We carried out a field expedition to Myanmar in 2008 and collected four relevant specimens, i.e., broad-leaved long-petioled Potamogeton specimens, including three non-flowering and one flowering ones in Shan state (Table 1). None of the specimens could be morphologically identified as any of three broad-leaved long-petioled Potamogeton species potentially distributed in Myanmar (Potamogeton distinctus, Potamogeton nodosus, and Potamogeton wrightii) due to either the lack of diagnostic floral characters or intermediate vegetative morphology or both. The morphological characters of the unidentified specimens were summarized to facilitate comparison with the three Potamogeton species (Table 1).

To evaluate the utility of selected DNA barcoding markers through performing molecular phylogenetic analyses, comparative materials of Potamogeton distinctus, Potamogeton nodosus, and Potamogeton wrightii were collected in Japan, Mexico, and Thailand (Table 2). As we failed to collect hybrids of Potamogeton distinctus, the nrITS data sets of two Potamogeton hybrids were obtained from GenBank: Potamogeton ×malainoides Miki (Potamogeton distinctus × Potamogeton wrightii) and Potamogeton distinctus × Potamogeton nodosus (Du et al. 2010). Besides, two outgroup species were selected following Lindqvist et al. (2006) and included into the sample set; those were Potamogeton lucens L. and Potamogeton perfoliatus L. Note that four out of the six comparative materials were previously used for molecular phylogenetic analyses (Ito and Tanaka 2013).

The voucher specimens are retained in either of the following herbaria: BKF; MBK; RAF; TI; TNS. Those of Du et al. (2010) are kept in HIB. Sequences were deposited at the DNA Data Bank of Japan (DDBJ) and their accession numbers and voucher information are given in Table 2.

For the newly obtained samples, total genomic DNA was extracted and sequencing of five plastid regions was performed using the procedure outlined by Ito et al. (2010). For the sequencing, previously used accessions were occasionally involved. We selected the following DNA regions that were used in previous molecular studies of Potamogeton as DNA barcoding markers: atpB-rbcL (Ito et al. 2007), rpl20-rps12 (Kaplan and Fehrer 2011), trnT-trnL, trnL, trnL-trnF (Ito and Tanaka 2013, Zhang et al. 2008), and nrITS (e.g., Du et al. 2010, Ito and Tanaka 2013, Kaplan and Fehrer 2011, Les et al. 2009). The atpB-rbcL (seven samples), rpl20-rps12 (nine), trnT-trnL (four), trnL (five), and trnL-trnF (five) regions of chloroplast DNA were amplified and directly sequenced using primers atpB-2F and rbcL-2R (Manen et al. 1994) for atpB-rbcL (779–787 bp), rpl-20 and 5'-rps-12 (Hamilton 1999) for rpl20-rps12 (794 or 813 bp), and Po-trnT2F (Ito and Tanaka 2013) and “b” (Taberlet et al. 1991) for trnT-trnL (807–809 bp), “c” and “d” (Taberlet et al. 1991) for trnL intron (593 bp), and “e” and “f” (Taberlet et al. 1991) for trnL-trnF (403 bp). Note that trnT-trnL was missing from Potamogeton sp. (N. Tanaka & al. 080662).

Sequences of the nrITS were obtained using primers ITS-4 and ITS-5 (Baldwin 1992) under the same conditions used for the phyB amplification in Ito et al. (2010). The total length was 713 bp. On direct sequencing of ten samples, overlapping double peaks were found at the same sites for complementary strands in the electropherograms. These products were cloned using a TOPO TA Cloning kit for Sequencing (Invitrogen, Carlsbad, California, USA). At least 16 clones per sample were chosen and their sequences were determined using the same procedure as that used in the first PCR followed by direct sequencing. For the cloned sequences, nucleotides that were not detected by direct sequencing were regarded as PCR errors.

Sequences of the atpB-rbcL, rpl20-rps12, trnT-trnL, trnL, trnL-trnF, and nrITS regions were manually aligned using the simple indel coding method of Simmons and Ochoterena (2000). Gaps associated with mononucleotide repeat units were removed from consideration in the phylogenetic analysis because of problems related to homology assessment (Kelchner 2000) and because technical artifacts might be responsible for the variation (Clarke 2001). One representative sequence was used for accessions having the identical combined sequence.

Phylogenetic analyses were independently performed for data sets of ptDNA (atpB-rbcL, rpl20-rps12, trnT-trnL, trnL, trnL-trnF) and nrITS, respectively. Phylogenetic inference was performed using maximum parsimony (MP) in PAUP* 4.0b10 (Swofford 2002) and Bayesian inference (BI; Yang and Rannala 1997) in MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003) as described by Ito and Tanaka (2013); the only differences were the best-fit model for BI analysis on ptDNA (F81) and nrITS (HKY). The Bayesian Markov Chain Monte Carlo algorithm was run for 1 million generations for both ptDNA and nrITS data sets. Four incrementally heated chains were used that started from random trees and sampled one out of every 100 generations. The first 25% of the sampled generations (250,000 generations for each data set, respectively) were discarded as burn-in, and the remaining trees were used to calculate a 50% majority-rule consensus tree and to determine posterior probabilities for branches. The data matrices and the MP trees are available from the TreeBASE (S14928).

The length of the combined five ptDNA regions alignment containing ten accessions totaled 3456 bp, of which two characters were parsimony-informative. Based on this data set, one MP tree (tree length = 27 steps; consistency index = 1.0; retention index = 1.0) and a BI 50% consensus tree were obtained. These trees showed congruent phylogenetic relationships and thus only the MP tree is presented here (Fig. 1).

The length of nrITS alignment composed of 20 accessions totaled 645 bp, of which six characters were parsimony-informative. In the phylogenetic analysis of nrITS data set, one MP tree (tree length = 43 steps; consistency index = 1.0; retention index = 1.0) and a BI 50% consensus tree were obtained. These trees showed congruent phylogenetic relationships and thus only the MP tree is presented here (Fig. 1).

In both ptDNA and nrITS trees, the three morphologically closely related species were well differentiated from one another. With Potamogeton lucens and Potamogeton perfoliatus as outgroup, Potamogeton wrightii and the clade of Potamogeton distinctus and Potamogeton nodosus were clustered (63 MP bootstrap (BS), 1.0 BI posterior probability (PP) in ptDNA; 87 MP BS, 1.0 PP in nrITS). Potamogeton nodosus from Mexico and Potamogeton nodosus-related nrITS sequence of Potamogeton distinctus × Potamogeton nodosus HDZY5-7 showed variation, yet the two sequences were clustered each other (62 MP BS, 0.99 BI PP). GenBank accessions of Potamogeton ×malainoides and Potamogeton distinctus × Potamogeton nodosus (Du et al. 2010) have diverged heterogeneous nrITS sequences, and non-hybrid species have homogenous nrITS sequences (Fig. 1).

Of the four broad-leaved long-petioled Potamogeton specimens from Myanmar, two were genetically identical to Potamogeton distinctus from Japan and Thailand (N. Tanaka & al. 080061, N. Tanaka & al. 080657; Figs 2, 3). Another specimen had Potamogeton wrightii haplotype and both of the heterogeneous nrITS sequences of Potamogeton ×malainoides (N. Tanaka & al. 080631; Fig. 4); the other of the remaining two exhibited Potamogeton distinctus haplotype and both of the heterogeneous nrITS sequences of Potamogeton distinctus × Potamogeton nodosus (N. Tanaka & al. 080662; Fig. 5).

The combined five ptDNA regions were separately analyzed to facilitate the utility as individual DNA markers. The comparison included nrITS. Between the closely related species, Potamogeton distinctus and Potamogeton nodosus, where two nucleotide substitutions were observed in nrITS, atpB-rbcL exhibited one nucleotide substitution, while trnT-trnL showed a difference in mononucleotide repeat unit (Tables 3, 4). Among the three species, in which ten nucleotide substitutions were found in nrITS, atpB-rbcL included one length variation (indel) and two nucleotide substitutions; trnT-trnL region had two mononucleotide repeat units, in which repeat numbers are differed.

Applying the comparative samples’ sequence data as DNA barcodes, the broad-leaved long-petioled Potamogeton specimens from Myanmar were genetically assigned. As a result, two out of four specimens were identified as Potamogeton distinctus, a widely distributed species in East Asia, Southeast Asia and the Pacific, including southern part of China and Thailand, but not in Myanmar (Wiegleb 1990). Here we document a new record for the flora of Myanmar.

The taxon-specific DNA barcoding also revealed two hybrids of Potamogeton in Myanmar, and among which was Potamogeton ×malainoides (Potamogeton distinctus × Potamogeton wrightii). This hybrid is known from China (Du et al. 2010), yet a difference is found between the Chinese and Myanmar cases in maternal lineage: Potamogeton ×malainoides from China has Potamogeton distinctus as a maternal parent (Du et al. 2010), but that from Myanmar has Potamogeton wrightii as a maternal parent. This kind of reciprocal hybridizations occasionally occur in Potamogeton, i.e., Potamogeton ×anguillanus, Potamogeton ×fluitans, Potamogeton ×inbaensis, Potamogeton ×lanceolatifolius, Potamogeton ×sudermanicus, and Potamogeton ×suecicus (reviewed in Ito and Tanaka 2013). In terms of morphology, Potamogeton ×malainoides in Myanmar showed both Potamogeton distinctus character, i.e., larger number of leaf veins, and that of Potamogeton wrightii, i.e., the acute to acuminate leaf tip (Table 1), and no major differences are found between the reciprocal hybrids (Du et al. 2010). In other cases of Potamogeton hybrids, reciprocal hybrids are partly distinguishable, e.g., reciprocal Potamogeton ×anguillanus shows no differences in morphology but exhibited differences in drought tolerance (Iida et al. 2007); Potamogeton ×inbaensis with different maternal lines is roughly distinguishable by leaf morphology (Amano et al. 2008).

The other hybrid of Potamogeton identified in Myanmar is Potamogeton distinctus × Potamogeton nodosus. This hybrid is also known from China, yet no maternal lineage was conclusively identified in the previous study (Du et al. 2010). The present study successfully identified Potamogeton distinctus as the maternal lineage of this hybrid for the first time. From the morphological point of view, it is difficult to evaluate the morphological intermediacy between the parental species as both species show large phenotypic plasticity in quantitative morphology, e.g., leaf petiole length.

Du et al. (2011) reported that nrITS is the most useful marker for DNA barcoding of Potamogeton. The present study verified its utility by distinguishing three closely related species, Potamogeton distinctus, Potamogeton nodosus, and Potamogeton wrightii, as well as hybrids among them (Fig. 1, Table 3). Meanwhile, in order to understand hybridization events precisely, we simultaneously used plastid DNA markers, including those used in previous molecular studies, i.e., atpB-rbcL (Ito et al. 2007), rpl20-rps12 (Kaplan and Fehrer 2011), and trnT-trnL, trnL, trnL-trnF (Ito and Tanaka 2013, Zhang et al. 2008). Given that atpB-rbcL showed higher utility than the others (Table 4), here we propose nrITS and atpB-rbcL as DNA barcoding markers for Potamogeton species. Note that trnT-trnL has similar resolution to distinguish closely related Potamogeton species, yet the differences are found only in mononucleotide repear units, which technical artifacts might be responsible for the variation (Clarke 2001).

The taxon-specific DNA barcoding method presented here will be applicable in elucidating further diversity of Potamogeton in other floras. With some modification on marker selection, this method will be also applicable for floras that focus on other taxa.