3.1 The haplotypes and the nucleotide polymorphisms of two cpDNAs intergenic regions ndhC-trnV and trnR-atpA
The analysis of haplotypes and the nuclotide polymorphisms in two cpDNAs intergenic regions indicated that the haplotype number (H) of ndhC-trnV and trnR-atpA were 17 and 20, respectively（Table 3）. The haplotype polymorphism (Hd) was 0.752 and 0.720, respectively; nucleotide polymorphisms (Pi) were 0.00325 and 0.00923, respectively. Tajima test (D) values were -1.88041 and -1.53301, both of which were negative, suggesting the evolution mode of sequences was negative selection. The P value of ndhC-trnV was P<0.05, indicated the significant differentiation between the number of segregative sites and the value of nucleotide polymorphisms in each sequences, and this intergenic region did not follow the neutral model during evolution. While the P value of trnR-atpA was P>0.10, indicated no significant difference between the number of segregative sites and the value of nucleotide polymorphisms in each sequences, and this intergenic region followed the neutral model during evolution.
Without considering missing/gaps and identical sites, the haplotype number of ndhC-trnV was 17. The haplotypes in Oriental pear were cH2-cH7, cH17 (cH stands for ndhC-trnV haplotype), and the main haplotype was cH2 and cH4 (table 4). The richest haplotypes were in P. pyrifolia with four haplotypes, maybe because the accessions were originated from different places, such as Zhejiang, Fujian, Japan, and so on, of which the different environment led to evolutions of them. Both P. bretschneideri and P. ussuriensis had two haplotypes of cH2 and cH4. P. betulafolia, P. calleryana and P. pashia has 1, 2, 3 haplotypes, respectively. The main haplotype of three of them was cH2. Though different haplotypes from wide populations were chosen as representative samples, the haplotypes were not rich for them in this study. There might be not too much divergence during the evolution of these two intergenic regions in chloroplast, and might also because Indels were not considered. The haplotypes in Occidental pear were cH8-cH16, and the main haplotype was cH8. There were two haplotypes in P. pyraster, P. cordata, P. nivalis, P. elaeagrifolia, P. salicifolia and P. syriaca, and only one haplotype in the rest species of Occidental pear, might because of limited accessions in each species. Moreover, P. caucasica 684 had cH4 haplotype which was one of the main haplotype in Oriental pear. Pears from Pakistan had haplotypes of cH5, cH8, cH15 and cH16. ‘bali 1’ was proposed to be Oriental pear, while ‘bali 2’, ‘bali 3’, ‘bali 4’ might be Occidental pear, but the species of them were still unknown.
Without considering missing/gaps and identical sites, the haplotype number of trnR-atpA was 19, of which rH1 (rH stands for trnR-atpA haplotype) was the haplotype of red flesh apple from Malus. There were six haplotypes in Oriental pear, which were rH3-rH6, rH8, rH18, and the main haplotype was rH5 (table 5). Haplotype in most of Oriental pears was simple, for example, P. bretschneideri, P. pyrifolia, P. ussuriensis and P. calleryana had only one haplotype of rH3, while P. pashia and P. betulafolia had two haplotypes, respectively.
There were 13 haplotypes in Occidental pear, which consisted of rH2, rH7-rH20 except rH8 and rH18, and the main haplotype was rH2 and rH11. The numerical order of haplotypes was: P. elaeagrifolia (four)> P. nivalis & P. spinosa & P. mamorensis (three)> P. communis & P. pyraster & P. cordata (two)> P. regelii & P. syriaca & P. gharbiana (one), which had richer haplotypes than Oriental pear. Three pear cultivars from Pakistan had haplotypes of rH2, and only one had rH6 haplotype. In the NN network, they had closer relationships with Oriental pears (Figure 3); P. caucasica 684 still had the haplotype of rH18 which was only in Oriental pear. Combined the results from ndhC-trnV above, the female parent of P. caucasica 684 might be from Oriental pear.
Without considering missing/gaps, the number of haplotypes in ndhC-trnV and trnR-atpA were 87 and 64, respectively, which suggested too much Indels between these two intergenic regions. Besides, there were too many A, T, or polyA, polyT structure in these two intergenic regions, which bring difficulty to PCR amplification and sequencing.
3.2 Sequence comparison between two cpDNA intergenic regions and NIA
cpDNA had the characteristics of maternal inheritance, therefore there was only one sequence after amplification. The aligned sequence length of two intergenic regions from cpDNA (ndhC-trnV and trnR-atpA) was 1066 and 1213 sites (including gaps); the number of parsimony informative sites was 94 (8.8%) and 36 (3.0%); the number of coding Indel sites was 121 and 120, respectively (Table 6).
NIA was one of low-copy nuclear genes, and one or two sequences were obtained after PCR amplification in Pyrus (Figure 4). The aligned sequence length of NIA was 2118 sites (including gaps); the number of parsimony informative sites was 435 (20.5%); the number of coding Indel sites was 108 (Table 4). All sequences were submitted to GenBank, and the phylogenetic trees were saved in TreeBase (www.treebase.org).
In the sequences of two intergenic regions from cpDNA and NIA, the best models for ndhC-trnV and trnR-atpA were F81+G, HKY, while NIA-i3 was TPM2uf+G (Table 4). These models were used in the following analysis.
3.3 The Neighbor-Net phylogenetic network analysis of two intergenic regions from cpDNA
3.3.1 The Neighbor-Net phylogenetic network analysis of ndhC-trnV
The phylogenetic trees constructed with ndhC-trnV by Maximum Likelihood method did not separate the Oriental pear from Occidental pear obviously (Figure was not shown), therefore Neighbor-Net phylogenetic network was constructed based on uncorrected-P genetic distance. The results were more complicated than the previous results (Zheng et al., 2014), which several splits cross connected to form a complicated network (different color showed in Figure 3a). Most accessions were connected by red splits, eg. fifty-five accessions with main haplotype of cH4 in Oriental pear, ‘bali 1’ with cH5 haplotype; thirty-nine accessions with main haplotype of cH8 in Occidental pear, ‘bali 2’ (cH16), ‘bali 4’ with cH15, P. salicifolia 869 (cH11), P. cossonii 828(cH12), P. cossonii 829 (cH12). P. pashia P5-11 (cH6), P. pashia P20-3 (cH7) in Oriental pear, and P. nivalis 1590 (cH9), P. elaeagrifolia (cH9), P. communis ‘cascade’ (cH10), P. mamorensis (cH1, cH13), P.pyraster (cH14) in Occidental pear, and outgroup Malus (cH1) were mainly connected by green and purple splits. P. mamorensis had the same haplotype with outgroup, suggested that it might be more primitive species. Moreover, in the NN network based on Neighbor Joining tree (Figure 3b), cH6, cH7, cH1, cH9, cH10 and cH14 were connected between cH4 and cH8 which were main haplotypes of Oriental pear and Occidental pear, indicated the accessions with these haplotypes might be transitional species between Oriental pear and Occidental pear.
3.3.2 The Neighbor-Net phylogenetic network analysis of trnR-atpA
The phylogenetic trees constructed with trnR-atpA by Maximum Likelihood method separated the Oriental pear from Occidental pear obviously (Figure was not shown), but there was not too much differences among each species, thus the phylogenetic relationships among them were not well solved. Neighbor-Net phylogenetic network was constructed for trnR-atpA, and the results were more complicated than that of ndhC-trnV (Figure 4). The results showed that the relationships of interspecies and intraspecies in Oriental pear from Occidental pear were complex, and Malus was closer with Occidental pear than Oriental pear. P. dimorphophylla 6, ‘bali 1’, P. caucasica 687, P. salicifolia (rH15) were connected with red splits. P. pyrifolia ‘Rentouli’ (rH4) and P. regelii 890 (rH19) were connected by blue splits. rH18 (P. caucasica 684 and P. betulaefolia ‘BT1’), rH7 (P. pyraster 881, P. spinosa 1608), rH17 (P. nivalis 256) and rH2 were linked indirectly by orange splits. rH8 (P. pashia P20-3) and rH16 (P. elaeagrifolia 2482) were joined together by green splits. rH10 (P. cordata 750) and Oriental pears were linked by aqua blue. Furthermore, in the NN network based on Neighbor Joining tree (Figure 4b), P. pashia P20-3 (rH8), P. caucasica 684 (rH18), P. betulaefolia ’BT1’ (rH18), P. elaeagrifolia 1482 (rH16), P. regelii 890 (rH19), P. elaeagrifolia 1490 (rH9) were in between the main haplotypes of Oriental pear from Occidental pear, to further suggest that P. pashia, P. betulaforlia, and P. regelii might be transitional species or relic species.
3.4 The phylogenetic analysis of NIA-i3
3.4.1 The phylogenetic trees of Pyrus constructed by NIA-i3
The preliminary experiment showed that NIA-i3 had two copies in both Malus and Pyrus. These two copies might be appeared before these two species evolved, latterly, Indels and mutation led to the differentiation of species during evolution. After clone and sequencing, one to five NIA-i3 sequences were obtained for each accession. Phylogenetic trees were constructed by Maximum Likelihood method to choose two sequences with less difference, which were used for the further analysis. However, two sequences with less difference were in one clade for some accessions, eg. P. bretschneideri (Cili), P. pyrifolia (Manding xueli, Hepi li), P. ussuriensis (Xiaoxiangshui, Manyuanxiang), P. Sinkiangensis (Kuerlexiangli), P. dimorphophylla 5, P. betulaeforlia ‘QS3’, P. xerophila 3, P. pyraster 1671, P. regelii 2513, P. salicifolia 2427, P. syriaca 908, P. spinosa 1516, P. mamorensis 1622, P. communis ‘comice’, more clones were needed to confirm if there was new genes, thus these species were not included in the following analysis and discussion except ‘Kuerlexiangli’.
After preliminary selection, two copies were kept for each accession for further analysis. A total of 93 accessions in Pyrus were used for constructing phylogenetic trees by Maximum Likelihood. The results separated the Oriental pear (A) from Occidental pear (B), of which there were seven clades (A1-7, B1-7) for them, respectively (Figure 5). Accessions with same Indels were clustered in the same clade (Figure 5), such as accession copy in clade A1 had one T missing (Del 1(T), and so forth); Del 1 (A) were showed in clade A2 and A3 except for P. pyraster 989_2; There were Del 6 (GGAGCT) and Del 1 (A) in clade A4, and Del 1 (C) in clade A5. Del 9 (CGCATATCA) was in clade B1, Del 4 (GTGT) in B2, Del 15 (TCCTTGAGCCTTGGA) in B6, Ins 5 (TTTCA) in B7b, Del 13 (CACAAGAATCTTC) in B7b. Though both ‘Yali’ and P. caucasica 684 had Del 6 (GGAGCT) and Del 1 (A), they did not joined together with P. ussuriensis (A4) missing these two Indels, which might be caused by more SNPs difference in P. ussuriensis.
The phylogenetic trees constructed by NIA had many polytomies, for instance, P. bretschneideri, P. pyrifolia and P. ussuriensis were still mingled together in clade A1 and A3, while European species, West Asian species and North African species in clade B5 were still not separated. The results also showed coevolution of species, for instance, one copy of P. ussuriensis (A4), P. calleryana (A2), P. pashia (A5), P. syriaca (B2), P. regelii (B3) and North African species (B7b) were in the same clade. In addition, two copies from part of P. ussuriensis accessions and both of two P. xerophila accessions were clustered in clade A3 and A4, respectively, demonstrating P. xerophila had close relationships with P. ussuriensis.
In Oriental pear, (1) majority of two copies of Chinese White Pear and P. Pyriforlia were in clade A1 and A3, the rest were in clade A2 (CWP ‘Huangjitui’_1, P. pyrifolia ‘Dalinaitou’_2), A5 (P. pyrifolia ‘Dalihuoba’_1), A7 (CWP ‘Yali’_2); (2) two copies of P. ussuriensis were mainly in clade A1, A3/A4, but two copies of P. ussuriensis ‘Manyuanxiang’ were in clade A3 and A4, respectively; (3)both copies of P. Sinkiangensis were in clade A1; (4) all copies of P. pashia were in clade A5, except for P. pashia ‘P10-3’_1 (A3)和P. pashia ‘P23-4’_1 (A6); (5)both copies of P. betulaforlia were in clade A3; (6) all copies of P. calleryana were in clade A1 and A2 except for P. calleryana_1 (A7), P. calleryana ‘DP8-1’_1 (A5) and P. calleryana ‘58’_2 (A3); (7) both copies of P. xerophila were in clade A3 and A4; (8) both copies of P. hondoensis were in clade A3 and A4; (9) ‘Yali’ (Del 6), P. calleryana_1 and P. caucasica 684 (Del 6) were in clade A7, which was at the bottom of the tree.
In Occidental pear, (1) clade B1 were composed of P. regelii 2587_1, P. nivalis 1590_1 and P. elaeagrifolia 1278_2 with the 9bp deletions (Del 9), which was at the bottom of the tree; (2) ten accessions with 4bp deletions (Del 4) were in clade B2, including European species, West Asia species, North African species; (3) P. regelii comprised of clade B3; (4) clade B4 and B5 included all accessions from Occidental pear, of which all copies were from West Asian species and North African species except for P. nivalis 725_2 in clade B4; (5) both copies of three P. nivalis accessions, P. spinosa 1615_2 and P. salicifolia 2797_1 with Del 15 and Del 2 were in clade B6; (6) North African species (P. cossonii, P. mamorensis) were in clade B7b with 5bp insertions (Ins 5) in copies; (7) the copies in clade B7c had 13bp deletions except for P. spinosa 1598_2 (Del 1).
3.4.2 The phylogenetic relationships of interspecies in Oriental pear
The phylogenetic relationships of Chinese White Pear, P. pyrifolia and P. ussuriensis were still not well solved (Figure 5). P. pyrifolia, some accessions from P. ussuriensis and Chinese White Pear, were mingled together (A1, A3), and accessions could not be well distinguished. However, some accessions of P.ussuriensis had close relationships with P. xerophila 2 and P. xerophila 4 (A3, A4). P.ussuriensis and P. xerophila distributed in the north of China, they had overlapped area in geography, which also suggested P. ussuriensis was different from P. pyrifolia and Chinese White Pear. P. xerophila 2 might be interspecific hybrid between Oriental pear and Occidental pear, of which the maternal parent was Oriental pear, and the male parent might be Occidental pear or at least had complex background referring to Occidental pear (Zheng et al. 2014). Considered of the results here, one of parents might be P. ussuriensis, therefore, more accessions from P. xerophila and other species of Pyrus were needed to validate further.
P. betulafolia, P. calleryana and P. pashia were primitive species. The number of ventricles in P. betulafolia and P. calleryana were 2~3, while that in P. pashia were 3~5, which probably suggested P. betulafolia and P. calleryana were more primitive species. Three accessions of P. betulafolia from Heinan Queshan (QS), Gansu Ningxian (NX) and Shandong Chengyang (CY) were clustered in different subclades of A3, having close genetic relationships with P. pyrifolia. However, they are in different clades (Group I, Group III, Group IV) in the NJ tree for P. betulafolia populations based on accD-psaI (Zong et al., 2014). The samples were widely chosen in this study and differences among intraspecies were more outstanding, so did P. calleryana and P. pashia. Majority of wild pea pear from Zhejiang province were in clade A2, which formed polytomies with other Oriental pears except from P. pashia. Most copies of wild P. pashia from Yunnan province were in clade A5, which connected Oriental pear and Occidental pear. Furthermore, P. pashia ‘P10-3’_1 clustered with other Oriental pear in clade A3, while P. pashia ‘P23-4’_1 and Occidental pear ‘comice’ were in clade A6, indicating wild P. pashia might be transitional species between Occident pears and Oriental pears, and more primitive speices might be from southwest mountain are by evolution (Rubtsov, 1944).
Though P. dimorphophylla and P. calleryana were similar in morphology, they were different species. Zheng et al. (2014) thought that the parents of P. dimorphophylla 5 (haplotype tH3aH2) and P. dimorphophylla 6 (tH1aH6) were distinct. Both copies of P. dimorphophylla 5 were in clade A3, and they were only closed with P. calleryana 58_2. One copy of P. dimorphophylla 6 had close relationship with P. calleryana 58_2, while the other one was with most of wild pea pear in clade A2, which suggested that P. dimorphophylla 5 and P. dimorphophylla 6 were probably from distinct parents, and they might have close genetic relationship with wild pea pear.
In Occidental pear, two species of P. communis (P. caucasica and P. pyraster) contained one probable interspecific hybrid between Oriental pear and Occidental pear (P. caucasica 684, P. pyraster 989). Copies of P. pyraster 881 (B5, B7c) were in different clades, illustrating a complicated genetic background. There were rich Indels (Del 2, Del 4, Del 9, Del 11, Del 15, Ins 1(A)) in three accessions of European species P. nivalis. Two copies of P. nivalis 1590 were in clade B1 (Del 9) and B6 (Del 2 and Ins 1(A)), with close relationships with West Asian species except for P. syriacaI. P. nivalis 256 and P. nivalis 1196 were thought to be hybrids between P. amygdaliformis and P. navalis. Two copies of them were both in clade B2 (Del 4, Del 11) and B6 (Del 15), with close relationships with P. syriaca、P. spinosa, which was in accordance with previous results (Zheng et al., 2014).
The leaf characters were variable in West Asian species (P. elaeagrifolia, P. salicifolia P. regelii, P. spinosa, P. syriaca). Most of them were named after leaf shape, and they had complex phylogenetic relationships and mingled with West Asian species and North African species in the phylogenetic trees. P. elaeagrifolia had complicated genetic background, and their copies were in clade B1, B2, B4, B5, B7, A2. P. elaeagrifolia 2817, a possible interspecific hybrid between Oriental species and Occidental species, had copies in clade A2 and B4. P. elaeagrifolia 1278 was closed with P. regelii, P. nivalis and P. spinosa and had copies in clade B1 and B7. P. salicifolia was rich in Indels (Del 2, Del 4, Del 13, Del 15), and copies were in B2, B4, B5, B6, B7, indicating an abundant background. Though the leaf shape of P. salicifolia 2427 was distinct from other accessions of P. salicifolia, two copies of it were clusterd with copies from other accessions of P. salicifolia in clade B7. P. regelii was one of the most primitive species, and one copy of three accessions from P. regelii was in clade B2, which comprised of polytomies with other clades of Occidental pear. The phylogenetic position of P. regelii was unclear, which need more gene sequences to illustrate. P. spinosa was thought to be simple background, however, relationships were showed with P. nivalis and other West Asian speicies. One parent of P. spinosa 1598 was P. betulaforlia possibly (Jiang et al., 2014), but both copies were in clade B7 without lineage of Oriental pear, more clones were needed to verify. One copy of three accessions from P. syriaca were in clade B2 (Del 4), the other one was in clade B2 (Del 11), B5 and B7 (Del 13), demonstrating intimate relationships wiht P. salicifolia.
It was reported that P. cossonii was differentiated from P. mamorensi, P. gharbiana in North African species, because of no flavonoid glycosides in P. mamorensis, or the phylogenetic results from LFY2int2-N. For these three North African species, the haplotypes were different (cH12, cH8, cH13 and cH1) based on ndhC-trnV, while P. gharbiana and P. mamorensis had the same haplotype of rH11 based on trnR-atpA. However, there was Ins 5 in all accessions of North African except P. gharbiana 834 and P. mamorensis 1622 based on NIA-i3, of which these copies were in clade B7. Two copies of two accessions of P. gharbiana were in clade B5 and B7, suggesting they had simple background. For P. cossonii, P. cossonii 753 and two accessions of P. gharbiana were in the same cluster, suggesting intimate genetic relationships of them. One copy of P. cossonii 829 had Del 13, with close relationships with P. pyraster and P. salicifolia. Copies of the rest two accessions from P. cossonii and two accessions from P. mamorensis were in clade B4 and B7, with close relationship. P. mamorensis 834 and P. elaeagrifolia were closely related. All of these results indicated intimate relationships of these three species.
3.4.4 The classification of hybrids between Oriental pear and Occidental pear and synonym
Two copies of P. hondoensis and Occidental pear P. caucasica 684, P. pyraster 989, P. elaeagrifolia 2817 were in clade A and B, respectively, indicating they were possible hybrids between Oriental pear and Occidental pear.
P. caucasica 684 was a hybrid originated from Ukraind, of which the materal parent was probably P. xerophila of Oriental pear, and the male parent was probably P. cossonii and P. caucasica (Zheng et al., 2014). One copy of P. caucasica 684 was in clade A7 with Del 6 and Del 1 (A) which were same as P. ussuriensis and P. xerophila, while the other copy was in clade B5 with P. cossonii, P. caucasica. The maternal parent of P. caucasica 684 was proposed to be P. betulaforlia (cH4rH18) or P. xerophila or P. ussuriensis according to the results from two chloroplast intergenic regions of ndhC-trnV and trnR-atpA. P. pyraster 989 was differed from the other two accessions of P. pyraster with the haplotype of tH15aH18 on the basis of accD-psaI and trnL-F. One copy of P. pyraster 989 was clustered with P. pyrifolia, P. dimorphophylla and P. betulaforlia, the other copy was in clade B5 with other Occidental pear species which was differed from the other two accessions from P. pyraster. However, the parent of P. pyraster 989 was unknown, more accessions from other species were needed.
West Asian species P. elaeagrifolia 2817 had two copies in clade A2 and B4, respectively, with intimate relationships with wild pea pear and P. salicifolia or P. caucasica. However, It only had close relationships with P. salicifolia, and no evidence showed it was a hybrids between Oriental pear and Occidental pear. P. hondoensis was similar with P. ussuriensis in morphology. One copy in clade A3 was closely related with Chinese White Pear, P. pyrifolia and P. ussuriensis, the other one in clade B4 had intimate relationships with P. salicifolia, P. elaeagrifolia, P. caucasica and P. mamorensis. There was no report that P. elaeagrifolia 2817 and P. hondoensis were hybrids, more genes were needed.
‘Kuerle Xiangli’ was thought to be hybrid between Oriental pear and Occidental pear all the time. Though 10 clones were sequenced, no sequence was in clade with Occidental pear, probably only one copy was exist, or other copy could not obtained, more clones need to be sequenced.
P. elaeagrifolia 768, P. regelii 890 and P. cossonii 828 were in different clades, however, they were thought to be synonym (Jiang et al., 2014), which was not accordance with results of ndhC-trnV and trnR-atpA.