Specimens for morphological observations and silica gel-dried leaf samples of “Ampelocalamus menglaensis” and Melocalamus yunnanensis were collected in August 2012 and April 2013 in southern Yunnan (Mengla, Xishuangbanna) and of “teng zhu” in October 2013 and August 2014 in southeastern Yunnan. The distribution map was created using DIVA-GIS (http://www.diva-gis.org).

The specimens collected in southern Yunnan were compared with specimens of Ampelocalamus actinotrichus from Hainan Island. The specimens of “teng zhu” from southeastern Yunnan were compared with those of Neomicrocalamus prainii from northwestern Yunnan and southern Tibet.

The inflorescence of Melocalamus yunnanensis from the Bamboo Garden of Xishuangbanna Tropical Botanical Garden was compared with specimens of this species from Jiangcheng, southern Yunnan. The glumes, lemma, palea, ovary, style, and stamens were observed under a hand-lens (30×).

The phylogeny of Arundinarieae based on LEAFY indicated that the genus Ampelocalamus is monophyletic, except for A. calcareus C. D. Chu & C. S. Chao, and sister to Drepanostachyum P. C. Keng and Himalayacalamus P. C. Keng (Yang et al. 2013). Thus, samples of the latter two genera were used as outgroup in our study. We aimed to elucidate the identity of “A. menglaensis”, not to infer phylogenetic relationships. Therefore, only those taxa closely related to A. actinotrichus were included (Table 1).

In order to confirm the identity of “teng zhu”, representatives of the subtribe Bambusinae were chosen according to Yang et al. (2008), including taxa of Bambusa Schreber, Bonia Balansa, Dendrocalamus Nees, and Melocalamus Bentham (Table 1).

Voucher specimens of all samples are deposited at the herbarium of the Kunming Institute of Botany, Chinese Academy of Sciences (KUN).

Total genomic DNA was extracted from silica gel-dried leaves using a modified CTAB procedure (Doyle and Doyle 1987). Primers and protocols for the PCR amplification of the nuclear gene LEAFY and four plastid regions (rbcL–psaI, rpl32–trnL, trnG–trnT(g), trnG–trnT(t)) followed prior studies (Yang et al. 2013, Zhang et al. 2013). PCR products were checked on 1% agarose gels, and purified using ExoSAP–IT (USB, Cleveland, OH, USA). Double–stranded and purified PCR products were sequenced by the dideoxy chain termination method with ABI PRISM Bigdye Terminator Cycle Sequencing Ready Reaction Kit with AmpliTaq DNA polymerase FS (Perkin Elmer, Waltham, MA, USA). PCR reactions and programs were chosen according to the recommendations of the handbook, with slight modifications in some cases. Bidirectional sequencing was performed on an ABI 3730xl automated sequencer.

Sequences were assembled and edited with SeqMan (DNA STAR package; DNA Star Inc., Madison, WI, USA), aligned by MUSCLE (Edgar 2004), and adjusted manually where necessary. Informative indels introduced by the sequence alignment were coded as binary characters using the simple indel coding method of Simmons and Ochoterena (2000). Due to no obvious conflicts among individual plastid trees, we used a combined data set of four plastid regions for phylogenetic analyses. All data matrices are available on request from the first author. Sequences newly obtained in this study have been deposited in GenBank (Table 1). Accession numbers initialed with ‘KM’ and ‘KF’ were downloaded directly from GenBank, and the others were obtained in this study.

For phylogeny reconstructions, we used three methods, namely maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference (BI). The MP and ML analyses were conducted with PAUP* version 4.0b10 (Swofford 2002). For the ML analyses, the best-fitting models were selected using jModeltest v2.1.4 under the Akaike Information Criterion (AIC) (Darriba et al. 2012). The TIM3 model and TPM1uf+G model were selected for LEAFY and combined plastid regions, respectively. MP and ML analyses were conducted with the same parameter setting for the heuristic search and the bootstrap calculation. The heuristic search was performed with 1000 random addition sequence replicates and TBR branch swapping, MULTREES option in effect. Strict consensus trees were calculated for all MP analyses. Branch support was estimated with 1000 bootstrap replicates (Felsenstein 1985) using the heuristic search method as described above (with 100 random addition sequence replicates). The same models were used for Bayesian analyses with MrBayes version 3.2.5 (Ronquist et al. 2012). Two independent runs were conducted simultaneously starting with random trees, and each run consisted of one cold and three hot chains. Chains were run for 150000 generations for LEAFY and 200000 generations for the combined plastid data set, and trees sampled every 100 generations. The average standard deviation of split frequencies between both runs reached a value below 0.01. The convergence of the chains and the number of trees to be discarded were determined using Tracer version 1.6 (http://tree.bio.ed.ac.uk/software/tracer). The initial 25% trees were discarded as burn-in, and the remaining trees were used to construct a 50% majority-rule consensus trees. MP and ML bootstrap values ≥ 50% and BI posterior probabilities ≥ 0.95 were labeled on the tree branches.

The aligned length of LEAFY was 738 bp, and six indels were coded as additional absent/present (0/1) characters, giving a total of 744 characters in the MP matrix, of which 26 were parsimony-informative. Sequences of the four plastid regions were obtained for all samples, except trnG-trnT(t) for the sample ZXZ11027. The combined plastid matrix was 3737 bp long and included 14 indel characters in the MP matrix, 113 of which were parsimony-informative. There were inversions in the rbcL–psaI and rpl32–trnL sequences (Zeng et al. 2010, Zhang et al. 2013) and gaps were introduced to separate the inverted regions to avoid overweighting the inversions. Those gaps were not scored and treated as missing data.

The 50% majority-rule consensus tree from BI for LEAFY is presented in Fig. 1. The ML tree and the MP strict consensus tree were consistent with the Bayesian tree except for a few branches with very low support (not shown). The statistical support was shown along the branches (MP/ML/BI). All individuals of Ampelocalamus formed a clade with high support (MP/ML/BI = 98/98/1.00). The samples of “A. menglaensis” (voucher numbers 12167, 13007, and 13012) fell with the Ampelocalamus actinotrichus samples collected from Hainan.

Figure 1. 

Phylogram of the 50% majority-rule consensus tree from Bayesian analysis of LEAFY sequences. Bootstrap values over 50% and Bayesian posterior probabilities over 0.95 are shown along branches (MP/ML/BI). Individuals of Ampelocalamus actinotrichus from Yunnan are in bold and indicated by solid triangles. Letters A and B after taxon names denote different alleles of LEAFY.

The 50% majority-rule consensus tree from BI for the combined plastid data set is presented in Fig. 2. The ML tree and the MP strict consensus tree were the same with the Bayesian tree. The statistical support is shown along the branches (MP/ML/BI). Individuals of “teng zhu” collected from southeastern Yunnan and Neomicrocalamus prainii ZXZ11027 from northwestern Yunnan fell in one clade with high support (MP/ML/BI = 89/89/1.00), and they subsequently formed a monophyletic group with N. prainii from Tibet (MP/ML/BI = 100/100/1.00).

Figure 2. 

Phylogram of the 50% majority-rule consensus tree from Bayesian analysis of the combined plastid data set (rbcL–psaI, rpl32–trnL, trnG–trnT(g), and trnG–trnT(t)). Bootstrap values over 50% and Bayesian posterior probabilities over 0.95 are shown along branches (MP/ML/BI). Individuals of Neomicrocalamus prainii from southeastern Yunnan are in bold and indicated by solid triangles.

Our collections (specimens 12167, 13007, and 13012) were identified as Ampelocalamus menglaensisnom. nud. by bamboo taxonomists. This species was initially published without proper description and designation of the type specimen, and it was reported that this bamboo was originally distributed in Mengla, Yunnan, China (Du et al. 1999). In the Flora Yunnanica (Sun et al. 2003), it was suggested that this species could represent a misidentification of Dinochloa puberula McClure and was not included in the genus Ampelocalamus. Later, Wang (2004) described this species briefly in Chinese with two colored photos in his book Ornamental Bamboos. Based on the brief description and our observations in the field, we inferred that this species should be Ampelocalamus actinotrichus. Major morphological diagnostic characters included: culms pendulous or scrambling; rhizomes pachymorph; internodes with sparsely brown setae initially; nodal sheath scars prominent; branches two to several, occasionally with one dominant branch replacing the main culm and scrambling to other plants; culm leaves tardily deciduous, much shorter than internodes, sheaths with sparse setae abaxially, auricles ovate, oral setae radiate and initially purple, margins of ligules with long and purple hairs, blades lanceolate; leaf sheaths glabrous or with sparse setae abaxially, auricles kidney-shaped, oral setae radiate and initially purple, ligules fimbriate, blades with long pubescence on both epidermises (Fig. 4).

Figure 4. 

Ampelocalamus actinotrichus from Hainan and Yunnan, China. A–C Individuals from Hainan A Culm leaf (MPF10001) B Branches and young culm (MPF10001) C Leaves (MPF10003) D–H Individuals from Yunnan D Culms climbing on trees (13012) E Culm leaf (13001, collected at the same locality with 12167) F Young culm (13001) G Leaves (13001) H Branches (13001).

Yang (2005) analyzed the anatomy of leaves and roots of 13 species of Drepanostachyum P.C. Keng and Ampelocalamus, including A. actinotrichus and A. menglaensisnom. nud.Ampelocalamus actinotrichus and A. menglaensisnom. nud had similar features but distinct from other species, such as two layers of mesophyll cells in leaf transverse sections, large fusoid cells, fewer prickles on abaxial and adaxial epidermises, microhairs developed. Moreover, our molecular phylogenetic analyses demonstrated that our collections from Yunnan grouped together with Ampelocalamus actinotrichus from Hainan, and there was a little genetic divergence between these two biogeographic entities (Fig. 1). On grounds of the aforementioned evidence we can confirm that Ampelocalamus menglaensisnom. nud. is conspecific with A. actinotrichus.

Ampelocalamus actinotrichus was recorded to occur only on the island of Hainan before our confirmation of its occurence in southern Yunnan (Fig. 5). During our field investigations in Mengla, Xishuangbanna, southern Yunnan we went to the border of China and Laos (collecting voucher specimens 13012 and 13013; 13013 was not used here for the molecular analyses). We were told by the local people that this bamboo could also be found in northern Laos. Chen and Wen (2008) reported that this bamboo is distributed in tropical montane rainforests of Mengsong, southern Yunnan, which is adjacent to Burma. It suggests that Ampelocalamus actinotrichus may be distributed in Burma. Additional field explorations are needed to clarify whether this species is more widely distributed on the Indo-China peninsula. The new distribution record of this species has a great significance in resolving the origin and divergence of the genus Ampelocalamus.

Figure 5. 

Distribution of Ampelocalamus actinotrichus and Neomicrocalamus prainii in China (based on specimens cited in this paper). Solid triangles: N. prainii; solid circles: A. actinotrichus; the solid line: the ecogeographical diagonal line from northwestern Yunnan to southeastern Yunnan (Li 1994).

Mengla and Mengsong are both parts of Xishuangbanna, in southern Yunnan. The flora of Xishuangbanna is part of the Indo-Malesian flora, and has a close affinity with floras of adjacent areas (i.e., southern China including tropical Guangxi and Hainan, Burma, Laos, Thailand, and Vietnam) (Zhu 1994, Zhu et al. 2001, Zhu and Roos 2004). The distribution of Ampelocalamus actinotrichus in both Hainan and southern Yunnan (and likely in Burma and Laos) provides a case at the species level for confirming the affinity of floras across southern China including Xishuangbanna (and likely adjacent areas).

From the point of view of morphology, individuals of “teng zhu” (specimens 13045, YXY150, YXY151) from southeastern Yunnan share many features of Neomicrocalamus prainii, including culms scrambling, nearly solid, branches many with the dominant branch equal in size to the culm and other small branches seldom branching again, culm leaf sheaths with purple–brown spots and white pubescence abaxially, culm leaf blades acicular, and others (Fig. 6). The habitats of “teng zhu” in southeastern Yunnan are mainly rocky mountains which are similar to those we found in Gongshan, northwestern Yunnan (voucher No. ZXZ11027). The molecular phylogenetic analyses illustrated that those individuals fell with Neomicrocalamus prainii from southern Tibet and northwestern Yunnan, especially close to the individual from northwestern Yunnan (voucher No. ZXZ11027) (Fig. 2). Based on molecular and morphological evidence, we concluded that “teng zhu” should be identified as Neomicrocalamus prainii.

Figure 6. 

Neomicrocalamus prainii (YXY150). A Habitat B Branches C–D Culm leaves.

Li (1994) inferred that an ecogeographical diagonal line from northwestern Yunnan to southeastern Yunnan was caused by a northward and clockwise rotation of the Shan–Malay Plate (Burma–Malaya Geoblock) since the Miocene, and that northwestern and southeastern Yunnan were once at the same latitude both with a tropical environment. One of the biological effects of this plate movement was that some species were discontinuously distributed at both ends of the diagonal and some were concentrated on the southwestern side (Li et al. 1999, Zhu and Yan 2002). The disjunct distribution of Neomicrocalamus prainii in northwestern and southeastern Yunnan are concordant with this biogeographical line (Fig. 5). Neomicrocalamus prainii is distributed at both ends of the diagonal. The distribution pattern of N. prainii may imply that this species has a tropical origin, and it provides another example for verifying the reality of the ecogeographical diagonal line in Yunnan.

This species was initially described as Racemobambos yunnanensis based on incomplete, poor specimens (Wen 1986). However, it is easily identified by observing the peculiar culm leaves. It was transferred to the genus Neomicrocalamus (Ohrnberger 1999), and subsequently subsumed into Melocalamus (Yi et al. 2007). Molecular studies also showed that it was not a member of Racemobambos or Neomicrocalamus and had a close relationship with Melocalamus (Yang et al. 2008). Main inflorescence features of Melocalamus include pseudospikelets 2-flowered, glumes 2, palea 2-keeled, lodicules 3, stamens 6, stigmas 2 or 3, plumose. Melocalamus yunnanensis as a member of the genus Melocalamus was confirmed by inflorescence characters reported in the current paper.