Background

Bambusoideae is a subfamily of the grass family Poaceae and is further divided into nine subtribes comprising more than 80 bamboo genera and about 1400 species worldwide. Fifty genera and more than 500 species are found in China, among which Phyllostachys pubescens (synonym: P. edulis) is commercially the most important species providing the third largest source of timber and the most predominant source of bamboo shoots. P. pubescens plantations cover an area of 3 million ha (approximately 2% of the total forest area), which has doubled over the last 30 years and taken on a more important ecological role (Fu 2001). Compounds extracted from P. pubescens have recently shown potential for the treatment of obesity and other diseases (Higa et al. 2012). However, various problems associated with P. pubescens plantations including its simultaneous flowering intervals of more than 60 years and recovers from a limited number of clones (Janzen 1976; Watanabe et al. 1982). Additionally, the little knowledge of its basic biology, genetics and breeding system bring about the practical difficulties associated with the identification and characterization of superior genotypes.

Molecular markers developing from microsatellites, also known as simple sequence repeats (SSRs) with characterization of high genome coverage, random dispersion, co-dominant inheritance, reproducibility and amenability to automation in high throughout genotyping, have gained considerable spotlight recently. By now, microsatellite markers have been developed for several other bamboo species, e.g. six loci for Bambusa arundinacea (Nayak and Rout 2005), eight loci for Sasa senanensis (Miyazaki et al. 2009) and eight loci for S. cernua (Kitamura and Kawahara 2009). We identified 19 GenBank microsatellite markers in P. pubescens and related species (Tang et al. 2010), and 15 expressed sequence tag (EST) SSR markers for Bambusa species (Dong et al. 2011). Recently, the Bamboo Full-Length cDNA Project (Peng et al. 2010) has generated a vast amount of publicly-available P. pubescens cDNA sequence data that can be used for gene discovery, comparative genomics/transcriptomics and marker development. Microsatellites derived from cDNAs or ESTs are highly transferable to closely related species (Zhang et al. 2005) facilitating the development of gene-based maps that may increase the efficiency of marker-assisted selection through the use of candidate genes (Rossi et al. 2003; Lu et al. 2006).

Here, we report the use of P. pubescens full-length cDNA (FL-cDNA) sequences to 1) analyze the association between SSRs and transposable elements (TEs) in the transcriptome; 2) develop and validate FL-cDNA SSR markers and determine their transferability to other bamboo species; and 3) apply the polymorphic SSR markers to estimate outcrossing rates in P. pubescens and identify bamboo interspecies hybrids.

Results and discussion

Association between SSRs and TEs in the P. pubescens transcriptome

We analyzed 10,608 P. pubescens FL-cDNA sequences available in NCBI GenBank, representing ~7171 kb of DNA. EST-trimmer was used to remove poly(A/T) runs, and the remaining sequence data were screened using MISA, identifying 2330 SSRs in 2014 cDNAs, the remaining cDNAs lacking SSRs. The sequences were clustered with CAP3, reducing the collection to 1614 non-redundant SSRs in 1382 cDNA contigs (Additional file 1: Figure S1). Peng et al. (2010) described the distribution of SSRs in the P. pubescens transcriptome in detail. Therefore, we selectively analyzed the non-redundant cDNA sequences and contigs with RepeatMasker to determine the association between SSRs and TEs because previous reports have shown that many SSRs are located in TEs (Richard et al. 2008), e.g. 50% of SSRs in the human genome (Scherer 2008), and that SSRs are closely associated with TEs in rice (Akagi et al. 2001; Temnykh et al. 2001) and barley (Wei et al. 2002). The results revealed 95 TEs, representing 13.52 kb (0.27%) of the total cDNA sequence data. Further analysis showed that 29 TEs contained a total of 39 SSRs, accounting for 822 bp (6.41%) of the total TE DNA sequences. In comparison, the non-redundant EST sequence data (7089 cDNAs refined from the original 10,608 sequences) contained 1614 SSRs, accounting for 2.60% of the total cDNA sequences in length. Therefore, SSRs were approximately 2.5-, 65.4-fold more abundant in TEs compared to cDNAs (Table 1) and whole genome (0.098% based on the analysis of genome survey sequences (GSSs; Tang et al. 2010)). It is possible that SSRs within TEs are also involved in the regulation of gene expression (Tomilin 2008).

Table 1 Association between FL-cDNA SSRs and transposable elements (TEs) in P. pubescens
Full size table

Some studies have also suggested associations between specific SSR motifs and particular TE families, e.g. (TA)n is often found in the 5′-UTR of Micron element transposase genes in rice (Akagi et al. 2001; Temnykh et al. 2001). We also investigated the distribution of SSRs among DNA transposons, and found they were most likely to occur in En/Spm elements (33.87% of the total En/Spm DNA sequence). Six SSRs were found in five En/Spm elements, with one element containing two SSRs (Table 2). Mutator elements were the next most likely to contain SSRs (14.19% of the total Mutator DNA sequence). Thirteen SSRs were found in 12 Mutator elements, again with one element containing two SSRs (Table 2). The situation was very different among retrotransposons, with only 0.30% of the total Ty1-copia DNA sequence and 0% of Ty3-gypsy DNA sequence made up of SSRs. More detailed investigation of specific repeat motifs showed that four of the six SSRs found in En/Spm elements were TA/AT repeats, and 10 of the 13 SSRs found in Mutator elements were CT/AG repeats. All 13 of the Mutator SSRs and six of the En/Spm SSRs were located in the 5′-UTR. It has been reported that TE molecular markers (mPing) showed significantly higher levels of polymorphism than all other molecular markers in closely-related rice cultivars (Monden et al. 2009). Considering that it is difficult to detect genetic variation in P. pubescens using ordinary markers (Lin et al. 2009; Tang et al. 2010), SSRs in TEs therefore appear to be promising markers for bamboo species.

Table 2 Distribution of SSRs in En/Spm and Mutator transposons
Full size table

Development and polymorphism assessment of FL-cDNA SSR markers for P. pubescens

Original collection of 10,680 P. pubescens FL-cDNA sequences produced 1382 cDNA contigs containing SSRs. Sequences containing mononucleotide repeat motifs were excluded, leaving 1051 cDNA sequences containing SSRs with 2–6 nt repeats motifs (Additional file 1: Figure S1). Following the procedure already adopted for rice (Temnykh et al. 2001). We were able to design primer pairs for 583 (55%) of these cDNAs, the remainder offering either insufficient flanking DNA (over half of the SSRs were found in the 5′ or 3′ UTRs) or flanking DNA that was unsuitable for primer design. Only 325 (24.1%) of the SSRs were type I repeats (>20 bp), which offer greater potential for marker development. The 100 most promising sequences were selected for PCR validation, including dinucleotide repeats with ≥12 repeat units, trinucleotide repeats with ≥8 repeat units, tetranucleotide repeats with ≥6 repeat units, pentanucleotide and hexanucleotide repeats with ≥5 repeat units and some compound SSRs with >24 repeats (Table 3). We found that 32 of the selected cDNAs were unsuitable because the PCR failed to generate a product (four cDNAs) or generated products lacking SSRs (28 cDNAs), but the remaining 68 sequences allowed the development of FL-cDNA SSR markers (Table 3). These contained 18 compound SSRs, 19 dinucleotide repeats, 18 trinucleotide repeats, four tetranucleotide repeats, three pentanucleotide repeats and six hexanucleotide repeats. Interestingly, although 45 of the cDNAs (66.2%) generated the anticipated PCR product, 16 (23.5%) generated products with more repeats than expected, five (7.4%) generated products with fewer repeats than expected, and two (PBM050 and PBM055) generated products with different repeats and flanking sequences than those anticipated. The unanticipated amplification resulted in three SSR markers (PBM036, PBM055 and PBM 077) containing type II repeats (12–19 bp in length) and one marker (PBM079) shorter than 12 bp. In total, 67 sequences were deposited in GenBank (accession nos GU644371–GU644438). Based on BLASTX analysis, putative functions were assigned to most (66.2%) of the cDNA sequences with significant similarity to known proteins, whereas 27.9% matched unknown/hypothetical proteins and 5.9% were novel sequences (Table 3).

Table 3 Characteristics of the P. pubescens SSR markers derived from FL-cDNAs
Full size table

One hundred and seven primer pairs finally yielded 68 FL-cDNA SSR markers for P. pubescens, which is towards the lower end of the 60–90% success rate previously reported in sugarcane (Cordeiro et al. 2001), barley (Thiel et al. 2003), wheat (Yu et al. 2004) and peanut (Liang et al. 2009). Squirrell et al. (2003) defined the successive loss of sequenced fragments and designed primers, until arriving at a final collection of “working SSRs” producing discrete bands of the expected size, as the “attrition rate”. Kofler et al. (2008) reported a high attrition rate when developing SSR markers from enriched libraries, BAC-end sequences and ESTs in rye, possibly reflecting the large number of TEs in the rye genome. Tero et al. (2006) found that the number of SSR markers was reduced when the markers were predominantly located within TEs. Squirrell et al. (2003) suggested that SSR marker development would be challenging in polypoid species and species such as wheat and rye with large numbers of TEs. P. pubescens has 2n = 48 chromosomes and is thought to be tetraploid (Li et al. 1999). The genome is >2000 Mb, which is approximately 5.4 times larger than diploid cultivated rice (Gui et al. 2007), and it contains a large number of TEs (Zhong et al. 2010; Zhou et al. 2010a, [b], [c]). The slightly higher attrition rate we encountered therefore seems reasonable when considering the chromosomal polyploidy, size and TE content of the genome. We also encountered a higher attrition rate in B. oldhamii (Li et al. 2001), a hexaploid bamboo species with a large genome (data unpublished) in which we developed 15 EST-SSR markers from 52 promising sequences selected from 3406 non-redundant ESTs (Dong et al. 2011).

We surveyed the allelic variability of the markers by genotyping 50 open-pollinated seedlings germinated from the year 2010 seedlot (Table 3). Among the 68 FL-cDNA SSR markers, only 22 (32.4%) showed polymorphism. The polymorphism information content (PIC) values of the 68 markers ranged from 0 to 0.51 with a mean value of 0.12. For the 22 polymorphic loci, the PIC values ranged from 0.19 to 0.51 with a mean value of 0.36, and the top ten markers in terms of polymorphism were PBM075, PBM069, PBM095, PBM046, PBM066, PBM080, PBm087, PBM044, PBM084 and PBM091. SSR polymorphism in P. pubescens is much lower than observed in cereals (Thiel et al. 2003; Yu et al. 2004), coffee (Aggarwal et al. 2007) and the rubber tree (Feng et al. 2009). Bamboo P. pubescens has a long flowering interval of more than 60 years (Janzen 1976; Watanabe et al. 1982). Therefore, open pollination (DNA recombination) appears to have limited the amount of replication slippage, which diversifies SSR alleles (Richards and Sutherland 1994; Jakupiak and Wells 1999). Clonal propagation in the interim periods of flowering has reduced the SSR diversity in bamboo (Nayak and Rout 2005). In a previous study, we discovered almost no allelic variation in the panel of 11 varieties and 17 provenances of P. pubescens using 19 GSS-SSRs (Tang et al. 2010).

Interspecific transferability and polymorphism of P. pubescens FL-cDNA SSR markers

Although more than 1000 bamboo species have been described, the vast majority of publically-available sequence data are derived from P. pubescens (Tang 2009). Therefore, the development of a set of transferable P. pubescens FL-cDNA SSR markers suitable for other bamboo species would help to accelerate genetic research and comparative genomics in the Bambusoideae subfamily. Previously, we developed 19 P. pubescens GSS-SSR markers and successfully transferred them to six other Phyllostachys species with an average transferability of 75.3% and 66.7% polymorphism (Tang et al. 2010). In B. arundinacea, 100% and 83.3% transferability were achieved with 6 SSR markers in eight other Bambusa species and 10 species of other genera, respectively (Nayak and Rout 2005). In B. oldhamii, we achieved an average 59.6% transferability and 51.4% polymorphism with 15 markers in 14 bamboo species including four species within the same genus (Dong et al. 2011). We tested the transferability and polymorphism of these 68 putative FL-cDNA SSR markers across 41 diverse species in six tribes of the Bambusoideae subfamily, as defined by Das et al. (2008) and Yang et al. (2008) (Additional file 2: Table S1 and Additional file 3: Table S2). Successful amplification became less likely with increasing phylogenetic distance from P. pubescens, with an 83.1% success rate within the genus Phyllostachys, a 79.4% success rate across genera within the subtribe Shibataeeae, and a 49.3% average success rate for other subtribes, ranging from 36.8–76.5% (Table 4 and Figure 1). In contrast, the number of markers showing polymorphism increased with phylogenetic distance, with 79.4% of markers showing polymorphism within the genus Phyllostachys, 91.3% showing polymorphism within the Shibataeeae, and 92.8% showing polymorphism when comparing other subtribes. Markers in coding sequences were on average the most transferrable (69.1%) and the least polymorphic (89.4%), compared to those located in 5′-UTRs (63.4% transferrable, 90.7% polymorphic) and 3′-UTRs (61.8% transferrable, 91.4% polymorphic). These trends were exacerbated with increasing phylogenetic distance. These matches the results from a metastudy of 601 loci in 35 plant species showing an average 89.8% transferability at the subgenus level, 76.4% at the genus level and 35.2% at the family level (Rossetto 2001). Interestingly, more than 17 (25%) of the markers were transferrable to more than 85% of the tested species (Additional file 3: Table S2). This success rate suggests that FL-cDNA SSRs and their flanking regions are sufficiently conserved (Zhang et al. 2005), and it is therefore possible to transfer P. pubescens FL-cDNA SSR markers to other bamboo species for evolutionary studies and phylogenetic reconstructions (Sharma et al. 2008).

Table 4 Transferability/polymorphism of P. pubescens FL-cDNA-derived SSR markers across species and genera in the Bambusoideae subfamily
Full size table
Figure 1
figure 1

Polyacrylamide gel electrophoresis bands representing microsatellites derived from FL-cDNA sequences, tested on a panel of selected bamboo species to evaluate transferability and polymorphism in locus of PBM042 (above) and PBM064 (nether). M: size marker. Mb, Cp, etc.: Bamboo species abbreviations are listed in Supplementary Table 1.

Full size image

Using polymorphic FL-cDNA SSR markers to estimate outcrossing rates and identify interspecific bamboo hybrids

Sexual propagation increases genetic diversity by creating progenies of different genotypes through recombination (i.e. outcrossing). This is advantageous for predominantly clonal plants such as most bamboo species, which rely mostly on vegetative regeneration interspersed with occasional flowering (Janzen 1976). The analysis of the reproductive system is therefore fundamental to elucidate primary genetic diversity and the structure of regenerating bamboo populations, and to adopt strategies for genetic improvement. Previous studies on the bamboo reproductive system based on field data and artificial pollination showed that self-compatibility is predominant in Sasa species (Nishiwaki and Konno 1990), and the selfing rate could approach 0.99 in Merostachys riedeliana (Guilherme and Ressel 2001). Outcrossing rate was estimated using SSR-based analysis as reported in S. cernua (Kitamura and Kawahara 2011).

Among the 22 polymorphic SSR markers described above, the ten most polymorphic (PIC ≥ 0.36) were used to detect polymorphisms in 50 open-pollinated half-sib seeds (year 2011) from three flowering sites in the Guangxi Province separated by at least 100 km. Polymorphism in the PBM044, PBM080 and PBM095 loci was identical in the seeds from all three flowering sites, whereas PBM084 and PBM091 featured additional alleles from Lipu, PBM069, PBM075, PBM087 and PBM091 featured additional alleles from Lingchuan, and PBM069, PBM075 and PBM084 featured additional alleles from Guanyang (Table 5). This indicated that flowering culms in different sites featured diverse SSR genotypes and produced genetically-diverse half-sib seed sources. Therefore, we used these eight polymorphic loci to estimate the outcrossing rates and other related genetic parameters for P. pubescens (Table 5). The overall estimates of tm and ts for three culms were 0.089 for both parameters, with no standard deviation. The estimates for individual culms showed small differences of 0.067 in Lipu and Lingchuan, and 0.133 in Guanyan, again for both parameters. Estimation of F is for the overall population was 0.195, indicating homozygote excess. We found that the outcrossing rate was 0.089, estimated from eight polymorphic multilocus datasets in P. pubescens, which is slight lower than the 0.148 reported in S. cernua using six multilocus SSR datasets (Kitamura and Kawahara 2011). This indicated that the reproductive system of P. pubescens predominantly involves self-fertilization with an adequate proportion of crossing to ensure genetic diversity as reported for S. cernua (Kitamura and Kawahara 2011).

Table 5 Seed number, estimated outcrossing rates and relative parameters for each of three P. pubescens flowering culms at 8 loci
Full size table

The grow-out test for bamboo interspecific hybrids is time-consuming and laborious because it involves growing plants to maturity (which takes at least 5 years), assessing several anatomical, morphological and floral (long-term interval) characteristics that distinguish the hybrid. The polymorphic SSR markers could also help in the rapid and accurate identification of interspecies hybrids, as reported in poplar (Rajora and Rahman 2003) and wheat-barley (Malysheva et al. 2003). To obtain proof of principle that our novel SSR markers are suitable for hybrid characterization, we next selected several highly-transferable and polymorphic FL-cDNA SSR markers. PBM032, PMB049, PMB063 and PMB064, each with a number of species-restricted alleles, were used to test uncharacterized bamboo samples. Marker PMB063 identified the parental species in one hybrid as P. kwangsiensis and P. bambusoides, because all sequenced bands contained the (TCCA)n motif although with a variable number of repeats (Figure 2). Similarly, marker PMB064 identified the parental species B. pervariabilis and Dendrocalamus latiflorus which are distantly related to P. pubescens, with a variable number of repeats in the (GAGT)n motif (Figure 3). As previously shown using GSS-SSR markers, such high levels of transferability and polymorphism within the Bambusoideae subfamily should allow the use of FL-cDNA SSR markers to identify interspecific hybrids and their parents, both within the genus Phyllostachys (Tang et al. 2010) and in more distant taxa within subtribe of Shibataeeae (Lu et al. 2009). We have also developed several putative EST-SSR markers in B. oldhamii and have used these to identify some other sympodial bamboo interspecies hybrids (Wu et al. 2009; Dong et al. 2011). The SSR markers developed in the present study were used to identify not only interspecific hybrids from monopodial Phyllostachys but also intergeneric hybrids with sympodial rhizomes, which are distantly related to P. pubescens. Our data confirmed that microsatellites, especially SSR markers based on cDNAs and ESTs, are ideal for the identification of bamboo interspecies hybrids.

Figure 2
figure 2

A, Microsatellite DNA fingerprints of P. kwangsiensis (line 1), P. bambusoides (line 3) and a presumed hybrid (line 2) at locus PBM063. B, Alignment of the nucleotide sequences of the microsatellite alleles at locus PBM014 amplified from P. kwangsiensis, P. bambusoides and two presumed hybrids. Nucleotides conserved among these sequences (relative to P. kwangsiensis) are shown by dots. The lines indicate the primer sequences used to amplify this microsatellite locus. The box highlights the microsatellite. The suffix numbers after bamboo species correspond to the DNA bands marked in part (a).

Full size image
Figure 3
figure 3

A, Microsatellite DNA fingerprints of Bambusa pervariabilis , Dendrocalamus latiflorus and their presumed hybrids at locus PBM064. B, Alignment of the nucleotide sequences of the microsatellite alleles at locus PBM064 amplified from B. pervariabilis, D. latiflorus and their presumed hybrids. Nucleotides conserved among these sequences (relative to B. pervariabilis) are shown by dots. The lines indicate the primer sequences used to amplify this microsatellite locus. The box highlights the microsatellite. The suffix numbers after the bamboo species correspond to the DNA bands marked in part (A).

Full size image

Conclusions

Our data provide insight into the association between SSRs and TEs in FL-cDNAs from the P. pubescens transcriptome, allowing us to develop and evaluate 68 FL-cDNA SSR markers that can be used in P. pubescens and partially for many other bamboo species, to estimate the reproductive system of P. pubescens and identify several interspecific hybrids. These FL-cDNA SSR markers enrich the molecular marker resources currently available for bamboo. When a large set of polymorphic markers becomes available, we can use genome-wide association mapping in bamboo, in the absence of structured populations, to identify markers for traits of interest that can be used for marker-assisted selection in the Bambusoideae subfamily.

Methods

Full-length cDNA mining and SSR/TE detection

We obtained 10,608 FL-cDNA sequences from NCBI Entrez (http://www.ncbi.nlm.nih.gov/) on July 1, 2010. These cDNA sequences were assembled from five cDNA libraries constructed from breaking-out shoots, young (40-cm) shoots and young leaves from plants, and shoots and roots from germinated seeds (Peng et al. 2010). We used EST Trimmer (http://pgrc.ipk-gatersleben.de/misa/download/est_trimmer.pl) to remove poly(A/T) runs from the 5′ and 3′ ends until there were no occurrences of (T)5 or (A)5 within a 50-bp range. Redundant sequences were eliminated and overlapping sequences were spliced together using CAP3 (http://seq.cs.iastate.edu/cap3.html) (Huang and Madan 1999).

After pre-treatment, we used MISA (http://pgrc.ipk-gatersleben.de/misa/misa.html) to screen for SSRs including mononucleotide repeats ≥10 bp in length, dinucleotide to hexanucleotide repeats with ≥6 repeat units, and interrupted composite SSRs with ≤100 bp of intervening DNA. Putative annotations were assigned to non-redundant ESTs containing SSRs using BLAST against the Moso Bamboo cDNA Database (http://202.127.18.228/mbcd/) and the Gramene Ontologies Database (http://archive.gramene.org/plant_ontology/). TEs were identified using RepeatMasker and RepeatProteinMask (http://www.repeatmasker.org) based on similar elements present in the rice genome, and SSRs within TEs were screened using MISA with the same parameters as above. Additional file 1: Figure S1 provides a flow chart for the data mining and marker development process.

Plant material and DNA extraction

We used P. pubescens samples collected from the Anji Bamboo Germplasm Garden, Anji, Zhejiang Province, to identify and characterize putative FL-cDNA SSR markers. The polymorphism of these SSR markers was evaluated using 50 seedlings germinated from an open-pollinated seedlots (mixed seed sources, mainly from different flowering sites in the counties of Lipu, Lingchuan and Guanyang, Guangxi Province in the year 2010). Another 50 seedlings were germinated from open-pollinated half-sib seeds (year 2011) from three flowering culms in the same three counties (>100 km between sites) and were used to estimate the P. pubescens outcrossing rate. We obtained 41 representative bamboo species from 38 genera within six subtribes mainly found in China to test the transferability and polymorphism of the FL-cDNA SSR markers (Additional file 2: Table S1). We obtained three Phyllostachys interspecific hybrids from Jiangxi Province, China, and two intergeneric hybrids from Yoshinaka Bamboo Germplasm Garden, Fukuoka, Japan, for the hybrid identification tests. Genomic DNA was extracted from young leaves using the hexadecyltrimethylammonium bromide (CTAB) method (Doyle and Doyle 1987), with some modifications.

Amplification and sequencing of SSR loci

Primer pairs designed according to the available cDNA sequences were synthesized by Shanghai Sangon Biological Engineering Technology & Services Co., Ltd. P. pubescens DNA was amplified in 20-μl reactions comprising 50–100 ng of template DNA, 0.2 μM of each primer, 200 μM of each dNTP and 1 unit of Taq DNA polymerase with 1× PCR universal buffer (10 μM Tris–HCl, pH 8.3 at 25°C; 50 μM KCl), and 1.5 μM MgCl2 (Shanghai Sangon Biological Engineering Technology & Services Co., Ltd). The reaction was heated to 95°C for 5 min using an ABI PE9700 thermocycler, followed by 30 cycles of 1 min denaturation at 95°C, 1 min annealing at 46–59°C depending on the primer pair (Table 3), and 2 min extension at 72°C, followed by a final hold at 72°C for 5 min. Amplified microsatellite loci were tested in 41 diverse species in six tribes of the Bambusoideae subfamily (Table 4) and interspecific hybrids (Figures 2 and 3). The annealing temperature was lowered by 2–5°C according to the evolutionary distance between species based on molecular markers (Das et al. 2008) and nuclear and chloroplast sequences (Yang et al. 2008), as suggested by Rossetto (2001). PCR products were separated on 6% polyacrylamide denaturing gels, and marker bands were revealed by silver staining as described by Panaud et al. (1996). Specific bands were excised directly from the silver staining polyacrylamide gel, purified using the EZ-10 Spin Column DNA Gel Extraction Kit (Biobasic Inc.) and ligated into the pUC18 vector (TaKaRa, Japan). Three positive clones for each bamboo species were selected for sequencing using BigDye terminator V3.1 in a cycle sequencing protocol according to the manufacturer’s specifications (PE Applied Biosystems, ABI PRISM 3100-Avant Automatic DNA Sequencer). Vector sequences were removed then edited using Vector NTI software (version 10.0, Invitrogen Co., USA). Sequences were deposited in NCBI GenBank (accession nos GU644371–GU644438).

Data analysis

The polymorphism information content (PIC) (Botstein et al. 1980) of our SSR markers was determined using Powermarker v3.25 (Liu and Muse 2005). All 68 selected primer pairs were used to amplify template DNA from 41 bamboo species covering 35 genera in six subtribes (Additional file 2: Table S1) and the statistical methods of Nayak and Rout (2005) and Sharma et al. (2009) were used to calculate the cross-taxon transferability and polymorphism (Additional file 3: Table S2), in which polymorphism is calculated only from the loci that were successfully transferred across taxa (Rossetto 2001). Single locus and multilocus outcrossing rates and relative parameters were analyzed separately under the mixed mating model of Ritland & Jain (1981) and Ritland (2002), implemented using MLTR v3.4 (Ritland 1996).