Introduction

The genus Pyrus L. (common name pear) is classified in the subtribe Pyrinae of the tribe Pyreae (Potter et al. 2007). These names have been corrected to Malinae and Maleae, respectively, based on recent changes to the International Code of Nomenclature for Algae, Fungi, and Plants (McNeill 2012), and about 20 primary species are generally accepted by most taxonomists. According to their original distribution areas, Pyrus is divided geographically into two groups: the Occidental pear and the Oriental pear (Bailey 1917). Occidental pears are further divided into three groups: West Asian species, North African species, and European species (Challice and Westwood 1973). Pyrus communis L. is the major cultivated species of Occidental pear, which has been produced widely throughout Europe, North and South America, and Africa. Cultivated pears native to Asia mainly belong to four species or types (Teng and Tanabe 2004): P. pyrifolia Chinese white pear group (sometimes mistakenly assigned as P. bretschneideri Rehd.), P. ussuriensis Maxim., P. pyrifolia Nakai, and P. sinkiangensis Yu, which is believed to be a hybrid of Chinese white pear or Chinese sand pear and Occidental pear (Teng et al. 2001). During the last decades, molecular markers, including random amplified polymorphic DNAs (RAPDs) (Monte-Corvo et al. 2000; Teng et al. 2001, 2002), amplified fragment length polymorphisms (AFLPs) (Bao et al. 2008; Monte-Corvo et al. 2000), and simple sequence repeats (SSRs) (Bao et al. 2007; Bassil and Postman 2010; Yamamoto et al. 2002; Yao et al. 2010), have been applied in Pyrus species to evaluate their genetic diversity and genetic relationships. However, the genetic relationships among some Pyrus species are still unclear owing to the extensive interspecific hybridization in Pyrus (Zheng et al. 2008, 2011, 2014). Recently, 42.4 % of the pear genome was reported to be long terminal repeat (LTR) retrotransposons (Wu et al. 2013), which implied that retrotransposons might be closely related to the evolution of Pyrus. Therefore, the development of new retrotransposon-based markers could be indispensable in further understanding the genetic relationship and origin of Pyrus.

Retrotransposons are a widespread class of transposable elements that exist in all plant species investigated to date (Sabot and Schulman 2006; SanMiguel et al. 1996; Wicker et al. 2007). They replicatively transpose by way of an RNA intermediate, and thus their copy numbers increase and occupy large fractions of the genomes, especially in higher plants. For example, 7.5 and 37.6 % of the Arabidopsis and apple genomes are estimated to be retrotransposons, respectively (Velasco et al. 2010). LTR retrotransposons is one of the most important retrotransposon families (Havecker et al. 2004; Peterson et al. 2002; SanMiguel et al. 1998; Vicient et al. 1999). LTRs are easy to find because of their presence as flanking sequences at the 5′ and 3′ ends of coding regions in the genome (Bergman and Quesneville 2007). They are suitable for developing new molecular markers because of their ubiquitous distribution, abundant copy number, and insertion polymorphisms (Flavell et al. 1992). Recently, retrotransposon-based markers, such as retrotransposon-based insertion polymorphisms (RBIPs) (Kalendar et al. 2011), inter-retrotransposon amplified polymorphisms (IRAPs) (Kalendar and Schulman 2006), retrotransposon-microsatellite amplified polymorphisms (REMAPs) (Kalendar and Schulman 2006), and sequence-specific amplification polymorphisms (SSAPs) (Waugh et al. 1997) have been developed in some species and widely used in studies of genetic diversity, phylogeny, genetic mapping, and cultivar identification.

RBIP identification involves the simple PCR-based detection of retrotransposon insertions using primers that flank the insertion site and primers from the insertion itself (Kalendar et al. 2011). Researchers have tried to develop RBIP markers in Rosaceae, but these efforts were constrained by the lack of known LTR sequences. Jing et al. (2005) carried out SSAP PCRs to obtain 52 RBIP markers, which are costly and time consuming. Kim et al. (2011, 2012) first isolated retrotransposons from a BAC library in Pyrus and developed 22 RBIP makers using BLASTN sequence similarity searches based on the LTR sequence of the copia-like retrotransposon Ppcrt4. Using these markers, 61 of 64 Japanese pear cultivars could be distinguished. However, a BAC library is too limited to identify the retrotransposons in pear, and the sequence homology analysis by BLASTN could not identify all the retrotransposons, especially retrotransposons that were specific to pear. Recently, the whole genome of Pyrus (Pyrus bretschneideri Rehd. cv. ‘Dangshansuli’) has been sequenced (Wu et al. 2013), which has opened the way for a faster and more economical approach to predicting retrotransposons and developing RBIP markers.

In the current study, we aimed to develop new RBIP markers across the whole pear genome using LTR harvest (Ellinghaus et al. 2008). We also validated 24 primers in 110 pear accessions covering nearly all species of the genus Pyrus to evaluate the genetic relationships among Pyrus species. The results will give new insights into the genetic diversity within the Pyrus species.

Materials and methods

Plant materials and DNA extraction

A total of 110 Pyrus accessions from Eurasia were used (Table 1). Accessions of Oriental pears were from China Pear Germplasm Repository, Xingcheng, Liaoning Province, China (CPGR), Gansu Pomology Institute, Gansu Academy of Agricultural Sciences, Lanzhou, Gansu Province, China (GPI), Tottori University, Tottori, Japan (TU), Wuhan Sand Pear Germplasm Repository, Wuhan, Hubei Province, China (WSGR), Zhengzhou Fruit Institute, Chinese Academy of Agricultural Sciences, Zhengzhou, Henan Province, China (ZZFI), and field collections. All accessions of Occidental pears were collected from the National Clonal Germplasm Repository, Corvallis, OR, USA (NCGR).

Table 1 Pyrus species and cultivars used in this study
Full size table

Total genomic DNA was extracted from the leaf tissues of the plants following the modified CTAB protocol described by Doyle and Doyle (1987). The DNA concentrations were diluted to 10–30 ng µL−1 after the quality and quantity were determined on 1 % (wv−1) agarose gels using standard DNA markers (Takara, Dalian, China).

Identification of LTR retrotransposons

The whole genome of P. bretschneideri Rehd. cv. ‘Dangshansuli’ (AJSU00000000) was used to predict retrotransposons. LTR harvest was run on the whole genome data using default settings to isolate LTR retrotransposons based on certain features, including two nearly identical LTR sequences flanked by target site duplications (TSDs) of 4–6 bp and some conserved sequence motifs, such as the poly purine tract (PPT) and the primer binding site (PBS) (Fig. 1a). The genomic sequences between any two putative LTRs were subsequently analyzed using BLASTN searches (the E value threshold was 10−5) against the GenBank non-redundant databases and repeat masking (Smit et al. 1996-2004) against Repbase (Kohany et al. 2006) to assign the putative LTR retrotransposons to the copia or gypsy family. The putative LTR retrotransposons in these two families were clustered (sequence length to be covered 70 %; percentage identity threshold 80 %) using Blastclust (Dondoshansky and Wolf 2002) to further classify the LTRs into the subfamilies. These LTR regions of the subfamilies were used to develop the RBIP primers.

Fig. 1
figure 1

Identification of LTR retrotransposons and RBIP primer design. a Characteristics of retrotransposons. TSD target site duplication, LTR long terminal repeat, PBS primer binding site, GAG gag protein domains, POL pol protein domains, PPT polypurine tract, F-primer forward primer, and R-primer reverse primer. b Alignment of five sequences between the 3′ LTR of Ppcr1 and the flanking sequences in the pear genome. F-primers are boxed; R-primers are underlined

Full size image

RBIP primer design

The LTR sequences in each subfamily were first aligned using ClustalW software (Larkin et al. 2007) to obtain conserved sequences and were then mapped to the pear genome using BLAST. Primer sets were designed using Primer3 software (Koressaar and Remm 2007) to amplify the junction regions between the 3′ LTR retrotransposon sequences and the flanking pear genome sequences (Fig. 1a, b). One primer in each set was designed within the 3′ LTR of the retrotransposon and labeled with a fluorescent chemical (FAM, HEX, PET, or NED); the other primer was designed within the pear genomic region.

PCR amplification and analysis

Twenty-four RBIP primer pairs from the Ppcr1 subfamily of copia retrotransposons were used for DNA profiling. PCR amplifications were carried out in a total volume of 15 µL [10 ng DNA template, 0.4 µM of each primer, 200 µM dNTPs, 2 mM MgCl2, and 0.5 U Taq DNA polymerase (Takara, Dalian, China)]. Amplifications were performed in a Mastercycler gradient PCR thermocycler (Eppendorf, Hamburg, Germany) programmed for an initial denaturation at 94 °C for 5 min, followed by 35 cycles of 94 °C for 40 s, 55 °C for 40 s, and 72 °C for 40 s, and then 72 °C for 6 min. Amplicons were pooled together with an internal size standard (GeneScan™ 500 LIZ; Applied Biosystems, Foster City, CA, USA) according to distinct dyes and expected fragment sizes, and subsequently separated and sequenced using an ABI 3700XL Genetic Analyzer (Applied Biosystems). Accessions were scored at each locus for the presence (1) or absence (0) of bands of the expected size.

Data analysis

The identification of a genetically homogeneous group in a data set is always challenging and is affected by several factors including the number of loci used, the magnitude and scale of gene flow, the variation at each locus, and the number of samples (Evanno et al. 2005). Therefore, we used two complementary methods to estimate the numbers of clusters at both non-hierarchical and hierarchical levels.

For non-hierarchical genotypic clustering, the number of homogeneous genepools (K) was modeled using the genotypes obtained from all 110 individuals in the software STRUCTURE version 2.3.3 (Pritchard et al. 2000). This revealed the genetic structure by assigning individuals or predefined groups to K clusters. Ten runs of STRUCTURE were performed by setting the number of clusters (K) from 1 to 10. Each run consisted of a burn-in period of 200,000 iterations followed by 200,000 Monte Carlo Markov Chain iterations, assuming an admixture model. The results were uploaded to the STRUCTURE HARVESTER web site (Earl and VonHoldt 2012) to estimate the most appropriate K value. The replicate cluster analysis of the same data resulted in several distinct outcomes for estimated assignment coefficients, even though the same starting condition was used. Therefore, we employed CLUMPP software (Jakobsson and Rosenberg 2007) to average the 10 independent simulations and illustrated the result graphically using DISTRUCT (Rosenberg 2004).

All data were used to calculate Dice’s similarity coefficients (Nei and Li 1979). An unweighted pair group method with arithmetic mean (UPGMA) dendrogram based on the similarity matrix was constructed using the NTSYS-pc program (Version 2.10e) (Rohlf 1998).

Results

Identification and classification of LTR retrotransposons

Putative LTR retrotransposons were identified from the pear genome using LTR harvest. First, we identified all the sequences that were flanked by two highly similar sequences (>90 %) between 100 and 1,000 bp in length. Using these criteria, 1,836 full-length LTR retrotransposons were mined, making up 2.3 % of the pear genome (size 512 Mb). The identified sequences were then masked against the Repbase database using RepeatMasker to classify the retrotransposons into different families. A total of 689 retrotransposons (37.5 %) were assigned to the Ty1-copia family, 353 retrotransposons (19.2 %) were assigned to the Ty3-gypsy family, and 794 retrotransposons were unclassified and, therefore, excluded from the subsequent analysis (Table 2). The copia retrotransposons (689) were further clustered into 195 subfamilies of which 53 % contained a single sequence, 15 % contained two sequences, and 32 % contained three or more sequences. The gypsy retrotransposons (353) were further clustered into 193 subfamilies of which 84 % contained a single sequence, 8 % contained two sequences, and 8 % contained three or more sequences.

Table 2 Number of putative full-length LTR retrotransposons in the pear genome
Full size table

Development and assessment of RBIP markers

To obtain more insertion sites, the conserved LTRs of the top five copia and gypsy subfamilies (KF806690–KF806699) were mapped against the Pyrus genome using BLAST to locate them in the genome and to obtain their flanking regions, which were used to develop markers (Table 3). A total of 196 RBIP primer sets were designed to amplify the specific junction regions between the LTR sequences and their flanking sequences (Fig. 1). The sizes of the PCR products were estimated to vary from 113 to 534 bp. Twenty-four RBIP primer pairs from Ppcr1 were used for DNA profiling (Table 4), and only the forward primer in the 3′ LTR region was designed with fluorescent chemicals to save costs. Additional RBIP primers developed from other retrotransposon subfamilies are listed in Supplementary data 1.

Table 3 Top five subfamilies in the copia and gypsy families in Pyrus
Full size table
Table 4 RBIP primers designed from the Ppcr1 subfamily of copia retrotransposons in Pyrus
Full size table

All of the 24 loci were polymorphic in our preliminary data analysis and were thus employed for the analysis of the main data set (Supplementary data 2). The data matrix contained 2,640 data points. All the primers showed single clear amplified bands of the expected sizes in ‘Dangshansuli’. Seven primers (Ppcr1M3, Ppcr1M8, Ppcr1M10, Ppcr1M14, Ppcr1M15, Ppcr1M16, and Ppcr1M23) amplified fragments in more than 50 % of the pear accessions. Ppcr1M3 successfully generated amplified fragments in all pear accessions except two Japanese pear cultivars (‘Chojuro’ and ‘Hakataao’). Ppcr1M8 amplified fragments in P. pyrifolia Chinese white pear group, P. pyrifolia and P. ussuriensis, but not in P. betulaefolia and P. communis. Primer Ppcr1M10 amplified fragments within all Oriental pear accessions except ‘Daxiangshui’, ‘Xiaoxiangshui’, ‘Korlaxiangli’, and ‘P. xerophila’. Three primers (Ppcr1M14, Ppcr1M15, and Ppcr1M16) produced amplicons mainly in Oriental pears and rarely in Occidental pears. Ppcr1M23 amplified fragments in most Oriental pear cultivars, but not in Occidental pears or Oriental wild pear. Ppcr1M20 amplified fragments in accessions originating from Europe, and Ppcr1M4 produced amplicons in accessions originating from West Asia and North Africa. Six primers (Ppcr1M1, Ppcr1M2, Ppcr1M7, Ppcr1M12, Ppcr1M18, and Ppcr1M22) failed to produce fragments in the Occidental pear species. Out of all the primers, Ppcr1M1 and Ppcr1M5 amplified fragments only in a few accessions.

Genetic relationships among Pyrus species and cultivars

Bayesian modeling of the number of homogeneous genepools (K) in STRUCTURE was used to infer the genetic structure of the Pyrus accessions. An evaluation of the optimum number of K following the procedure described by Evanno et al. (2005) indicated two clear optimal values for ∆K, at K = 2 and 4 (Fig. 2), suggesting that a model with two genepools captured a major split in the data and that substantial additional resolution was provided under a model with K = 4. Barplots of the proportional allocations in STRUCTURE to each genepool for K = 2 and 4 are shown in Fig. 3. The plots showed that these two models were related to each other hierarchically, such that the green cluster in the two-genepool model was subdivided into two (blue and red) genepools in the four-genepool model.

Fig. 2
figure 2

Modeling of cluster number for Pyrus using STRUCTURE. LnP (K) and Delta K were calculated in accordance with the method of Evanno et al. (2005)

Full size image
Fig. 3
figure 3

Genetic relationships among the 110 accessions of Pyrus revealed by a Bayesian modeling approach under K = 2 (top) and K = 4 (bottom)

Full size image

The primary split in the data (K = 2) divided the accessions into two clusters. One cluster (yellow in Fig. 3) included Oriental pear cultivars belonging to P. pyrifolia Chinese white pear group, P. pyrifolia and P. ussuriensis. The other cluster (green) was comprised of major samples from P. betulaefolia, P. xerophila, and Occidental pears. The accessions that demonstrated a high level of admixture belonged to P. pashia and P. calleryana. The model with four genepools was also supported by the STRUCTURE results. Under this model, the green cluster in the model of K = 2 was further divided into two genepools (Fig. 3). One (red) contained the wild Asian pears including P. pashia, P. betulaefolia, P. dimorphophylla, and P. calleryana. The other (blue) was composed of the Occidental pears and an Oriental species, P. xerophila. The remaining two genepools, which contained cultivars of P. pyrifolia Chinese white pear group, P. ussuriensis and P. pyrifolia, overlapped substantially with one another, and the hierarchical levels in these two clusters could hardly be recognized. In the P. pyrifolia Chinese white pear group, all accessions appeared to be from the two major genepools. Several accessions in P. ussuriensis and P. pyrifolia showed admixed origins, such as ‘Baozhuli’ with three major genepools and ‘Yaguang’ with four genepools. Some Occidental accessions, such as ‘Cascade’, ‘Comice’, ‘P. spinosa 1598’ and ‘P. regelii 2513’ should be noted because they displayed admixtures of the genepool.

An UPGMA dendrogram was generated from the genetic similarity values to elucidate the genetic relationships among Pyrus species and cultivars (Fig. 4). Oriental pear accessions and Occidental pear accessions had the lowest genetic similarity values, while most of the cultivars from Asia shared relatively closer affinities. The dendrogram could be divided into two major groups. The majority of Group 1 was composed of Oriental pear accessions, including cultivars and wild accessions, and two Occidental cultivars, ‘Cascade’ and ‘Comice’, while Group 2 included the Occidental species and an Oriental species, P. xerophila. This was similar to the STRUCTURE results at K = 2 that classified all pear accessions into two groups, but the Oriental wild species were clustered with the Occidental pears (Fig. 3). However, in the model of K = 4, the majority of Oriental wild pears was separated from the Oriental pear cultivars and Occidental species.

Fig. 4
figure 4

Dendrogram of 110 pear species and cultivars based on their genetic similarity coefficients. UPGMA cluster analysis based on Dice’s similarity coefficients (Nei and Li, 1979) was used to generate the dendrogram. ‘○’: P. pyrifolia Chinese white pear group; ‘●’: P. pyrifolia; ‘∆’: Japanese pear; ‘▲’: P. ussuriensis; ‘□’: P. sinkiangensis; ‘♠’: P. communis; ‘♣’: Species originating from Europe; ‘♦’: Species originating from West Asia; and ‘◊’: Species originating from North Africa

Full size image

Within Group 1, four subgroups (I–IV) could be identified (Fig. 4). The dendrogram indicated the P. pyrifolia Chinese white pear group, P. ussuriensis and P. pyrifolia, were generally mingled and concentrated in subgroup I. Two P. communis cultivars, ‘Cascade’ from the USA and ‘Comice’ from France, clustered with ‘Korlaxiangli’ in subgroup II (Fig. 3). Subgroup III consisted of P. calleryana, P. pashia, and P. dimorphophylla. In subgroup IV, four samples of P. betulaefolia were clustered together. In Group 2, three subgroups (V–VII) were identified. Accessions of P. xerophila native to China clustered into subgroup V. Subgroup VI included pear species from Europe, and subgroup VII consisted of species from West Asia and North Africa, except P. cordata, which originated from Europe.

Discussion

Isolation and characterization of LTR retrotransposons

Genome data enable the prediction of retrotransposons that were previously unreachable. Recently, LTR retrotransposons have been predicted in many plants using genome data. For example, 2,226 LTR retrotransposons were found in the rice genome (Baucom et al. 2009) and 1,479 LTR retrotransposons were identified in the Populus genome (Cossu et al. 2012). Recently, with the completion of the Chinese white pear genome, 42.4 % of the genome was reported to be LTR retrotransposons (Wu et al. 2013), while 2.3 % of the pear genome was identified to be full-length LTR retrotransposons in the present study. The obvious difference in the number of LTRs (42.4 vs. 2.3 %) might be attributed to the different approaches and parameters that were used in the two studies. Additionally, only putative full-length LTR retrotransposons with two very similar LTR sequences were isolated in this study. Our results showed that full-length copia retrotransposons were more common than gypsy retrotransposons in the pear genome (Table 2). The numbers of full-length LTR retrotransposons in the different subfamilies were generally low, and gypsy subfamilies had more single sequences than the copia subfamilies. Only 10 subfamilies contained more than 10 LTR retrotransposons (Table 3). These findings are consistent with the results reported in other plants with different genome sizes (Schnable et al. 2009; Cavallini et al. 2010; Cossu et al. 2012).

Characteristics of RBIP markers and diversity of LTR retrotransposons in Pyrus

New bioinformatics software offers exciting perspectives for the development of new markers based on whole genome sequences. In the pear genome, the most abundant retrotransposon families were gypsy and copia, accounting for 25.5 and 16.9 % of the genome, respectively. However, simple sequence repeats contribute only 0.22 % of the genome (Wu et al. 2013), which indicated that RBIP markers were more ubiquitous than SSR markers in pear. Although a large number of SSR markers have been developed from pear genomic DNA (Fan et al. 2013), almost all the SSRs have di-nucleotide repeat motifs, and “stutter bands” are frequent (Diwan and Cregan 1997). In addition, repeat instability and an increased mutation rate in repeat lengths might be a problem for SSR markers (Wierdl et al. 1997; Yamada et al. 2002), and these do not occur in RBIP markers. RBIP markers also allow DNA profiling and the evaluation of genetic diversity, which was similar with SSR markers. However, RBIP markers amplify a single locus in samples, which is different from SSR markers that potentially amplify two or possibly more homologous loci. Compared with other retrotransposon-based markers (IRAP, REMAP, and SSAP), which display polymorphisms in band sizes owing to retrotransposon insertion, RBIP markers can detect the presence or absence of the retrotransposons in a locus produced by the integration of an element (Kalendar et al. 2011).

In our developed RBIP markers, each primer was related to the insertion of Ppcr1 retrotransposons in a particular locus. Our results showed that Ppcr1 LTR retrotransposons existed in all pear species, even if the species were believed to be ancestral to Pyrus, such as P. betulaefolia and P. calleryana (Zheng et al. 2011), which suggested that Ppcr1 retrotransposons widely existed in the pear species for a long time. In the wild Asian pears and Occidental pears, several primers did not amplify fragments, suggesting that Ppcr1 retrotransposons were not found in these loci. However, the insertion of Ppcr1 retrotransposons was extensively detected in many Asian pear cultivars. These results implied that Ppcr1 retrotransposons replicated many times in the development of cultivated Asian pears, which might be one reason why 42.4 % of the genome was reported to be LTR retrotransposons in ‘Dangshansuli’.

Genetic relationships among Pyrus species

This study was the first report to assess the genetic relationships of nearly all species of Pyrus based on DNA markers. The results of STRUCTURE and the dendrogram were in good agreement with the previous results of Oliveira et al. (1999), Monte-Corvo et al. (2000), Teng et al. (2001, 2002), and Bao et al. (2007, 2008), who, using a limited number of Pyrus species, divided Pyrus into Oriental pears and Occidental pears based on AFLP, RAPD, and SSR markers. These results support the traditional view that the genus Pyrus consists of two geographic species groups: Oriental pears and Occidental pears.

Four wild Oriental pea pear species composed subgroup III and subgroup IV in the UPGMA dendrogram (Fig. 4). P. dimorphophylla, P. calleryana, and P. pashia were clustered together in subgroup III. Four accessions of P. betulaefolia from different places clustered together in subgroup IV and were located between Occidental pears and other Oriental pears, which is consistent with the results obtained from RAPD markers (Teng et al. 2002) and AFLP markers (Bao et al. 2008). In the model with two genepools in STRUCTURE (Fig. 3), we noticed that the green cluster mainly contained Occidental pears and two wild Asian species, P. xerophila and P. betulaefolia, native to North China (Yu 1979). These results indicated that the Occidental pear species’ cluster was linked to the Oriental pear cluster through P. betulaefolia, as had been proposed by Challice and Westwood (1973). However, in the model with four genepools in STRUCTURE (Fig. 3), P. dimorphophylla, P. calleryana, P. pashia, and P. betulaefolia were allotted predominantly to one cluster, inferring their relatively close relationship. Two accessions in P. dimorphophylla appeared to be of different origins, P. dimorphophylla 5 with four genepools and P. dimorphophylla 4 with two major genepools, suggesting P. dimorphophylla had a complex genetic background.

Pyrus xerophila is a wild Oriental pear mainly found in Gansu Province, China, where cultivars of P. communis have been introduced through the Silk Road and cultivated for a long time (Teng and Tanabe 2004). In a previous study, we proposed that P. xerophila might be an ancient genetic recombinant that arose by interspecies hybridization involving Oriental and Occidental species based on the analysis of Adh and LEAFY sequences (Zheng et al. 2011). In the results of the RAPD markers (Teng et al. 2001), P. xerophila clustered loosely with ‘Beijing Baili’, a famous P. ussuriensis cultivar and shared several common RAPD bands with the Occidental pears, implying some relationship between P. xerophila and P. ussuriensis, as well as the Occidental pears. In this study, STRUCTURE results (K = 4) (Fig. 3) showed that P. xerophila was composed of one major genepool that mostly appeared in Occidental pears, and the UPGMA dendrogram (Fig. 4) also indicated that P. xerophila was clustered with the Occidental pears, which further confirmed their close genetic relationship. However, a more extensive sampling of P. xerophila should be carried out to elucidate the origin of P. xerophila.

Pyrus hondoensis, native to Japan, was once classified as a variety of P. ussuriensis. In the UPGMA dendrogram, P. hondoensis and two P. ussuriensis cultivars, ‘Mangyuanxiang’ and ‘Yaguang’, were clustered together in subgroup I, which supports our previous results with RAPD, AFLP, and SSR markers (Teng et al. 2002; Bao et al. 2007, 2008).

All of the Occidental pear accessions, except ‘Cascade’ and ‘Comice’, were clustered into two subgroups (Fig. 4). Accessions in Subgroup VI belonged to the European species, except for P. cordata, which occurs in the western margin of Europe, southwestern England, western France, and north-west of the Iberian Peninsula (Challice and Westwood 1973). P. cordata 1588 clustered with species from West Asia and North Africa in subgroup VII and was far from the other European species (Fig. 4), but it has a similar genetic background to some of the European species (K = 4, Fig. 3), which supports the view that P. cordata is a relict species linked to North African, West Asian, and East Asian species (Challice and Westwood 1973; Aldasoro et al. 1996). P. caucasica and P. pyraster were reported to have very similar botanical characteristics (Challice and Westwood 1973), and they were once treated as a variety of P. communis. In this study, accessions from P. caucasica and P. pyraster appeared in two subclades of subgroup VI, indicating that they have a close relationships with each other. Three of four accessions in P. caucasica were clustered with P. pyraster 1671 in one subclade, while P. caucasica 694 was with the other accessions of P. pyraster.

Subgroup VII consisted of pear accessions from West Asia and North Africa. In the results of STRUCTURE (K = 4), ‘P. spinosa 1598’ and ‘P. regelii 2513’ appeared to be of an admixed origin. The tomentose indumentum on both sides of the young leaves was a common feature in P. spinosa and was usually lost in the mature leaves in summer, which is very similar to those of P. betulaefolia. P. spinosa might be of a hybrid origin involving wild Asian pears, probably P. betulaefolia. Two accessions of P. regelii, ‘P. regelii 2513’ with two genepools and ‘P. regelii 890’ with one genepool were shown to be of different origins. More samples are needed for further studies. P. salicifolia was thought to be a synonym for P. nivalis or a hybrid of P. nivalis and P. communis (Aldasoro et al. 1996). However, in the UPGMA results, two accessions of P. salicifolia did not cluster with P. nivalis but grouped with species from West Asia and North Africa (Fig. 4). In the case of P. cossonii, native to North Africa, one genotype clustered with ‘P. spinosa 1598’ and P. salicifolia 2797, and the other grouped with ‘P. regelii 890’ and ‘P. elaeagrifolia 768’, which reflected its complex genetic background. Another North African species, P. mamorensis, was shown by STRUCTURE (K = 4) analysis (Fig. 4) to have at least two genepools, of which one genepool (red) mostly appeared in Oriental wild pears, suggesting that this species was more related to the Oriental wild species as indicated by Challice and Westwood (1973).

Hybrid origins of cultivars

Although cultivars of P. ussuriensis differed morphologically from other cultivated pears native to East Asia, all accessions from P. pyrifolia Chinese white pear group, P. pyrifolia, and P. ussuriensis were clustered together into subgroup I in the UPGMA dendrogram, and the STRUCTURE analysis also showed that their genotypes were assigned as admixture of two genepools (Fig. 3, K = 4). The results in this study were different from our previous results with RAPD (Teng et al. 2002) and SSR markers (Bao et al. 2007), where the cultivars from P. ussuriensis clustered independently from other species’ accessions. Iketani et al. (2012) reported that P. pyrifolia Chinese white pear group and P. ussuriensis from China also could not be distinguished strictly by SSR markers. In Pyrus, interspecific hybridization has been considered the major mode of evolution (Iketani et al. 2012; Zheng et al. 2008, 2011, 2014). P. pyrifolia Chinese white pear group and P. ussuriensis are extensively cultivated in North China. There was no reproductive barrier among these species and horizontal gene transfer may cause their poor resolution in the phylogenetic tree. Introgressive gene flow and hybrid speciation might be expected to have a higher prevalence between closely related species and populations. The STRUCTURE analysis (K = 4) showed some accessions in P. pyrifolia with three major genepools, of which one genepool (red) appeared in the Oriental wild species, such as ‘Baozhuli’; ‘Huobali’; ‘Jiuzhong’; and ‘Mandingxueli’, suggesting that these accessions might be of a hybrid origin involved with Oriental wild species. Four P. ussuriensis accessions (‘Daxiangshui’, ‘Mangyuanxiang’, ‘Nanguoli’, and ‘Yaguang’) also appeared to have three or four genepools, inferring their hybridized origin. These results confirmed that hybridization extensively existed among cultivars of P. pyrifolia Chinese white pear group, P. pyrifolia and P. ussuriensis.

‘Korlaxiangli’, a famous cultivar of P. sinkiangensis, was clustered with two cultivars of P. communis, ‘Cascade’ and ‘Comice’, in subgroup II in the UPGMA dendrogram, showing the genetic background of P. communis. Our previous study showed that P. sinkiangensis is of hybrid origin and at least P. communis, P. armeniacaefolia, and Chinese white pears or sand pears have been involved (Teng et al. 2001). ‘Cascade’ was reported to be a hybrid between ‘Max Red Bartlett’ and ‘Comice’ (NCGR-Corvallis Pyrus Germplasm, http://www.ars.usda.gov/main//Docs.htm?docid=11372). The results of STRUCTURE also showed that these two cultivars displayed an admixed genepool, which inferred that ‘Comice’ and ‘Cascade’ might be of a hybrid origin involved with Asian pears.

Synonymous cultivars or accessions

In the Oriental group, two P. pyrifolia cultivars, ‘Puguali’ and ‘Yangdanxueli’, showed 100 % similarity in the UPGMA dendrogram and STRUCTURE results. Among all 24 primers used in this study, 15 primers amplified the same fragments in these two cultivars (Supplementary data 2). All of the results suggested that these two cultivars might be synonyms. In the Occidental group, P. elaeagrifolia 768 from Turkey, P. regelii 890 from Russia, and P. cossonii 828 showed 100 % similarity. A similar situation was also observed between P. pyraster 1671 from Romania and P. caucasica 2816 from Armenia. Because only a few primers amplified successfully in these Occidental accessions, we could not determine if they were synonyms. More RBIP makers will be needed to confirm their relationships. P. mamorensis 834 and 835, and P. gharbiana 789 and 794, all from Morocco showed 100 % similarity, and might be treated as the same genotype.

Conclusions

In summary, the present study predicted retrotransposons in Pyrus and provided a new approach to develop RBIP markers. Ten LTR retrotransposon subfamilies in pear were identified and the Ppcr1 retrotransposon was proven to be duplicated many times in the development of cultivated Asian pears. Results inferred from the data of RBIP markers confirmed that Pyrus could be divided into Occidental and Oriental groups, and that extensive hybridization events occurred during the development of Asian pear cultivars. P. xerophila, which is native to Gansu Province, China, was proven to be an interspecies hybrid involving Oriental and Occidental species. Some Occidental pear cultivars or species, such as ‘Cascade’, ‘Comice’, and ‘P. spinosa 1598’, were observed to be interspecific hybrids between Occidental and Oriental species, and some accessions, such as ‘Puguali’ and ‘Yangdanxueli’, were identified as synonyms. New findings in this study will be important to further understand the phylogeny of Pyrus. More RBIP markers will be needed to produce a comprehensive understanding of the complex relationships and evolution in the Pyrus species.