,j,k

a

a,d

a

a,d

a,b

a,b

a

a

a,e

spelt (Asia)

a,c

c

a

a

a,f

a,c

d

a,b

a,d

a

a,b

a

a

a

spelt (Europe)

a

c,f

a

a

a,f

a

d

a

a,c,d

a,b

a

a,b

a

a

subsp. vavilovii

a

c

a

a

a,f

a

a,d

a

d

a

a

a

a

a

^† T. aestivum haplotypes which were not found in Ae. tauschii are in bold.
^‡ Pairs of Ae. tauschii haplotypes shared with T. aestivum are underlined.

The Glu1 locus was investigated with five restriction endonucleases. Triticum aestivum was polymorphic in the XbaI and EcoRV digests. The same DNA fragments were observed in Ae. tauschii (Table 1). In Ae. tauschii, the XbaI haplotypes a and b had frequencies 0.21 and 0.07, respectively, and the EcoRV haplotypes a and b had frequencies 0.94 and 0.06, respectively (not shown) indicating that the b haplotype is rare in Ae. tauschii. Of 172 Ae. tauschii accessions, XbaI haplotype b was found in 12 accessions and EcoRV haplotype b was found in 10 accessions (Table 2). Of these, eight shared the two haplotypes, indicating that the XbaI and EcoRV b haplotypes are in a strong linkage disequilibrium in Ae. tauschii.

Fig.4. Autoradiograms of Southern blots of DNAs of Ae. tauschii and T. aestivum accessions and Chinese Spring cytogenetic stocks digested with DraI (left) and KpnI (right) restriction endonucleases and hybridized with pPSR928. Accessions sharing haplotype a and those sharing haplotype b are indicated at the bottom. Genome allocation of major bands is indicated. Note the absence of the Chinese Spring fragments assigned to chromosome 2D in the disomic substitution line 2E(2D). Note also that the D-genome DNA fragments of the Ae. tauschii and T. aestivum accessions with haplotype a differ from those shared by Ae. tauschii and T. aestivum accessions with haplotype b in both digests.

Seven types of the x subunit of HMW-glutenin and eight types of the y subunit were found with SDS-PAGE (not shown). Most of the Ae. tauschii accessions with the DNA haplotype b had the HMW-glutenin x subunit 5 (Table 2, Fig. 5). The haplotype encoding subunit 5 was found to be rare in Ae. tauschii, frequency (f) = 0.06 (Table 3).

None of the accessions with the b DNA haplotype had the x subunit 2 or any other x-type subunit frequent in Ae. tauschii. Thus, the XbaI and EcoRV haplotypes b and the Glu1x allele encoding subunit 5 usually occur together and must be in a strong linkage disequilibrium in Ae. tauschii.

The same linkage disequilibrium exists in T. aestivum. Seed storage proteins were extracted from seeds of 25 accessions of the Asian and European spelt, subsp. macha, subsp. vavilovii and bread wheat cultivars of both western and eastern origin and fractionated by SDS-PAGE. Eight accessions had HMW-glutenin subunits 5 + 10 (haplotype Glu1d) and 17 had HMW-glutenin subunits 2 + 12 (haplotype Glu1a). All Glu1d accessions had EcoRV haplotype b and all Glu1a had EcoRV haplotype a (XbaI digests were not investigated because it was difficult to score the XbaI fragment in the wheat genetic background). These parallels between the T. aestivum D genome and Ae. tauschii showed that T. aestivum Glu1a and Glu1d haplotypes encoding HMW-glutenin subunits 2 + 12 and 5 + 10, respectively, were both contributed to T. aestivum by Ae. tauschii.

In Ae. tauschii, the Glu1a haplotype encoding HMW-glutenin subunits 2 + 12 occurs in Transcaucasia, the southeastern Caspian region and north-central Iran. The Glu1a haplotype was observed only in the strangulata genepool (Table 3). The haplotype encoding the pair of subunits 5 + T5, which corresponds to the wheat haplotype Glu1d (Fig. 5), was also observed only in the strangulata genepool; its highest frequency was in southwestern Caspian Iran (Table 3). Likewise, the x gene encoding subunit 5 occurred in the highest frequency in southwestern Caspian Iran (Table 3).

Discussion

Multiple contributions of Ae. tauschii to the wheat D genome

Pairs of haplotypes shared by Ae. tauschii and the T. aestivum D genome were found at five of the 55 investigated loci. Within each pair, haplotypes differed in multiple restriction sites. If these were merely restriction site differences or small independent insertions or deletions, it would be extremely unlikely that these parallel polymorphisms originated by reverse mutations in wheat. Several reverse mutations would have to occur in each case to convert one haplotype into the other, which is very unlikely. If, however, each of the polymorphisms is due to a single large DNA insertion that occurred in Ae. tauschii, an excision of the inserted DNA during wheat evolution could convert a haplotype into the ancestral one and, thus, generate a shared haplotype pair between wheat and Ae. tauschii.

This scenario can be discounted for the shared polymorphism at the Glu1 locus because the Glu1a and Glu1d haplotypes differ by several independent mutations. Should the difference between the wheat Glu1a and Glu1d haplotypes be due to reverse mutations, a minimum of three independent reverse mutations, HMW-glutenin subunit 5 to 2 (or vice versa), HMW-glutenin subunit 10 to 12 (or vice versa) and EcoRV haplotype b to a (or vice versa) would have to have occurred during wheat evolution. This is extremely unlikely.

Three of the five loci in which polymorphisms are shared between T. aestivum and Ae. tauschii are linked on chromosome 2D. Two, Xpsr666 and Xpsr928, are on the short arm and one, Xpsr901, is closely linked to them on the long arm (for map positions of these loci see Dubcovsky et al. 1996). Linkage among the polymorphic loci would hardly be expected if the polymorphisms were caused by reverse mutations, since these should be random. An intriguing coincidence is that chromosome 2D played a critical role in the evolution of the free-threshing T. aestivum. The free-threshing character is based on two mutations, the dominant mutation of q to Q on the long arm of chromosome 5A, and a recessive mutation of Tg to tg at the end of the short arm of chromosome 2D (Kerber and Rowland 1974). Chromosome 2D was substituted from four accessions of Ae. tauschii into bread wheat cv. Chinese Spring, which is QQtgtg. All four disomic substitution lines had adhering glumes in spite of having Q allele (J. Dvorák, unpublished). Clearly, Q itself is insufficient to cause a free-threshing habit. Therefore, the evolution of the free-threshing habit required fixation of a recessive tg mutation in T. aestivum. One can imagine a scenario in which a Tg to tg mutation happened more than once. Selection for the tg alleles could maintain hitchhiking linked polymorphisms on chromosome 2D in the wheat genepool.

Table 2. Lists of Ae. tauschii accessions (out of 172 investigated) having the rare XbaI and EcoRV haplotypes b at the Glu1 locus and the HMW-glutenin subunits encoded at the Glu1 locus in these accessions.

Xbal	EcoRV	HMW-glutenin subunits
KU20-10	KU20-10	5 + T5
KU2104	KU2104	5 + 12
KU2090	KU2090	5 + T5
	KU2106	5 + null
KU2160	KU2160	5 + T5
KU2110	KU2110	5 + 12
	KU2001	3 + 12
AL8/78-2	AL8/78-2	T6 + T8
AL9/78-3, -4, -6, -7		T6 + T5
AL10/80-1, -2	AL10/80-1, -2	5 + T9

Fig. 5. SDS-PAGE profiles in indicated Ae. tauschii accessions (KU), disomic substitution line in which the Cheyenne 1D chromosome was substituted for 1D of Chinese Spring (DSCnn1D), and Chinese Spring (CS). Subunits encoded by Glu1 haplotypes are indicated in parentheses. The x HMW-glutenin subunits (2 and 5) and y HMW-glutenin subunits (10, T5, and 12) are indicated by arrowheads. Chinese Spring profile shows subunits 2 + 12 (haplotype Glu1a), the profile of DSCnn1D shows subunits 5 + 10 (haplotype Glu1d), the profile of Ae. tauschii accessions KU2122 shows subunits 5 + T5, the profile of Ae. tauschii KU2121 shows subunits 2 + T5, and that of Ae. tauschii KU2110 shows subunits 5 + 12. Note that the mobility of subunit 5 in wheat and Ae. tauschii is identical. Likewise the mobility of subunits 10 and T5 is indistinguishable from each other in this gel.

Fig. 6. Phenograms produced by the neighbor-joining method using genetic distances based on 55 loci (top) and selected 27 loci (bottom). The magnitude of divergence between groups was computed as Nei's genetic distance D (Nei 1978) with the GDA program (Lewis and Zaykin 1997). The phenograms were constructed with the GDA program. A scale showing a Nei's genetic distance is shown. The phenogram based on 27 loci is longer than that based on 55 loci because it involves only highly informative loci (from Dvorák et al. 1998). In each phenogram, T = subsp. tauschii and S = subsp. strangulata.

Table 3. Frequencies of haplotypes encoding the indicated pairs of HMW-glutenin subunits in Ae. tauschii accessions grouped by geographic region and botanical subspecies (the frequencies were zero in the remaining regions).

Geographic region	Subspecies	2 + 12	5 + T5	5 + any subunit
Transcaucasia	tauschii	0.00	0.10	0.10
Transcaucasia	strangulata	0.29	0.00	0.03
Southwest Caspian Iran	tauschii	0.00	0.14	0.36
Southeast Caspian Iran	tauschii	0.00	0.00	0.00
Southeast Caspian Iran	strangulata	0.33	0.04	0.10
North-central Iran	tauschii	0.33	0.00	0.00

Sources of the wheat D genome and the geographic origin of T. Aestivum

Most of the evidence accumulated thus far suggests that the Triticum aestivum D genome is more related to the strangulata genepool than to the tauschii genepool (Nishikawa 1974; Jaaska 1978, 1980, 1981; Nakai 1979; Nishikawa et al. 1980; Lagudah et al. 1991, Dvorák et al. 1998). On a geographic region basis, the D genomes of all forms of T. aestivum were found to be most closely related to accessions of the strangulata genepool collected in Transcaucasia and southwestern Caspian Iran (Dvorák et al. 1998; see Figure 6). On the morphological basis, the former accessions belong to subsp. strangulata but the latter, collected by Kihara et al. (1965) in the coastal southwestern Caspian Iran, belong to subsp. tauschii. The latter population is composed of morphological varieties meyeri and typica (Kihara et al. 1965). On the genetic basis, however, there is no appreciable genetic distinction between the two varieties in this region and both belong to the strangulata genepool (Dvorák et al. 1998). Therefore, characteristics attributed to var. meyeri are pertinent to the entire southwestern Caspian population. Lagudah et al. (1991) failed to find the Nor3a restriction pattern, for which wheat is monomorphic (Clarke et al. 1989; Lagudah et al. 1991; Dvorák et al. 1998), and the Glu1a haplotype in var. meyeri. While the Nor3a restriction pattern does occur, and in a high frequency, in southwestern Caspian Iran (Dvorák et al. 1998), the Glu1a haplotype was indeed not found in that region (present data). The Glu1a haplotype is present in the Transcaucasian and southeastern Caspian subsp. strangulata. The Glu1x gene encoding the HMW-glutenin subunit 5 occurs with a high frequency in Transcaucasia and southwestern Caspian Iran and the Glu1 haplotype encoding HMW-glutenin subunits 5 + T5, which is ancestral to wheat haplotype Glu1d, occurs with the highest frequency in southwestern Caspian Iran. Thus, variation at the Nor3 and Glu1 loci is consistent with inferences based on genetic distances (Dvorák et al. 1998) indicating that the strangulata genepool in Trancaucasia and southwestern Caspian is the most likely source of the T. aestivum D genome. Earlier, Tsunewaki (1966) and Nakai (1979) placed the origin of T. aestivum to southwestern Caspian Iran and the neighboring mountainous Azerbaijan on the basis of the distribution of the waxy bloom and esterase alleles and Jaaska (1981) to Transcaucasia on the basis of the distribution of aspartate aminotranferese and aromatic alcohol dehydrogenase alleles.

To investigate individual regions of Transcaucasia, Transcaucasian accessions were divided into four geographic regions: Georgia, Armenia, Nakhitshevan and Azerbaijan (Dvorák et al. 1998). The D genome of T. aestivum appeared to be most closely related to the strangulata genepool in Armenia (Dvorák et al. 1998). Most of the Armenian accessions were collected in the vicinity of Jerevan and along the Razdan River (Kihara et al. 1965; V. Jaaska, pers. comm.). Although genetic distances point to Armenia, other areas in Transcaucasia and potentially western coastal Caspian Iran may have played a role in the evolution of the T. aestivum D genome as well (Dvorák et al. 1998). First, differences among genetic distances between wheat and strangulata genepool accessions grouped by geographic regions in Transcaucasia and southwestern Caspian Iran are not large. Second, it was shown here that multiple Ae. tauschii sources contributed to the evolution of the T. aestivum D genome. There is no reason to suppose that these multiple Ae. tauschii sources were from a single geographic area or, in fact, that they all were from only the strangulata genepool. One of the five loci at which wheat and Ae. tauschii share polymorphism is the Xpsr901 locus. Only two of the 172 investigated Ae. tauschii accessions matched the restriction sites of the T. aestivum haplotype b. Both came from a single population from Nakhichean and both belonged to the tauschii genepool. Third, it is not known to what extent the distribution and genetic structure of the present-day populations of Ae. tauschii were affected by agriculture, particularly the cultivation of wheat in Transcaucasia and the nearby Iranian regions.

Genetic distances calculated from RFLP in the D genome suggest a very close relationship of both western and eastern accessions of bread wheat with the Asian spelt (Dvorák et al. 1998). These distances are about a tenth of the genetic distances between bread wheat and the European spelt (Dvorák et al. 1998). This finding agrees with Jaaska's (1978) conclusion that the European spelt differs from the Asian spelt. The D genome of the European spelt appears, in turn, more related to bread wheat than is that of subsp. macha and subsp. vavilovii (Dvorák et al. 1998). Although these findings could be interpreted to mean that bread wheat evolved from the Asian spelt, they could be equally well interpreted to mean that the Asian spelt was derived recently from bread wheat by mutation or hybridization of free-threshing bread wheat with hulled tetraploid wheats (see Introduction). Additional work is therefore needed to resolve this dilemma. However, monophylesis of all T. aestivum groups in the dendogram based on variation in the D genome (Fig. 6) and the observation that many polymorphisms unique to T. aestivum are shared among the various forms of T. aestivum (Table 1) strongly suggests that all forms of T. aestivum likely have evolved from a common hexaploid genepool (Dvorák et al. 1998).

Agriculture appears in Transcaucasia in the 6th millenium (Mellaart 1975). Findings of free-threshing hexaploid wheat in archaeological sites in Anatolia between 6000 and 7000 BC (Hillman 1978; de Moulins 1993) require rapid evolution of hexaploid wheat upon the arrival of wheat cultivation to Transcaucasia. If this timetable is correct, the antiquity of this hexaploid genepool would conceptually provide sufficient time for the evolution of modern free-threshing and hulled forms of T. aestivum by mutation or hybridization from this ancient genepool.

There are two basic scenarios by which the D-genome genepool could have been formed. Several amphiploids could have originated and their hybridization and recombination could have led to the evolution of a single genepool. Alternatively, a single amphiploid could have originated and founded a hexaploid population. Plants in this hexaploid population could have hybridized with Ae. tauschii and the hybrids could have been a bridge for geneflow from Ae. tauschii to T. aestivum. This scenario seems unlikely. While hybridization between tetraploid wheat and Ae. tauschii is easy and fertile amphiploids are produced by self-pollination of triploid hybrids owing to high production of unreduced gametes (Kihara et al. 1950), hybridization between hexaploid wheat and Ae. tauschii is difficult and production of hybrid plants usually requires embryo rescue. Therefore, the recurrent appearance of hexaploid amphiploids in the fields of tetraploid wheat or mixed tetraploid/hexaploid wheat was a likely source of geneflow from Ae. tauschii to the hexaploid genepool.

If multiple amphiploids contributed to the formation of the genepool ancestral to all present-day forms of T. aestivum, it is puzzling why shared polymorphisms between T. aestivum and Ae. tauschii are rare. An obvious example is rRNA gene locus Nor3 which is highly polymorphic in Ae. tauschii but appears to be monomorphic in T. aestivum (Lagudah et al. 1991; Dvorák et al. 1998). Since the Nor3a haplotype is rare in Ae. tauschii it is unlikely that this haplotype was present in every amphiploid that was involved in the formation of T. aestivum. This contradiction can be accounted for if it is assumed that the D-genome genepool was subjected to significant evolution prior to the differentiation of the modern forms of T. aestivum. The dates of the appearance of free-threshing wheat in archaeological sites provide ample time for this phase of T. aestivum evolution. During this phase, some polymorphic loci may have become monomorphic. Once the hexaploid genepool became large, most alleles contributed to the genepool by subsequent amphiploids would have a general tendency to be lost, unless they were selected for or hitchhiked with selected alleles. New amphiploids would be particularly disadvantaged in the fields of free-threshing wheat because their adhering glumes would tend to eliminate them during threshing.

Acknowledgments

This project is a contribution to the International Triticeae Mapping Initiative (ITMI). The authors acknowledge a grant received from the US Department of Agriculture -National Research Initiative Competitive Grant Program No. 96-35300-3822 to J. Dvorák and multi-investigator grant 92-37310-7664 from DOE/NSF/USDA Joint Program on Collaborative Research in Plant Biology, and a gift from Anheuser-Busch Co.

References

Andrews, A.C. 1964. The genetic origin of spelt and related wheats. Zuchter 34:17-22.

Clarke, B.C., L.B. Moran and R. Appels. 1989. DNA analyses in wheat breeding. Genome 32:334-339.

Cole, E.W., J.G. Fullington and D.D. Kasarda. 1981. Grain protein variability among species of Triticum and Aegilops: quantitative SDS-PAGE studies. Theor. Appl. Genet. 60:17-30.

de Moulins, D. 1993. Les restes de plantes carbonisées de Cafer Hoyuk. Cahiers de l'Euphrate 7:191-234 [in French].

Dekaprelevich, L.L. 1961. Die Art Triticum macha Dek et Men. im Lichte neuster Untersuchungen uber die Herkunft der Hexaploiden Weizen. Z. Pflanzenzuchtg. 45:17-30.

Dubcovsky, J., M.C. Luo, G.Y. Zhong, R. Bransteitter, A. Desai, A. Kilian, A. Kleinhofs and J. Dvorák. 1996. Genetic map of diploid wheat, Triticum monococcum L, and its comparison with maps of Hordeum vulgare L. Genetics 143:983-999.

Dvorák, J. 1980. Homoeology between Agropyron elongatum chromosomes and Triticum aestivum chromosomes. Can. J. Genet. Cytol 22:237-259.

Dvorák, J. and H.B. Zhang. 1990. Variation in repeated nucleotide sequences sheds light on the phylogeny of the wheat B and G genomes. Proc. Nat. Acad. Sci. USA 87:9640-9644.

Dvorák, J. and H.B. Zhang. 1992. Reconstruction of the phylogeny of the genus Triticum from variation in repeated nucleotide sequences. Theor. Appl. Genet. 84:419-429.

Dvorák, J. and K.C. Chen. 1984. Phylogenetic relationships between chromosomes of wheat and chromosome 2E of Elytrigia elongata. Can. J. Genet. Cytol. 26:128-132.

Dvorák, J., M.-C. Luo, Z.-L. Yang and H.-B. Zhang. 1998. The structure of Aegilops tauschii genepool and the evolution of hexaploid wheat. Theor. Appl. Genet, (in press).

Dvorák, J., P. di Terlizzi, H.B. Zhang and P. Resta. 1993. The evolution of polyploid wheats: Identification of the A genome donor species. Genome 36:21-31.

Dvorák, J., P.E. McGuire and B. Cassidy. 1988. Apparent sources of the A genomes of wheats inferred from the polymorphism in abundance and restriction fragment length of repeated nucleotide sequences. Genome 30:680-689.

Hillman, G.C. 1978. On the origins of domestic rye - Secale cereale: the finds from aceramic Can Hasan III in Turkey. Anatolian Studies 28:157-174.

Jaaska, V. 1978. NADP-dependent aromatic alcohol dehydrogenase in polyploid wheats and their relatives. On the origin and phylogeny of polyploid wheats. Theor. Appl. Genet. 53:209-217.

Jaaska, V. 1980. Electrophoretic survey of seedling esterases in wheats in relation to their phylogeny. Theor. Appl. Genet. 56:273-284.

Jaaska, V. 1981. Aspartate aminotransferase and alcohol dehydrogenase isozymes: Intraspecific differentiation in Aegilops tauschii and the origin of the D genome polyploids in the wheat group. Plant System. Evol. 137:259-273.

Jaaska, V. 1995. Isoenzymes in the evaluation of germplasm diversity in wild diploid relatives of cultivated wheat. Pages 247-257 in Biodiversity and Wheat Improvement (A.B. Damania, ed.). John Wiley, Chichester, UK.

Jakubziner, M.M. 1958. New wheat species. Pages 207-217 in Proc. 1st Int. Wheat Genetics Symposium (B.C. Jenkins, ed.). Public Press, Winnipeg, Canada.

Johnson, B.L. 1972. Protein electrophoretic profiles and the origin of the B genome of wheat. Proc. Nat. Acad. Sci. USA 69:1398-1402.

Jones, S.S. 1991. Chromosome ID in wheat (Triticum aestivum L.): Linkage map, quantitative traits, and effects of telosomes on recombination values. PhD Thesis. Department of Agronomy & Range Science, University of California, Davis, USA.

Jones, S.S., J. Dvorák and C.O. Qualset. 1990. Linkage relations of Gli-D1, Rg2, and Lr21 on the short arm of chromosome 1D in wheat. Genome 33:937-940.

Jones, S.S., J. Dvorák, D.R. Knott and C.O. Qualset. 1991. Use of double-ditelosomic and normal chromosome 1D recombinant substitution lines to map Sr33 on chromosome arm 1DS in wheat. Genome 34:505-508.

Kerber, E.R. and G.G. Rowland. 1974. Origin of free threshing character in hexaploid wheat. Can. J. Genet. Cytol. 16:145-154.

Kihara, H. 1924. Cytologische und genetische Studien bei wichtigen Getreidearten mit besonderer Rucksicht auf das Verhalten der Chromosomen und die Sterilitat in den Bastarden. Mem. Coll Sci. Univ. Kyoto Ser. B 1:1-200 [in German].

Kihara, H. 1944. Discovery of the DD-analyser, one of the ancestors of Triticum vulgare (Japanese). Agric. & Hort. (Tokyo) 19:13-14.

Kihara, H., H. Yamashita and M. Tanaka. 1965. Morphologic, physiological, genetical, and cytological studies in Aegilops and Triticum collected in Pakistan, Afghanistan, Iran. Results of the Kyoto University scientific expedition to the Korakoram and Hidukush in 1955. Pages 4-41 in Cultivated Plants and their Relatives (K. Yamashita, ed.). Kyoto, Japan.

Kihara, H., M. Okamoto, M. Ikegami, J. Tabushi, H. Suemoto and Y. Yamane. 1950. Morphology and fertility of five new synthesized hexaploid wheats. Seiken Ziho 4:127-140.

Kuckuck, H. 1959. Neuere Arbeiten zur Entstehung der hexaploiden kulturweizen. Z. Pflanzenzuchtg. 41:205-226 [in German].

Kuckuck, H. 1964. Experimentelle Untersuchungen zur Entstehung der Kulturweizen. Z. Pflanzenzuchtg. 51:97-140 [in German].

Kuckuck, H. and E. Schiemann. 1957. Uber das Vorkommen von Speltz und Emmer (Triticum spelta L. und T. dicoccum Schubl.) im Iran. Z. Pflanzenzuchtg. 38:383-396 [in German].

Lagudah, E.S. and G.M. Haloran. 1988. Phylogenetic relationships of Triticum tauschii the D genome donor to hexaploid wheat. 1. Variation in HMW subunits of glutenin and gliadins. Theor. Appl. Genet. 75:592-598.

Lagudah, E.S., R. Appels and D. McNeil. 1991. The Nor-D3 locus of Triticum tauschii; natural variation and genetic linkage to markers in chromosome 5. Genome 34:387-395.

Lewis, P.O. and D. Zaykin. 1997. Genetic data analysis: Computer program for the analysis of allelic data. Version 1.0. GDA Home Page at http://chee.unm.edu/gda.

Liu, Y.-G. and K. Tsunewaki. 1991. Restriction fragment length polymorphism (RFLP) analysis in wheat, II. Linkage analysis of the RFLP sites in common wheat. Jpn. J. Genet. 66:617-633.

Lubbers, E.L., K.S. Gill, T.S. Cox and B.S. Gill. 1991. Variation of molecular markers among geographically diverse accessions of Triticum tauschii. Genome 34:354-361.

Mac Key, J.M. 1966. Species relationship in Triticum. Pages 237-275 in Proc. 2nd Int. Wheat Genet. Symp. (J.M. Mac Key, ed.), Hereditas, Lund, Sweden.

McFadden, E.S. and E.R. Sears. 1946. The origin of Triticum spelta and its free-theshing hexaploid relatives. J. Hered. 37:81-89, 107-116.

Mellaart, J. 1975. The Neolithic of the Near East. Thames and Hudson, London, UK.

Morris, R. and E.R. Sears. 1967. The cytogenetics of wheat and its relatives. Pages 19-87 in Wheat and Wheat Improvement (K.S. Quisenberry and L.P. Reitz, eds.). Am. Soc. Agronomy, Madison, Wisconsin, USA.

Nakai, Y. 1979. Isozyme variation in Aegilops and Triticum, IV. The origin of the common wheats revealed from the study on esterase isozymes in synthesized hexaploid wheats. Jpn. J. Genet. 54:175-189.

Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590.

Nesbitt, M. and D. Samuel. 1996. From staple crop to extinction? The archaeology and history of hulled wheats. Pages 41-100 in Hulled Wheats. Promoting the Conservation and Use of Underutilized and Neglected Crops. 4. Proc. of the First International Workshop on Hulled Wheats, 21-22 July 1995, Castelvecchio Pascoli, Tuscany, Italy (S. Padulosi, K. Hammer and J. Heller, eds.). IPGRI, Rome, Italy.

Nishikawa, K. 1974. Alpha-amylase isozymes and phylogeny of hexaploid wheat. Pages 851-855 in Proc. 4th Int. Wheat Genetics Symposium (G. Kimber, ed.). Columbia, Missouri, USA.

Nishikawa, K., Y. Furuta and T. Wada. 1980. Genetic studies on alpha-amylase isozymes in wheat. III. Intraspecific variation in Aegilops squarrosa and birthplace of hexaploid wheat. Jpn. J. Genet. 55:325-336.

Schiemann, E. 1951. New results on the history of cultivated cereals. Heredity 5:312-314.

Sears, E.R. 1966. Nullisomic-tetrasomic combinations in hexaploid wheat. Pages 29-44 in Chromosome Manipulations and Plant Genetics (R. Riley and K. R. Lewis, eds.). Oliver & Boyd, Edinburgh, UK.

Swaminathan, M.S. 1966. Mutational analysis of the hexaploid Triticum complex. Pages 418-437 in Proc. 2nd Int. Wheat Genet. Symp. (J.M. Mac Key, ed.). Hereditas Suppl. 2, Lund, Sweden.

Tsunewaki, K. 1966. Comparative gene analysis of common wheat and its ancestal species. II. Waxiness, growth habit and awnedness. Jpn. J. Bot. 19:175-229.

Tsunewaki, K. 1968. Origin and phylogenetic differentiation of common wheat revealed by comparative gene analysis. Pages 71-85 in Proc. 3rd Int. Wheat Genetic Symposium (K.W. Finley and K.W. Shepherd, eds.). Canberra, Australia.

Yen, C., J.L. Yang, X.D. Liu and L.R. Li. 1983. The distribution of Aegilops tauschii Cosson in China and with reference to the origin of the Chinese common wheat. Pages 55-58 in Proc. 6th Int. Wheat Genetics Symposium (S. Sakamoto, ed.). Kyoto, Japan.

Introgression of Durum into Wild Emmer and the Agricultural Origin Question - M.A. Blumler

Introduction

The transition from hunting/gathering to farming was a crucial watershed in human history, with enormous ramifications continuing right down to the present day (Diamond 1997a, 1997b). In recent decades, a highly productive burst of archaeological investigations into the origins of agriculture, complemented by equally fruitful cytogenetic studies of crop phylogeny, has shed considerable light on this crucial time period. Nonetheless, our understanding of the reasons for the transition to agriculture remains limited, and subject to legitimate debate (Blumler and Byrne 1991). A major problem is that archaeologists have found it extremely difficult to locate sites that display the transition from the use of wild plants and animals to the raising of domesticated species (Blumler 1992a). Most sites show hunting/gathering only, farming only, or hunting/gathering followed abruptly by an agricultural economy that is clearly at least in part intrusive from elsewhere (e.g. Hillman et al. 1989). For instance, Pickersgill and Heiser (1977:829) pointed out that the famous site of Tehuacan, Mexico, displays “domesticates but not domestication.” That is, domesticated crops apparently filtered into Tehuacan from some other place, where their wild progenitors previously had been taken under cultivation and domesticated. This is true not only of Tehuacan, but of all or almost all other archaeological sites as well (Smith's 1997 re-analysis of the Oaxacan squash remains may be an exception). Thus, we have a great deal of information about the spread of agriculture, but not about its inception. Until archaeologists locate the sites where domestication actually occurred, where hunter-gatherers made the decision to farm, independently (i.e. free from the influence of already existing farmers), it will be difficult to move beyond speculation concerning causes.

This difficulty was largely unanticipated, because archaeologists (and a few influential crop geneticists) had assumed that most crops were domesticated independently many times by many different small human populations. Consequently, it should not be so difficult to locate a site showing the complete transition from wild to domesticated organism. For instance, Harlan (1975) asserted that emmer wheat (Triticum dicoccum) no doubt was domesticated many times in various parts of the Fertile Crescent. This assumption rested upon a theoretical stance that Harlan, and most others concerned with agricultural origins, adopted in the 1950s in the context of a highly polarized debate over the importance of diffusion (the people on the 'other side' in this debate were geographers, notably Carl Sauer). It is obvious from Harlan's writings on the subject that this was one of those debates that generated far more heat than light, with scurrilous charges of racism being bandied about in attempts to discredit opposing viewpoints. Sadly, as often happens in such circumstances, scholars became locked into extreme and inflexible viewpoints (Blumler 1993, 1996). Only recently, as a profusion of new, cytogenetic and distributional evidence concerning the wild progenitors of crop species has become available, has it become possible to re-examine the issue dispassionately (Zohary 1989; Blumler 1992a). This new evidence demonstrates very clearly that, in most regions where agriculture began, primary crops were domesticated only once or a very few times (Table 1). Moreover, while Sauer's (1952) specific hypothesis concerning agricultural origins is probably invalid, his underlying premise that diffusion is far more important than independent invention is undeniably correct (Blumler 1996).

On the one hand, that so many primary crops were domesticated only once or twice makes the archaeologist's search for the agricultural transition analogous to attempting to find a needle in a haystack. On the other hand, it also means that it should be possible to narrow down the location of domestication by cytogenetic analysis, whereupon archaeological exploration might become more successful. With a single origin, only a small subset of the wild genepool is involved in domestication. If the progenitor genotype(s) is geographically restricted, as is often the case, a plausible center of origin can be delineated. For instance, Gepts (1990, this volume) has concluded that the common bean (Phaseolus vulgaris) was domesticated in Jalisco, Mexico and in northwestern Argentina. Since several additional crops also seem to have arisen in these two geographical areas, archaeological exploration is clearly called for (Blumler 1992a). Similarly, Heun et al. (1997) identified forms of wild einkorn (T. boeoticum) from Mt. Karacadag in southeastern Turkey as the nearest genetic relatives of einkorn (Triticum monococcum). Several archaeological sites in the vicinity of Karacadag provide evidence that einkorn was a very early domesticate there.

While this approach is promising, several limitations need to be kept in mind: (1) it assumes that the important ancestral genotypes are still extant, (2) it assumes no change in wild genotype distribution in the 10,000 years since domestication, despite considerable evidence for dramatic climatic changes during this time period, and (3) introgression from cultivars can cloud the picture. I focus on the latter problem in this paper. In brief, the difficulty is that presumed 'wild' individuals may be genetically similar to the domesticate because of geneflow from the crop into wild populations, and hence be mistaken for the progenitor. Therefore, it is important to recognize introgression where it occurs, and to determine what the genetic composition of the introgressed individuals would have been before they received genes from the domesticate, prior to carrying out the phylogenetic analysis (Blumler 1994).

Table 1. Most likely number of domestication events for putative primary crops and animals (Blumler 1992a, 1996), illustrating that most species apparently were domesticated only once in any given hearth region.

Domesticate		Most likely number of domestications
Near East
	Emmer	1
	Einkorn	1
	Barley	2
	Pea	1
	Lentil	1
	Chickpea	1
	Broad bean	1
	Bitter vetch	2
	Flax	1
	Sheep	1
Africa
	Finger millet	1
	Pearl millet	1
	Yam bean	2
Andes
	Common bean	>3
	Potato	1
Mexico
	Maize	1
	Common bean	1
	Tepary bean	1
	Scarlet runner	2
	Peppers	1
	Amaranths	1
	Squash (Cucurbita pepo)	2†
Other
	Sunflower	1
	Rice	2
	Sweet potato	1
	Chicken	1
	Cattle	2
	Dog	2

^† Includes possible eastern US domestication event.

Wild emmer wheat

Wild emmer, a tetraploid, is the ancestor of most wheat cultivated today. It is distributed in the Fertile Crescent, from Palestine and Jordan to southeastern Turkey, northern Iraq and western Iran. It is more common in the western portion of this arc, while another, morphologically almost identical, wild tetraploid species (Triticum araraticum) is common in the eastern portion. When Harlan and Zohary (1966) proposed that emmer had been domesticated in or near the Upper Jordan Valley, it was not yet recognized that T. dicoccoides occurs in the eastern and northern portions of the Fertile Crescent. This fact became known in time for Zohary (1969) to modify his position, though in a note added in proof to his paper. Although the paper is widely read, many archaeologists have perhaps (understandably) overlooked the key footnote in which he pointed out that the species may have been domesticated almost anywhere in the Fertile Crescent, and not necessarily in the Upper Jordan Valley. Consequently, there appears to be some confusion about the geneticists' conclusions, with some scholars interested in agricultural origins continuing to believe that the Upper Jordan Valley has been identified as the most likely locus of domestication.

Indeed, many wild individuals from this region do look a lot like domesticated wheat. Especially in that portion of the valley just north of the Sea of Galilee, plants resemble domesticates in being robust, large-seeded, early maturing, and also in such specific traits as lack of anthocyanin pigment. Moreover, electrophoretic studies demonstrate that the wild emmer populations in this sector of the valley are sharply differentiated genetically from all other investigated Palestinian populations, so much so that a case could be made for classifying these populations as a separate species (Nevo et al. 1982; Golenberg 1986); here, it is categorized as the Upper Jordan Valley (UJV) race of wild emmer. This race disappears rapidly as one travels from UJV to the east or west, and is replaced by populations that are similar in their electrophoretic alleles to other Palestinian populations (Golenberg 1989; Blumler 1994).

In UJV, wild emmer is reported to be extremely abundant, especially where grazing is light, suggesting that it is well adapted to truly wild conditions (Zohary 1969; Noy-Meir 1990). Moreover, wild emmer is known to be scarcely if ever weedy (Harlan and Zohary 1966; Blumler and Byrne 1991), and is reported to have only limited geneflow (Golenberg 1987), so that introgression would seem to be less likely than in other crop progenitors that are common in cultivated fields. On the other hand, hybrid swarms (including one of several thousand individuals) were securely identified in UJV, on the edge of formerly cultivated fields (Zohary and Brick 1961). Little wheat is grown in UJV today, but formerly the region was long a center of durum production. For instance, the Biblical parable of the loaves and fishes is situated in UJV. More recently, up until 1948, there were many Palestinian villages in UJV, the denizens of which would have cultivated durum. Most of the Palestinian villages were destroyed during the 1948 war although a Bedouin settlement was left intact (Falah 1996). The continuing existence of this village, with its emphasis on pastoralism, seems to have given some scholars the false impression that traditional land use in UJV was pastoral when in fact farming played a major role. Thus, domesticated individuals should have had opportunities to come in contact with wild plants, especially along rocky, untillable field margins. Scholarly opinion concerning the resulting frequency of hybridization varies greatly, with some believing that there has been essentially no introgression, and others asserting that there was a great deal.

Harlan and Zohary were not the first to propose Palestine as the place where wheat farming began. N.I. Vavilov traveled through the region in 1926; he noticed “a peculiar subspecies of wild wheat [that] accompanies cultivated hard wheat in Palestine” (Vavilov 1957:98), and concluded that it must be the wild progenitor because of its similarity to the domesticate. So distinct from other forms of wild emmer that he classified it as a subspecies, his description fits UJV. Yet he encountered his new subspecies to the south, at the edge of the Esdraelon Plain:

“At the base of the hills, from which flows the underground River Esdraelon, we observed large growths of wild wheat in mixture with wild double-rowed barley. This was an abandoned lot with soft fertile ground positioned right next to the field. The wheat here had a very different appearance from that which we collected in Hauran, Syria. Spikes were large, but with rough spikelets and large seeds. This was no longer an extreme of a xerophyte, but essentially a plant close to cultivated wheat.

“Investigating the fields of the Esdraelon Valley, we found wild wheat in large quantities on edges, along boundaries. No doubt it represents the closest wild source of cultivated wheat, especially hard wheat.” (Vavilov 1962:141)

These observations, taken together, suggest introgression. The Esdraelon is a region that supported extensive durum cultivation in Vavilov's time, but today is given over to other crops. It appears that Esdraelon populations of wild emmer were located at the base of rocky, untillable slopes, at the edges of deep soil and cultivation -precisely where hybridization between a crop and a non-weedy species would be most likely to occur. One expected outcome of introgression would be high variability. Another would be possible reduction or elimination of the stands after cultivation ceased. Israeli geneticists have collected wild emmer intensively throughout their territory, but have never reported populations from the Esdraelon. Other judaicum specimens discussed by Jakubziner also seem to be from areas of intensive cultivation. For instance, he describes judaicum from Mt. Hermon, at Majdal es-Shams, a population center in a wheat-growing region.

Approaches to the identification of introgression

Several lines of investigation are proving fruitful in identifying introgression in wild emmer, and should be useful in other species as well (Box 1). First, while introgressed individuals should resemble domesticates, they should be more similar to modern varieties than to primitive ones. Progenitors, on the other hand, should resemble primitive forms of the domesticate more than later-evolving derivatives. For example, wild emmer gave rise initially to emmer, which is rarely cultivated today, and not at all in Palestine. Emmer later evolved into durum and several other tetraploid wheats, while bread wheat is an allopolyploid derivative. In recent centuries, durum has been the major wheat cultivated in Palestine, though there has been a shift toward more bread wheat since 1948. Introgression from the latter would be limited because of the difference in ploidy level, though it does occur (Dorofeev 1969). Thus, one would predict that UJV plants will resemble durum more than emmer if introgressed, while they will be more similar to emmer if they represent the genotypes originally taken under cultivation.

Second, the ecological conditions to which wild plants must adapt are very different from those which a domesticate faces, and this results in varying degrees of selection for morphological and other characteristics. For instance, indehiscence is strongly selected against in the wild, since the seed needs to disperse from the top of the plant and find its way to the soil in order to germinate. In contrast, indehiscence is usually selected under cultivation (Harlan et al. 1973; Blumler and Byrne 1991). The presence of indehiscent individuals in wild populations, therefore, suggests very recent introgression. Similarly, domesticated cereals characteristically exhibit uniform, high rates of germination, while wild synaptosperms such as wild emmer have pronounced dormancy polymorphisms. A synaptosperm is a species that disperses more than one seed in each diaspore; wild emmer spikelets typically have two seeds. If both seeds germinate at once, their seedlings will compete. Selection, then, favors a somatic polymorphism in which one seed is non-dormant, and the other is dormant for perhaps a year (Zohary 1969; Blumler 1991, 1992a). Selection against non-dormancy should not be as severe as against indehiscence in the wild, so the trait should persist for a longer time period after the introgression has ceased. Other domesticated traits, discussed below, probably are selected against to a still slighter degree, and persist still longer, but on the other hand become less reliable indicators of introgression for precisely the reason that selection does not work strongly against them. By considering a range of such traits, one should be able to test whether introgression has occurred in a given population.

Finally, introgressed plants should exhibit a predictable spatial pattern. They should be more common in and around current or former cultivated fields, especially where cultivation abuts on untillable land that is good habitat for the wild species. Introgression is more likely to be significant also where the amount of cultivation is great in comparison with the size of the wild populations: theoretically, then, geneflow from the crop might overwhelm selection for wild traits. Wild populations can be sampled along transects running away from cultivated land, to determine if the putative introgressed alleles decrease with distance. If so, introgression is likely, especially if there is no corresponding environmental gradient along the transect.

Comparing wild emmer with durum and emmer

Jakubziner's (1932) monograph remains the most authoritative description of the morphological variation in emmer and durum, and is useful also because of its treatment of Palestinian durum landraces, which would be the most likely parents in introgression events. Durum and emmer are different from each other, especially in spikelet characteristics. In addition to the characters described in Jakubziner, I added a few cryptic (i.e. physiological rather than morphological) traits investigated by others. To compare with wild emmer, I examined the herbarium specimens at important repositories such as Edinburgh and the Hebrew University, and also visited a few populations in the field. While typical (horanum) wild emmer is not at all similar to durum, UJV varies in the direction of durum in several respects (Blumler 1994). Grain shape, glume shape, first glume tooth, glume pubescence, spikelet width and glutenin A1-1 allele all are at least occasionally like durum, while there are no characters for which UJV is demonstrably more similar to emmer. In general, UJV is highly variable but intermediate between durum and horanum wild emmer, as one would expect if it were the product of hybridization (Blumler 1994).

Box 1. Approaches to identifying introgression. 1. Morphological comparisons wild progenitor should be more similar to emmer than durum 2. Ecological considerations certain traits are likely to be adaptive in the wild, others in the cultivated field 3. Spatial pattern introgression is more likely in or around former cultivated fields

The glutenin data of Levy and Feldman (1988) are particularly informative. Glutenin A1-1 is present in domesticated emmer, but absent in durum. It is almost always present in wild emmer, but is generally absent in UJV, which suggests that introgression from durum may have been massive. Of the 19 wild populations outside UJV that Levy and Feldman examined, glutenin A1-1 was 100% present in all but two: a population just south of UJV, and Majdal es-Shams. As mentioned above, Jakubziner (1932) reported judaicum from Majdal es-Shams. Both populations are in agricultural areas, whereas most wild emmer populations are on rocky slopes where cultivation could only have been practised in a few select spots. Interestingly, the glutenin A1-2 locus presents a different pattern: absent in most wild emmer populations, absent in durum and emmer, but present in bread wheat and in some UJV plants. Since Levy and Feldman did not study Palestinian durum landraces, it would be interesting to determine if they also possess A1-2 alleles.

Ecological influences on introgression

As discussed above, a few domesticated traits such as indehiscence and lack of dormancy are selected against in the wild. On the basis of ecological research, I believe that several additional traits are selected against, though less strongly than indehiscence (Box 2). I believe, too, that a few domesticated traits are sometimes advantageous in wild environments (Blumler 1994). Most of these traits are spikelet characters, because of the importance of seed dispersal and self-implantation in the ecology of the wild plant. Zohary and Brick (1961) and Zohary (1969) described the 'arrow-shaped' spikelets of wild emmer, and pointed out that they are so structured as to enable them to drill into the soil where they are protected from predation until the fall rains initiate germination. The glumes tightly invest the two grains, which are elongate in order to fit into the arrowhead shape. I examined specimens that differ from typical forms in being more like domesticates. Any change to rounder grains, more grains per diaspore, or more divergent glumes makes the structure less streamlined and therefore, presumably, less able to drill into the soil.

I also investigated the drilling ability of the spikelets under various microenvironmental conditions (Blumler 1991). I found that the spikelets generally drill into soil cracks or adjacent to rocks or other obstructions where the vegetation has been mostly removed (e.g. under grazing); on the other hand, in more-or-less undisturbed conditions where the dead plants remain standing at season's end, the two awns of the spikelet tend to catch in the litter, and prevent the diaspore from penetrating all the way to the soil surface. Seedling establishment is nonetheless excellent under these circumstances, as the grains end up close enough to the surface that their roots can penetrate the soil after germination. Thus, it seems possible that the changes in spikelet morphology induced by introgression would be selected against where there is grazing, but might not be under undisturbed conditions. The somewhat more awkward spikelets of hybrids, with round grains or three grains, probably still can drop through litter to just above the soil surface, whereas they may be less efficient at drilling into cracks on bare ground. Further research is needed to test this hypothesis.

There are predictable effects of domestication on grain size and shape. As the spikelet is no longer needed to disperse the grains, the tightly investing glumes open up over the course of evolution, allowing the florets to expand outwards or to increase in number. Since grain size and shape reflect that of the floret (Millet 1986), the 'naked' grains such as durum are often round, whereas the glume wheats such as emmer remain elongate. Those species or varieties that increase the number of seeds per spikelet, such as bread wheat, produce relatively small grains, whereas those with fewer grains per spikelet produce large ones, such as some forms of durum. Thus, with introgression from durum, wild grain size is likely to increase.

But whether this will be selected against should depend on site conditions. In annual species, seed size is strongly correlated with site productivity, both among and within species (Blumler 1992b). While the within-species variation may be largely phenotypic, the variation between species is so great that it must be primarily genetic. On infertile sites, annuals have difficulty producing enough photosynthate to fill large seeds, whereas they should be able to produce small seeds relatively easily. Sinnott (1921) showed in cultivated annuals that seed size increases phenotypically as productivity (plant size) increases, eventually leveling off. I studied several Jerusalem wild emmer populations on hard limestone, where soil depth, and hence productivity, vary over short distances, providing a natural analogue to Sinnott's experiment (Blumler 1992b). Results were identical to his: grain size increased with plant height up to a point, and then appeared to level off (Fig. 1). However, some populations leveled off at a low mean grain weight (<20 mg), whereas others did not do so until they were considerably larger (>35 mg). The populations with low mean grain weight occurred on sites that were relatively low in productivity over most of the site, though with a few high-productivity pockets, whereas the larger-grained populations were on deeper, more productive soil with a few small areas of infertile substrate. When seeds from wild populations were planted in a uniform garden, however, there was no relationship, within populations, between weight of grain planted and grain produced. That is, the within-population variation was mostly phenotypic. The relevant implication is that the large grain size of most domesticated genotypes should be advantageous on fertile sites but not where productivity is low. Since most cultivated fields have deep soil, they are productive compared with natural stands of wild emmer. Thus, in the places where introgression is likely to occur, large grain size may not be disadvantageous, and may even be beneficial.

Box 2. Ecological aspects of introgression. 1. Dormancy selection for polymorphism within spikelet 2. Drilling ability of spikelet selection for elongate grains, parallel glumes, narrow spikelets, two seeds/spikelet, dehiscence 3. Productivity selective and phenotypic effects on grain size 4. Pigment selection for coleoptile pigment, glume pigment

Finally, it appears that there may be a selective advantage to anthocyanin and other pigments in wild plants, but not in domesticates. The ecology of anthocyanin is poorly understood and even less studied - one suggestion is that it makes plants less palatable to herbivores - but there is no question that loss of pigment is a feature not only of wheat domestication, but also of the domestication and evolution of many other crops. Exceptions occur when humans intentionally select pigmented forms for ornamental purposes (e.g. 'Sangre de Cristo' maize, amaranths). Most wild emmer populations are black glumed, with anthocyanin pigment present in various vegetative parts, often including the coleoptile. Most durum lacks pigment.

With these ecological considerations in mind, and also the morphological comparisons discussed in the previous section, we can construct a list of traits that are likely to be characteristic of domesticates but not of truly wild plants (Box 3). Included will be those characters controlled by genes linked to the traits of ecological concern. For instance, the rounding of the glumes that often occurs in durum because of the change from elongate to rounded grains typically is associated with specific changes in the glume teeth that seem likely to be of no selective value. Glume pubescence also is characteristic of durum, particularly the Palestinian landraces (Jakubziner 1932; Poiarkova and Blum 1983), and especially the varieties with rounded glumes, but it is difficult to understand how it could be adaptive. Wild emmer populations are mostly glabrous, but the sibling species T. araraticum is usually pubescent. It seems likely, therefore, that glume pubescence is a neutral or nearly neutral character. I noticed, during the course of my examination of herbarium specimens, that wild emmer with pubescent glumes occurs, without exception, in populations that show more reliable indicators of introgression such as rounded glumes or three-seeded spikelets. Consistent with this conclusion, Oppenheimer (1963) reported that pubescent glumed individuals of wild emmer are less dormant than those with glabrous glumes. Perhaps there may be pubescent individuals somewhere that are not the result of introgression but they have not been collected.

There are numerous additional traits, not listed in Box 3 because of insufficient data, that merit further investigation. For instance, glaucous glumes and absence of leaf sheath cilia both appear to be durum traits (Jakubziner 1932), but I have little information on their distribution in the wild. More speculatively, it seems possible that pale glumes, which are common in durum but infrequent though widespread in wild emmer, may be a reliable indicator of introgression. In my Jerusalem studies, for instance, I noticed that pale-glumed forms of wild emmer make larger grains and grow in more productive sites than populations that produce black glumes. I have yet to encounter a pale-glumed form of wild emmer that produces the 20-mg grains of some black-glumed types, except under stress.

Regardless, populations from UJV that I examined either in the field or the herbarium all exhibit more than one of the indicators listed near the top of Box 3. Possibly typical is the population at Migdal described by Jakubziner (1932), which included the varieties (with some peculiar variants) listed in Table 2. All of the peculiar variants are typical of durum. Of course, in this case there are no data on characters that he did not investigate, such as glutenin alleles.

Spatial analysis of UJV

As mentioned above, all UJV-type populations occur in close proximity to extensive areas of former durum cultivation. More precise spatial analysis of allele frequencies should allow a more rigorous test of the introgession hypothesis. Fortunately, this is possible because of the existence of a detailed data set from Ammiad, a kibbutz located at the former site of the Palestinian waystation of Jubb Yusuf. A multidisciplinary team has carried out detailed ecological and genetic studies of the population, located on karstic limestone hills that form the western boundary of UJV, as well as the adjacent UJV alluvium (Anikster and Noy-Meir 1991). The site is located only about 100 meters from the large hybrid swarm described by Zohary and Brick (1961) and D. Zohary (pers. comm.), but the Ammiad team (with the exception of Felsenburg et al. 1988) assumed that no plants in the population are introgressed. Plants were gathered along transects, and analyzed genetically, while ecologists characterized the four primary microenvironments occurring there. If UJV plants are truly a locally evolved race, adapted to the basalt and alluvium of UJV itself, and not to other environments, then UJV alleles should occur at Ammiad where transects extend into the alluvium, but only incidentally on the adjacent slopes, in microsites that mimic UJV. On the other hand, if UJV is the result of introgression, UJV alleles should decline with distance from the alluvium regardless of microenvironment. The full analysis will be reported in a separate paper. Here, I will discuss only the most dramatic evidence.

This evidence comes from Transect 'C', which is in extreme karstic terrain. Noy-Meir et al. (1991) concluded that the microenvironment along Transect C differs greatly from the microenvironment on the UJV alluvium. In fact, according to Noy-Meir et al. these are the two most differentiated environments at the site. However, the base of Transect C is geographically close to the alluvium of a wadi that runs into UJV, and cultivation would have been practised in the wadi in the past (D. Zohary, pers. comm.). Thus, on ecological grounds one would predict C to have few if any UJV-type plants, while it might have many such plants if the introgression hypothesis is correct.

In fact, and in contrast to the remainder of the population, the section of C nearest to the alluvium (accessions 160-180) is laden with plants that have many UJV alleles (Table 3). As one goes up the transect away from the wadi, the UJV alleles suddenly drop out at accession 160, which is half way up a limestone cliff. The accessions on the transect above the cliff are almost completely devoid of UJV alleles, suggesting that the cliff forms a topographic barrier to their dispersal upslope from the wadi. Gene frequencies in lower C are very similar to the two UJV populations that have been intensively studied, Yehudiyya and Tabgha (Table 3). Lower C plants also resemble UJV morphologically. In contrast, the remainder of the Ammiad population exhibits gene frequencies that are very similar to non-UJV populations. Given that C is ecologically so dramatically different from UJV, it is difficult to explain this pattern as the result of natural selection. Moreover, if selection favors UJV alleles on this particular karstic site, then UJV plants should turn up on extreme karstic limestone elsewhere as well. Karstic limestone is extremely widespread in Palestine, and commonly supports wild emmer populations, but as far as is known, does not support UJV plants elsewhere.

Fig. 1. Relationship between plant height and grain weight in a population of wild emmer in Jerusalem (Blumler 1992b). Seed weight appears to increase linearly up to a plant height of approximately 75 cm and a weight of 21 mg; it appears to be stable at greater plant heights.

Box 3. Some morphological and other indicators of introgression in wild emmer. Traits are ranked according to the reliability with which they suggest that introgression has occurred. Indehiscent plants are almost certainly introgressed, whereas plants that produce large grains may not be.
Indehiscence	Poorly developed second glume tooth
Rounded grains	Absence of glutenin A1-1
Three-grained spikelets	Absence of pigment
Low grain dormancy	Short grain brush
Pubescent glumes	Large grains
Curved glume tooth

Table 2. Some varieties of wild emmer in a collection from Migdal in UJV (Jakubziner 1932).

Variety		Glumes	Peculiar variants
horanum
	kotchii	white, glabrous	glaucous glume
	palestinicum	white, pubescent
judaicum
	arabicum	white, glabrous	glaucous glume with long curved tooth; glabrous leaf sheath
	vavilovi	black, glabrous	long curved apical tooth
	fulvovillosum	white, pubescent	large tooth

Although the C plants are very similar to those from UJV populations, they are not identical; for instance, compare allele frequencies at loci Pgi-A, Pgi-B and Glu-A1-1 in Table 3. Nor are Yehudiyya and Tabgha identical to each other. At most loci there is a characteristic allele that predominates throughout Palestine, and also at Ammiad, but not in UJV. Within UJV, on the other hand, the predominant allele at any given locus is highly variable from site to site, and alleles sometimes are common at one site but absent at others. This is as one would expect if UJV resulted from repeated introgression events with diverse durum landraces, but makes little sense if UJV is the result of natural selection in a purely wild setting. UJV appears ecologically fairly uniform, in contrast to karstic limestone sites where soil depth, moisture and shading typically vary greatly over short distances.

Discussion

While these results are still preliminary, the convergence of several lines of evidence strongly indicates that massive introgression has occurred in UJV, giving rise to the UJV 'race' of wild emmer. Does this mean that UJV is not the locus of domestication? At present, we have no information that bears on that question. The evidence for massive introgression at UJV merely means that its populations are no more or less likely than other wild emmer populations to be the progenitor. UJV-type plants reported elsewhere, such as on Mt. Hermon, also seem to be introgressed. If pale glumes occur in the wild only after introgression, then many additional populations and subpopulations have been altered by interactions with domesticated wheat. Introgression appears to have been more massive in Palestine than elsewhere in wild emmer's range. This may reflect the abundance of low-elevation, warm-winter, highly productive habitats in Palestine, though it may also be an artifact of the greater density of exploration and scientific study in Palestine compared with other parts of the Fertile Crescent.

Stands of wild cereals in and around UJV are known to be especially productive (Harlan and Zohary 1966), so large grain size might be present naturally in wild populations, or alternatively, might be selected in the wild after introgression from domesticates. On the other hand, most wild emmer populations occur on much less productive sites, where introgression of genes for large seed size might be selected against. This is a possible explanation for the apparently massive introgression in UJV, and the smaller degree of introgression elsewhere. Moreover, the elimination of Palestinians from UJV in 1948 was followed by a long period of limited exploitation of the land. Settlers initially grazed the region very lightly because they were unsure how much pressure the vegetation could support. Characters such as large seed size and low dormancy should be favored under relatively undisturbed conditions when stands are dense. Recently, grazing pressure has been increasing, and wild emmer populations appear to be declining in consequence. Under these circumstances, selection against domesticated traits may be more severe.

That introgression apparently has been so massive in wild emmer, a non-weedy species, raises questions concerning the extent of introgression in other wild crop relatives. Considerable concern and debate exists about the likelihood that genetically engineered crops might pass the engineered genes to wild relatives, with subsequent effects on ecosystems. Ecologists tend to express serious concern (Tiedje et al. 1989), whereas geneticists sometimes dismiss the possibility, perhaps because of an underlying assumption that wild plants are already well-adapted to natural environments and therefore unlikely to be improved in fitness by geneflow from domesticates. But the example of wild emmer suggests that genes from cultivated plants can indeed affect the genetic make-up and perhaps the fitness of wild plants.

Geneticists have studied wild emmer to address theoretical issues, such as the comparative roles of natural selection and neutral mutation in evolution. Nevo et al. (1982, 1986) concluded that natural selection was at least partly responsible for allozyme distribution patterns, but Nevo's statistical methodology is known to be invalid (Heywood and Levin 1985). In a more rigorous series of experiments, Golenberg (1986, 1989) found no evidence for an adaptive basis for allozyme distribution in UJV. Neither Nevo nor Golenberg considered the possibility that durum genes might have introgressed into their populations. The evidence presented here for massive introgression suggests that the odd electrophoretic variants in UJV are contaminants from durum, currently hitchhiking on other traits upon which selection is working. Similarly, the Ammiad study site was selected in part because of its high genetic variability; but when the putative introgressed alleles are subtracted, genetic diversity becomes very low (Blumler, unpublished). Consequently, the assumption of the researchers that there has been local-scale adaptation to differing microenvironments at Ammiad (Nevo et al. 1986) is probably incorrect.

Conclusion

To return to the agricultural origin question, one implication of these results is that introgression may be more massive than realized in other wild progenitors, such as einkorn and bean, as well. Einkorn probably has not been cultivated near Karacadag for some time, and would have been rare for a still longer period, so any introgression that might have occurred presumably would have been much more ancient than the introgression in UJV. Assume for the sake of argument that massive introgression did occur, say, 3000 years ago. By now, the most obvious indicators of introgression, such as indehiscence, would have been selected out of the wild populations. Only more-or-less neutral introgressed alleles might still be present. But since the analysis of Heun et al. (1997) was of DNA that probably is more-or-less neutral, it does not seem possible to reject introgression entirely. I should emphasize that I am quite comfortable with the conclusions of Heun et al. (1997), and also with those of Gepts (this volume), because they conform to my own views on agricultural origins; for instance, Gepts' report of a northwest Argentinian origin for Andean common bean was anticipated by my paper (Blumler 1992a). But at the same time, given the amount of introgression I am finding in wild emmer, I cannot but feel that the introgression issue needs to be addressed more fully in all other crops as well.

Table 3. Allele frequencies in Triticum dicoccoides in Palestine^†.

Locus	Ammiad		Transect C 160-180		Yehudiyya (UJV)		Tabigha (UJV)‡		Other Palestine		Ammiad, exc. 160-180
	N	Freq.	N	Freq.	N	Freq.	N	Freq.	N	Freq.	N	Freq.
6Pgd-2	191		21		96		39		354		170
M		0.890		0.524		-		0.333		0.958		0.935
S		0.110		0.476		1.000		0.667		0.006		0.065
Pept-3	146		8		96		0		49		138
M		0.918		0.500		0.016		-		1.000		0.942
S		0.082		0.500		0.984		-		-		0.058
Pgi-A	230		20		39		37		375