Twenty-nine accessions were used in this study (Table 1), which included: nine accessions of T. petropavlovskyi; six accessions of the other three unique Chinese endemic wheat landraces; one accession of T. carthlicum, T. dicoccoides, and T. turanicum; two accessions of T. durum, T. turgidum, T. polonicum, and T. compactum; T. aestivum cv. Norin-10; and the synthetic hexaploid wheat (SHW-DPW). The artificial synthetic amphiploid between the dwarfing Polish wheat T. polonicum from Xinjiang and Aegilops tauschii (AS60) was produced and named SHW-DPW by Kang et al. (2008), to simulate the hypothesis no. 2 of origin of T. petropavlovskyi. The tetraploid T. polonicum cv. dwarfing Polish wheat was collected from Tulufan, Xinjiang, China. It is the only dwarf mutant of T. polonicum in China. Aegilops tauschii (AS60) originated in the Middle East. Voucher specimens were deposited in the Triticeae Research Institute herbarium, at Sichuan Agricultural University (SAUTI).

Crosses were made in the field at the Triticeae Research Institute, SAUTI. T. petropavlovskyi (AS360) and SHW-DPW were used as maternal plants to cross with tetraploid and hexaploid wheat plants. Florets were emasculated and covered with a cellulose bag. Hand-emasculated spikes were pollinated two days later, and maternal stigmas were brushed with freshly broken anthers from the paternal species. Hybrid seeds were counted, germinated on filter paper in petri-dishes, and then transplanted in pots at the two-leaf stage.

For cytological procedures, spikes were fixed in Carnoy’s II solution (absolute ethanol: chloroform: glacial acetic acid, 6:3:1, v/v) for 24 h, transferred to 70% ethanol and stored in a refrigerator. Pollen mother cells (PMCs) at metaphase I (MI) were squashed and stained with 1.5% carbolic acid-fuchsin solution. Sixty cells at MI were observed from each hybrid, and the calculation of mean pairing frequency (c-value: the mean frequency with which two related chromosome arms pair) was made according to Alonso & Kimber (1981). Micrographs were taken from permanent meiosis preparations using the Olympus BX-51 camera system.

The percentage of seed set and fertility of F₁ hybrids were converted to angle by arcsine transformation, and the transformed data was then subjected to analysis of variance using the DPS (Data Processing System) 3.01 computer package (http://www.statforum.com/). Seed set, fertility of F₁ hybrids and c-value means were compared using the Duncan’s multiple range test (Seraj et al., 1997; Pitkanen, 2000). Differences in seed set and fertility were analyzed at 1% probability threshold.

T. petropavlovskyi was used as maternal parent and crossed with Triticum species and the synthetic hexaploid wheat (SHW-DPW). The results are shown in Table 2. All crosses produced seeds and resulted in mature hybrid plants. The seed sets for two combinations of T. petropavlovskyi × T. polonicum were 34.0% and 43.3%. Statistical analysis indicated that seed sets of crosses between T. petropavlovskyi and two T. polonicum accessions were the highest among the crosses between T. petropavlovskyi and tetraploid wheat (p<0.01). Among the crosses between T. petropavlovskyi and hexaploid wheat, the seed set of T. petropavlovskyi × T. aestivum cv. Chinese Spring was the highest (p<0.01).

Using SHW-DPW as maternal parent, nine crosses with tetraploid and twelve crosses with hexaploid wheat were made (Table 2), all of which produced seeds. In hybrids between SHW-DPW and tetraploid wheat, the seed set of SHW-DPW × T. polonicum was 11.3%, which was at the ordinary level comparing with the other crosses between SHW-DPW and tetraploid wheat. Statistical analysis suggested that the seed set of SHW-DPW × T. carthlicum was significantly higher than that of the other crosses (p<0.01). Considering hybrids between SHW-DPW and hexaploid wheat, the seed set of SHW-DPW × T. aestivum cv. Changning white wheat was 62.5%, the highest among such accessions (p<0.01).

Among the orthogonal and reciprocal crosses between SHW-DPW and T. petropavlovskyi, the seed set of T. petropavlovskyi × SHW-DPW was 16.7%, the lowest one among crosses between T. petropavlovskyi and hexaploid wheat. The seed set of SHW-DPW × T. petropavlovskyi was non-significant in relation to the crosses between SHW-DPW and hexaploid wheat not obvious (Table 2).

Fertility of all F₁ hybrids is shown in Table 2. The F₁ hybrid plants between T. petropavlovskyi and tetraploid and hexaploid wheat grew well. However, the hybrids T. petropavlovskyi × T. turanicum (AS2279), T. petropavlovskyi × T. durum (AS2349), and T. petropavlovskyi × T. aestivum cv. Norin-10 failed to produce seeds. Fertility of hybrids T. petropavlovskyi × T. turgidum (AS2277) and T. petropavlovskyi × T. aestivum ssp. yunnanense was significantly higher than that of the other hybrids between T. petropavlovskyi and tetraploid and hexaploid wheat (p<0.01).

The F₁ hybrids between SHW-DPW and tetraploid and hexaploid wheat also grew well (Table 2). However, five hybrid plants failed to produce seeds: SHW-DPW × T. durum cv. Langdon, SHW-DPW × T. dicoccoides (AS847), SHW-DPW × T. turgidum (AS2277), SHW-DPW × T. turanicum (AS2279) and SHW-DPW × T. aestivum cv. Yinong white wheat. Fertility of hybrids from the SHW-DPW×T. dicoccoides (AS838) was 79.8%, the highest one among hybrids between SHW-DPW and tetraploid wheat (p<0.01). Statistical analysis indicated that the fertility of SHW-DPW × T. aestivum ssp. tibetanum was the highest among all hybrids between SHW-DPW and hexaploid wheat (p<0.01).

In the seven hybrids (2n = 5x = 35) between T. petropavlovskyi and tetraploid wheat, the meiotic configuration patterns in T. petropavlovskyi × T. dicoccoides and T. petropavlovskyi × T. durum cv. Langdon were similar, with a low frequency of trivalents (Table 3; Fig. 1A, 1B). Chromosome pairing at MI in T. petropavlovskyi × T. polonicum (PI190951), however, was the highest, with an average 13.70 bivalents per cell, the most frequent configurations being 7 I + 14 II (Fig. 1C). In T. petropavlovskyi × T. polonicum (AS304), an average 13.07 bivalents and 60% of cells with 13 or 14 bivalents were observed (Fig. 1D). The c-value of T. petropavlovskyi × T. polonicum (PI190951) was the highest among all crosses with tetraploid wheat (p<0.01) (Table 3).

In all nine hexaploid hybrids (2n = 6x = 42), mean chromosome pairing ranged from 18.50 to 20.85 bivalents (Table 3). In the combinations between T. petropavlovskyi and the Chinese endemic wheat landraces, mean pairing configuration ranged from 19.85 to 20.85 bivalents per cell (Fig. 1E). The c-value of T. aestivum cv. Changning white wheat was higher than that of the other combinations (p<0.01) (Table 3). An average 18.65 bivalents per cell was observed at MI in hybrids of T. petropavlovskyi × SHW-DPW, most cells containing 18 or 19 bivalents (Fig. 1F).

Seventeen hybrids were produced with tetraploid and hexaploid wheat plants having SHW-DPW as female parent. Between SHW-DPW and tetraploid wheat, mean chromosome pairing ranged from 10.70 to 13.85 bivalents per cell (Table 3). Trivalents were observed only in the SHW-DPW × T. durum combination (Fig. 1G). The highest number of bivalents was observed in the SHW-DPW × T. polonicum combination, with an mean pairing configuration of 7.30 I + 13.85 II and c-value of 0.88 (Fig. 1H). The c-value of SHW-DPW × T. polonicum was significantly higher than that of the other crosses (p<0.01) (Table 3).

Chromosome pairing at MI in SHW-DPW × T. aestivum ssp. tibetanum showed an average 19.35 bivalents per cell with a c-value of 0.76 (Fig. 1I). In the crosses between SHW-DPW and the Sichuan white wheat complex, a large number of univalents (average of 10.25 per cell) was observed in hybrids in SHW-DPW × T. aestivum cv. Chinese Spring (Fig. 1J).

Mean chromosome pairing in SHW-DPW × T. petropavlovskyi ranged from 17.20 to 19.30 bivalents. Nearly 20 bivalents were observed in the combination between SHW-DPW and T. petropavlovskyi (AS362) (Fig. 1K). The meiotic configuration of this hybrid was 3.40 I + 19.3 II. The c-value of SHW-DPW × T. petropavlovskyi (AS362) was significantly higher than that of the other combinations (p<0.01) (Table 3).

Meiosis in hybrids was more irregular at later stages, especially in hybrids of tetraploid wheat accessions. Lagging chromosomes and chromosome bridges along with fragments were observed at anaphases I and II in some hybrids (Fig. 1L).

In previous cytological studies, seed set, fertility, and meiotic pairing behavior indicated that T. petropavlovskyi is more closely related to T. aestivum cv. White head than to other hexaploid wheat. In addition, the relationship between T. petropavlovskyi and T. polonicum was found to be distant compared to other tetraploid wheat landraces (Yao et al., 1983; Chen et al., 1985). In the present study, the statistical analysis of data on seed set, fertility, and meiotic pairing behavior indicated that T. petropavlovskyi × T. polonicum and T. petropavlovskyi × T. aestivum cv. Chinese were significantly higher than that with any other cross (p<0.01). These results indicate that the relationships between T. petropavlovskyi and T. polonicum, and Sichuan white wheat complex are closer than the other tetraploid and hexaploid wheats, which is in agreement with the results of Yao et al. (1983) and Chen et al. (1985). Moreover, based on the results of seed set, fertility and meiotic pairing behavior of the hybrid between SHW-DPW and tetraploid, and hexaploid wheat, we found that SHW-DPW might be different from T. petropavlovskyi.

In short, the relationship between T. petropavlovskyi and domestic wheat species, especially those from the Sichuan white wheat complex, is closer than that with exotic wheat landraces. Furthermore, the chromosome pairing results indicated that T. polonicum might have played a role in the origin of T. petropavlovskyi, and that SHW-DPW and T. petropavlovskyi are different from one another.

It has been reported that the spike of T. petropavlovskyi is similar to that of T. polonicum (Udaczin & Miguschova, 1970; Kang et al., 2010). The genes for long glume in T. polonicum and T. petropavlovskyi were located on the long arm of chromosome 7A and are allelic (Dorofeev et al., 1979). A phylogenetic classification with molecular markers indicated that T. petropavlovskyi is more closely related to T. polonicum than T. durum and T. turgidum (Akond & Watanabe, 2005). The phylogenetic relationship analysis of Acc-1 and Pgk-1 gene pointed out that the T. petropavlovskyi and T. polonicum from Xinjiang are clustered in one group (Kang et al., 2010; Chen et al., 2013). Based on our results, the hybrids between T. petropavlovskyi and tetraploid wheat showed that the bivalents, seed set, and fertility of F₁ hybrids were significantly higher in the cross T. petropavlovskyi × T. polonicum compared to the other cross combinations. Our findings show that T. petropavlovskyi is more closely related to T. polonicum than to any other tetraploid wheat.

Molecular analyses indicated that T. petropavlovskyi is genetically distinct from three other Chinese endemic wheat landraces (Wei et al., 2002). UPGMA clustering, estimated from AFLP, suggested a similar genomic constitution of T. aestivum and T. petropavlovskyi (Akond & Watanabe, 2005). Phylogenetic relationship analysis of Acc-1 sequences provided additional evidence of a close affinity between T. petropavlovskyi and exotic landraces of T. aestivum (Kang et al., 2010). In contrast, our results indicate that the relationship of T. petropavlovskyi with native T. aestivum is closer than with exotic T. aestivum. The relationships of T. petropavlovskyi with the other three Chinese endemic wheat landraces and with exotic primitive wheat need further research. In addition, SHW-DPW was used as parental plant in crosses with tetraploid and hexaploid wheat and T. petropavlovskyi. Based on the results of seed sets, fertility of F₁ hybrids and chromosome pairing, we speculate that SHW-DPW is different from T. petropavlovskyi, and consider the possibility of hypothesis no. 2 that T. petropavlovskyi originated from an independent allopolyploidization event seems unlikely. Based on cytological analyses (Yao et al., 1983; Chen et al., 1985), agronomic and morphological studies (unpublished), and the results of the present study, we also discard the hypothesis no. 1 that T. petropavlovskyi is derived from a single mutation in T. aestivum. We consider most likely the hypothesis no. 3 that T. petropavlovskyi probably derives from a natural cross between T. aestivum and T. polonicum via either spontaneous introgression or breeding effort.