Spontaneous wheat-Aegilops biuncialis, Ae. geniculata and Ae. triuncialis amphiploid production, a potential way of gene transference

Some F1 hybrid plants between three species of theAegilops genus and different hexaploid wheat Triticum aestivum cultivars show certain self-fertility, with averages of F1 hybrids bearing F2 seeds of 8.17%, 5.12% and 48.14% for Aegilops biuncialis, Aegilops geniculata and Aegilops triuncialis respectively. In the Ae. triuncialis-wheat combination with ‘Astral’ wheat cultivar, the fertility was higher than that found in the other combinations. All the F2 seeds studied were spontaneous amphiploids (2n=10x=70). The present study evidences the possibility of spontaneous formation of amphiploids between these three Aegilops species and hexaploid wheat and discusses their relevance for gene transference. Future risk assessment of transgenic wheat cultivars needs to evaluate the importance of amphiploids as a bridge for transgene introgression and for gene escape to the wild. Additional key words: Aegilops, hybrid self-fertility, risk assessment, Triticum aestivum, wild relatives.

intergeneric and interspecific crosses. But the picture is quite different and numerous crops are known to have wild relatives that can hybridize with them somewhere in the world. Gene flow between cultivated species and their weedy and wild relatives has been documented in species such as oilseed rape (Brassica napus L.) (Jørgensen and Andersen, 1994), maize (Zea mays L.) (Doebley, 1990), sorghum (Sorghum halepense (L.) Pers) (Arriola and Ellstrand, 1996), sunflower (Helianthus annuus L.) (Arias and Rieseberg, 1994) and sugarbeet (Beta vulgaris L.) (Bartsch and Pohl-Orf, 1996).
Transgenic wheat varieties are being developed and field-tested and probably in the next few years certified cultivars will be commercially available. Potential risks should be examined before their widespread adoption. Gene transfer between cultivated wheat Triticum aestivum L. and the weedy Aegilops cylindrica Host. is known to occur and scientific literature confirms that herbicide resistance genes can move from herbicide tolerant wheat to Ae. cylindrica through hybrids (Seefeldt et al., 1998;Zemetra et al., 1998;Guadagnuolo et al., 2001;Wang et al., 2001). Other wild Aegilops species like Ae. geniculata Roth., Ae. biuncialis Vis. and Ae. triuncialis L. also form natural intergeneric hybrids with bread wheat where they grow in sympatry (van Slageren, 1994;Loureiro et al., 2006;Zaharieva and Monneveux, 2006), a phenomenon underlining the close genetic links of the two genera. Hybrids between Ae. geniculata and Ae. triuncialis and wheat have been found in several countries of Europe, mainly in Spain and France, while Ae. biuncialis-wheat natural hybrids have been described in Lebanon (van Slageren, 1994). These natural hybrids are highly sterile, although seeds may occasionally be found in Ae. geniculata hybrids (van Slageren, 1994;Loureiro et al., 2008).
Crop-to-wild transgene transference may occur through recombination in homoeologous chromosome pairs, translocation, or chromosome retention (Schoenenberger et al., 2006). Another possible transference mechanism is the formation of amphiploids. Spontaneous chromosome doubling usually results from unreduced female and male gametogenesis and the union of those unreduced gametes leads to the formation of a spontaneous amphiploid from an interspecific hybrid (Ramsey and Schemske, 2002). It is known that amphiploidy has played a fundamental role in the evolution in the Triticeae tribe: durum wheat Triticum turgidum L. probably evolved in one step as an amphiploid between the wild grasses Ae. speltoides Tausch and T. urartu Tumanian ex Gandilyan; that durum wheat crossed naturally with Aegilops tauschii Coos, which resulted in the first hexaploid wheat, T. spelta L. (Stebbins, 1946;Maan and Sasakuma, 1977;Jauhar, 2003). In the amphiploids every chromosome of the original hybrids is represented twice and each chromosome has a homologous partner to pair at meiosis. Thus, meiosis is regular and fertile derivatives can be obtained from hybrids whose sterility was caused by chromosomal unbalance and irregularity (Bretagnolle and Thompson, 1995;Ramsey and Schemske, 1998;David et al., 2004). In their review on hybridization between wheat and its relatives, Zaharieva and Monneveux (2006) mentioned that Tschermak and Bleier (1926) were the first to obtain an amphiploid species as a result of the spontaneous doubling of the chromosomes of a wheat hybrid produced by the cross of T. turgidum dicoccoides with Ae. geniculata. Amphiploidy is a mechanism that can also provide a valuable genetic resource for the introgression of desirable genes from alien species to cultivated wheats. Induced amphiploidy was indeed used to transfer Ae. geniculata chromosomes carrying disease resistance into bread wheat Ganeva, 1998, 1999).
In nature, amphiploids could serve as an effective bridge for gene flow over the interspecific and intergeneric barriers. David et al. (2004) reported an estimated frequency of 10 -6 of spontaneous amphiploidy between the tetraploid wheat T. turgidum and Ae. geniculata in field sympatric populations, that was higher (10 -3 ) in nursery conditions and with different genotypes. Genomic in situ hybridization proved that fertile amphiploids had arisen through unreduced gametes and that some of them carried wheat−Ae. geniculata recombinant chromosomes.
The objective of this paper is to report the production and frequency of spontaneous amphiploids in the selfprogenies of hybrids between hexaploid wheat T. aestivum and the Aegilops species Ae. biuncialis, Ae. geniculata and Ae. triuncialis.

Material and methods
Experiments were carried out at the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) experimental station in Madrid, Spain (40º 27´North; 3º 44´West).
Ae. biuncialis, Ae. geniculata and Ae. triuncialis are relatively easy to hybridize with wheat. Aegilops × T. aes-tivum hand crosses made during several years (from 2004 to 2007) yielded abundant F 1 hybrid seeds, at rates from 30 to 70% of the pollinated flowers (Loureiro, 2005). The plant material employed in the experiments was the F 2 seed progeny produced by selfing of the F 1 hybrids between the wheat wild relatives Ae. geniculata (2n=4x=28 chromosomes, genomic constitution MMUU), Ae. biuncialis (2n=4x=28 chromosomes, UUMM) and Ae triuncialis (UUCC), and the T. aestivum (2n=6x=42, AABBDD) cultivars 'Chinese Spring' (CS), 'Castan' and 'Astral'. Nomenclature of the wild Aegilops spp. is according to van Slageren (1994) and the genomic constitution is according to Kimber and Tsunewaki (1988). The F 1 hybrids were sown each year in autumn in a greenhouse in which lateral walls were open and temperature and relative humidity were similar to outdoor conditions. There was no temperature or humidity control. Two to three days prior to anthesis, hybrid spikes from each plant were covered with cellophane bags to avoid any cross-pollination. The fertility of the hybrids was estimated as the number of seeds obtained per 100 spikelets after selfing. A Kruskal-Wallis test was carried out to determine significant differences between the F 2 seed set in the different Aegilopswheat combinations, followed by a Mann-Whitney U-test for multiple comparisons. A confidence level of 95% (p < 0.05) was considered significant.
Seeds collected on hybrids were placed on filter paper moistened with distilled water in 9-cm diameter Petri dishes and once germinated were subjected to cytological analysis to confirm amphiploidy. Root meristems for mitotic chromosome number counts were collected from each germinated seed and were pre-treated in α-bromonaphthalene at 4ºC during 16 hours, fixed in a 90% acetic acid solution during 30 min, washed twice with 95% ethanol and stored in 70% ethanol. After a minimum of 10-14 days, root meristems were ready to be stained in Schiff reactive for 60 min after a 10-12 min hydrolysis at 60ºC HCl 1N and squashed in a 1% Belling's aceto-carmin solution prior to light microscopy observation.
The F 2 plantlets were further grown to maturity in the greenhouse.

Results
A total of 182, 150 and 26 hybrids between hexaploid wheat and Ae. biuncialis, Ae. geniculata and Ae. triuncialis respectively, were studied. These F 1 firstgeneration hybrids were pentaploids (2n=5x=35) (Figure 1A). F 1 plants were grown to maturity in order to study the self-fertility. Some of the plants gave F 2 seeds, with a percentage of F 1 plants bearing at least one F 2 seed that varied significantly in the distinct Aegilops-wheat combinations (Kruskal-Wallis test: H=24.06, p<0.001) ( Table 1). The highest self-fertility was obtained for the Ae. triuncialis-wheat combination with frequencies of 1.82 seeds in 100 spikelet and percent averages of 48.14% F 1 plants with F 2 seeds. All the hybrids obtained between Ae. triuncialis and the wheat cultivar 'Astral' were fertile (Table 1), their fertility varied from 1.06 to 8.88 seeds in 100 spikelets among the five plants studied. The fertility of the hybrids with 'Astral' was significantly higher than the obtained with the other wheat cultivars (Mann-Whitney U tests, p < 0.05) that did not show differences among them (Mann-Whitney U tests, p > 0.05). Fertility was lower for Ae. biuncialis and Ae. geniculata-

A B
wheat combinations with values of 0.09 (0.07-0.85) and 0.08 (0.05-0.16) seeds in 100 spikelets respectively, and without significant differences among these two combinations according to the Mann-Whitney U test (p = 0.097). Not all the F 2 hybrid seeds obtained were germinated and used for amphiploidy study; some of them were conserved for subsequent studies (Table 1). F 2 seeds were mostly well formed with a germination rate of 89%, and all the germinated Aegilops × wheat F 2 hybrid seeds showed 70 chromosomes ( Figure 1B). On the assumption that all the obtained F 2 seeds are amphiploids, the mean frequency of amphiploid formation (new decaploid F 2 plants per single pentaploid F 1 hybrid) would be of 0.049 for Ae. biuncialis, 0.047 for Ae. geniculata and 1.34 for Ae. triuncialis-wheat hybrid combinations.
The decaploid F 2 plants were grown in the greenhouse. The two F 2 Ae. biuncialis-wheat plants were not viable and died at the seedling stage. In the case of the Ae. geniculata and Ae. triuncialis-wheat hybrid plants, the phenotype of the F 2 amphiploids resembled that of the normal Aegilops × T. aestivum F 1 hybrids with regard to their leaf size, height and spike morphology (Figure 2). The plants showed a vigorous vegetative growth.

Discussion
The hybrids between tetraploid Aegilops-species and hexaploid wheat are often considered self-sterile; however, seeds can occasionally be found (Maan and Sasakuma, 1977;van Slageren, 1994). This high sterility may be due to the expected highly disrupted meiosis in these F 1 hybrids, whose parental plants differ in chromosome number and in chromosome pairing homology due to the relative divergences between Aegilops genomes (U, C and M in the case of the species studied) and wheat genomes (A, B and D) (Sears, 1941). Structural differences in chromosomes between parents and cytoplasmic differences may also cause sterility in certain hybrids (Stebbins, 1950;Maan and Sasakuma, 1977).
Our data reveals certain degree of self-fertility in some Aegilops-wheat F 1 hybrid combinations. The most plausible explanation would be the fecundation of an unreduced female gamete (2n=5x=35) of an F 1 hybrid with another unreduced male gamete (2n=5x=35), resulting in a decaploid plant (2n=10x=70). As happens with the self-fertility of intergeneric hybrids, the formation of unreduced gametes was also considered to be infrequent in general on the assumption that the occurrence of 2n gametes  Table 1. Fertility of the three Aegilops × wheat F 1 hybrids by selfing and evidence of formation of amphiploids. The fertility of the hybrids was estimated as the number of seeds obtained per 100 spikelets after selfing in plants is rare and sporadic. But contrary to this assumption, Harlan and de Wet (1975) showed that almost all plant species can produce 2n gametes in some frequencies and that all polyploids have originated through functional 2n gametes. Many of the intergeneric hybrids have disturbed chromosome pairing and produce high frequencies of 2n male and female gametes. Both phenomena have been reported to occur in the F 1 hybrids between T. turgidum and Ae. tauschii (Xu and Joppa, 1995). There is clear evidence that viable seeds are produced by functional 2n gametes in durum wheat haploids (Jauhar et al., 2000;Jauhar, 2003). Similarly, David et al. (2004) observed that fertile Ae. geniculata × T. turgidum durum hybrids always produced total or partial amphiploids in their offspring.
In the case of our Ae. triuncialis-wheat F 1 hybrids from 'Astral' wheat, the F 2 seed set was quite elevated and all plants obtained were fertile. The number of spikes and spikelets per plant was greater in these hybrids, which could contribute to their higher fertility. This wheat genotype effect is also a factor that has an influence on the frequency of 2n gamete formation (Maan and Sasakuma, 1977;Bretagnolle and Thompson, 1995;Ramsey and Schemske, 2002). However, the Ae. triuncialis-wheat F 1 hybrid plants flowered later due to the Aegilops parent's life cycle and suffered higher temperatures during their meiosis, so it cannot be excluded that this increase in their fertility may be the result of the effects of those temperatures which can also induce meiotic abnormalities promoting unreduced gamete production (Sax, 1937;Ramsey and Schemske, 1998).
Hybridization rates between wheat and Ae. biuncialis and Ae. geniculata under semi-natural conditions are around 0.3% (Loureiro et al., 2007) while the frequencies of amphiploid formation obtained in this study vary from 3.7 to 28.5 per 100 F 1 hybrids studied between Ae. biuncialis and wheat and from 2.5 to 10 amphiploids for Ae. geniculata-wheat hybrids, depending on the Aegilops-wheat parental combination. Thus, finding spontaneous amphiploids in nature would not be completely exceptional. In the case of Ae. triuncialis, although the frequency of hybridization with wheat under natural conditions is unknown, it is a matter of concern that between 11 and 620 amphiploids may be formed per 100 F 1 hybrids. As stated above, fertility rates were significantly higher with 'Astral' and these data on amphiploid formation cannot be generalized for all Ae. triuncialis-wheat hybrid combinations.
However, the production of unreduced gametes and the formation of a new polyploid is only one step towards the establishment of this polyploid, which must be competitive in order to persist. Even if the fertility and/or viability of F 1 s were low, these traits often increase in each successive hybrid generation (Rieseberg, 1997). Therefore if the F 1 sets results in viable seed at all, this will provide a second generation of hybrids that might be more fertile. These will breed, among themselves and with their parents, and might provide a third hybrid generation comprising more and fitter individuals than those of the second one. Once gene dispersal has occurred, it is important to understand whether these genes will persist and establish in the natural populations of free-living relatives; since

A B
there is a lack of information it follows that more studies are necessary. Herbicide-resistant wheat, primarily glyphosate tolerant wheat (Blackshaw and Harker, 2002;Zhou et al., 2003), has the potential to improve the efficiency of weed management. The risk of gene transfer to wild and weedy relatives that grow in sympatry and with overlapping flowering times should be studied, in order to evaluate the likelihood that such an event occurs. Ae. geniculata, Ae. biuncialis and Ae. triuncialis are colonizing species with the capacity to develop large stands, up to many hectares, that could increase their invasiveness under the selection pressure of the herbicide if they acquire the herbicide resistance gene. The risk of introgression of the wheat transgenes into the genome of these Aegilops species is related to the occurrence of meiotic recombination during chromosome pairing in the meiosis of the F 1 hybrids. The introgression may be achieved either by hybridization with transgenic wheat and recurrent backcrossing with the Aegilops parent or by the "bridge" of the amphiploids. This second route of spontaneous amphiploid production, via doubling the chromosomes can permit a good bivalent pairing and overcome the sterility of the F 1 hybrids. In the case of Ae. geniculata-wheat, the F 2 fertility varied between 0 and 36 seeds in 100 spikelets among combinations, with one Ae. geniculata × 'Castan' F 2 plant that reached up to 66.7 seeds in 100 spikelets (Loureiro et al., 2008), while the F 2 amphiploids between Ae. triuncialis and wheat are also fertile and, indeed, plants producing up to 11 F 3 grains can be found (Loureiro, unpublished). In both cases F 3 seeds were well-developed.
Further studies of the recombination in the meiosis of the amphiploid hybrids are essential in order to determine whether a transgene may be transferred to the wild genome. In addition, the possible fertility of these amphiploids increases the likelihood of them becoming a new species in which a transgene would be maintained easily. Data on amphiploid frequency and fertility will without doubt be useful in assessing the potential risks of future transgenic wheat cultivars.