Bamboo IdentificationHome

powered by FreeFind

Download for reading PDF documents

More depth on classification

Plant classifications started by defining groups on the grounds of possession of shared highly visible morphological characteristics, such as number of stamens. Ever since it became suspected that plants may have evolved similar characteristics independently at different times, there has been an ambition to make classification systems more natural, reflecting real relationships between plants. Sometimes, however, this has become such a holy grail that the original intention of classifying plants in order that they can be named and recognized has been rather forgotten. The relative importance of phenetics: plant characteristics such as morphology, as opposed to phylogenetics: plant relationships as expressed in genes, is still hotly disputed. Nevertheless, analysis of similarities in molecules such as DNA has helped enormously in interpreting the relationships and evolution of most plants, including bamboos, and trees that summarised the early results of temperate bamboo DNA analysis in Europe and DNA analysis in China are presented here. Since then other analyses have produced phylogenies, and a small forest of trees that portray them, slowly obtaining better statistical support for groupings that make sense from a morphological perspective.

Natural classifications aim to group plants on characteristics that have been
panda thumb
inherited from common ancestors mainly so that there will be many other shared characteristics, perhaps not all so immediately obvious or important. This likeliness to share other characteristics as well is of ‘predictive’ value. If one species has chemical constituents that give it some beneficial pharmacological properties for example, then species that are closely related in a natural classification are likely to have similar properties. This could mean that closely related bamboo species will have quite similar site requirements, or uses, or maybe, like species of Himalayacalamus as opposed to species of Drepanostachyum, and those of Chimonocalamus as opposed to Chimonobambusa, they will all just taste good. Looking for a natural classification in the bamboos is particularly appropriate as Giant Pandas, through their novel thumbs, have become a symbol of evolution and the Darwinian theory of evolution through natural selection of the fittest.

Bamboo classifications have traditionally relied on phenetics, and as bamboos are grasses, a family in which vegetative structures are not generally well developed, they have emphasized inflorescence structures in the past. More recently, and especially in China, conflicting systems of classification have developed in the bamboos, according to whether inflorescence or vegetative characteristics are considered more important. These systems are being tested by molecular analysis, which hopes to demonstrate how bamboos have evolved, primary by DNA sequencing.

Molecular analysis has often been able to show where substantial proportions of older systems of classification were decidedly artificial, which means that they combined unrelated plants on the basis of shared morphological characteristics, which had evolved independently. For example, the Himalayan genus Drepanostachyum and the S American genus Chusquea were quite recently combined in a tribe called Chusqueeae on the  grounds that they both have very many branches. Similarly, the S American genus Otatea was included along with the Asian genera Chimonocalamus, Yushania and Drepanostachyum in a large genus called Sinarundinaria, on the basis of their open inflorescences. Molecular evidence has suggested that in both these cases the genera are not really closely related at all, and the similarities are the result of parallel evolution.

Taking as an example a group of temperate bamboo genera, those from the Himalayas, it can be seen that they have been grouped (or ‘lumped’) in many different ways by different taxonomists. This is portrayed in fig. 2, which lists those genera with open inflorescences on the left, and the 3 genera with compressed inflorescences on the right. Red loops enclose genera that have been considered to represent a single genus by different authorities over the decades. All the bamboos in these genera were included in just one genus, Arundinaria at first. As many as 11 separate genera have since been recognized (Wang 1997), but most authorities have lumped at least some of these together. Sometimes they have been lumped on the basis of similarities in their inflorescences, essentially whether they are open or compressed (vertical loops). At other times they have been lumped on the basis of other, vegetative characters, despite the differences in their inflorescences (horizontal loops). With as many different treatments as this, it is easy to see how the naming and identification of these bamboos became a nightmare. Fortunately, modern techniques of molecular analysis, such as DNA sequencing, have helped us to see which of these groupings were justified, and which were not.

Trees of most likely relationships between our current plants derived from statistical data such as molecular sequences are called cladograms, and the branches or ‘clades’ of closely related plants should be close together. Several investigations have demonstrated that the S American, or ‘neotropical’ bamboos are on very different branches to those from Asia. Evidence from patterns of plant dispersal and distribution can often be used to test molecular evidence when it suggests parallel or convergent evolution. Given that bamboos seed so infrequently and have no effective dispersal mechanisms, the inclusion of bamboos from Asia and S America together is highly suspect, reinforcing our confidence in molecular evidence that suggests that similar characteristics have evolved independently on the two continents.

While molecular data can highlight such mistakes as these, what the results have not always been able to establish to date, in other plant groups as well as the bamboos, is exactly how the resultant trees of most likely relationships, or cladograms, should be carved up to give taxa at different ranks. The true relationships between the plants do not always conform to our concepts of species, genera, or subtribes etc. Considerable interpretation is often required, in terms of where boundaries should lie, in order to retain the practical functionality of the binomial system of genera and species established by Swedish botanist Linnaeus in the 18th Century. The old question of whether a long line of continuous variation should be broken up into separate taxa, or combined into one very large heterogeneous taxon has not been solved. Neither phenetics nor phylogenetics alone can satisfy all objectives, combining complete objectivity with perfect utility.

The modern approach is to start by looking for an underlying molecular phylogeny, a likely network of evolutionary relationships. Then, hopefully, it can be broken down into groups according to major branches and gaps in evolutionary pathways, which ideally will be supported by morphological characters. Concerted evolution should over generations emphasize differences in both genes and morphology, by gene recombination and also the extinction of a proportion of offspring. This would allow a statistical analysis of the genes and morphology to highlight well-defined groupings. In many bamboos however, a large number of closely related and poorly separated groupings is emerging as the underlying theme. There is often little apparent means of separation, nor clear grounds for merging them either. Possible explanations for this include the infrequent flowering of bamboos, and relatively recent diversification in some groups, but reticulate evolution through cross-breeding is another likely factor.

One such DNA-based tree of likely but still hypothetical relationships for many of the temperate bamboo genera in cultivation is given in fig. 3. This was produced during PhD research by Grainné Ní Chonghaile in Dublin and Kew in 1999-2001, supervised by Dr Trevor Hodkinson & myself. Sequencing of either chloroplast or nuclear genes on their own did not provide enough information for the temperate bamboos, but when the results of the two were combined a useful phylogeny has started to appear. While it suggests plausible relationships from our knowledge of the morphology and distribution of the species, the statistical support for these precise relationships is low, and there are very many other, fairly similar trees that were equally likely.

Interpretation of all available data to provide the best guidance for classification of Asian woody bamboos has been undertaken, and a paper (Stapleton et al. 2009) has been written to justify the bamboo classification system used in the Flora of China. In the paper it is suggested that the tree in fig. 3 is quite sufficient to demonstrate that molecular data suggests there is no support at all for practically all the large tribes, subtribes or genera based on inflorescence form that were widely followed until very recently. It would appear that such large groups are all extremely unlikely to represent natural relationships. Supertribe Bambusatae Keng f. (1992) is not acceptable, as it combined genera Hibanobambusa, Shibataea, Phyllostachys, Chimonobambusa, and Schizostachyum. Tribe Chusqueeae as applied by Keng & Wang (1996) included Drepanostachyum and Chusquea. Subtribes Shibataeinae (Nakai) Soderstrom & Ellis (1988), and Phyllostachydinae Keng f. (1992) are not acceptable as they combined genera such as Phyllostachys, Shibataea, and Chimonobambusa.

Taxonomists have often disagreed about how large genera should be. Some like big genera and call the others ‘splitters’, some prefer smaller ones and call the others ‘lumpers’. Thamnocalamus as interpreted very broadly by Clayton & Renvoize (1986; 1999), Chao & Renvoize (1989), and Li (1997) and defined by compressed inflorescences, is not acceptable as it also included the unrelated genera Fargesia and Himalayacalamus. Similarly Sinarundinaria as interpreted broadly by Clayton & Renvoize (1986; 1999), Chao & Renvoize (1989), and Li (1997) with the criterion of loose, open inflorescences is not acceptable as it included unrelated bamboos from Ampelocalamus, Drepanostachyum, and Yushania. Moreover there is a fundamental flaw with Sinarundinaria in that its type species is Fargesia nitida, a species that has compressed rather than open inflorescences.

Similarly, the inclusion of Neomicrocalamus in Racemobambos (Clayton & Renvoize 1986, 1999; Chao & Renvoize 1989; Li 1997; Keng & Wang 1986), on the grounds of similarities in inflorescence, seemed highly unlikely to be a sound decision, given their long branch lengths. The advice given by Kew taxonomists to its discoverer, the forester George Gamble, to include it in Arundinaria in 1896, despite its 6 stamens, was not very helpful, and its inclusion in Thamnocalamus (Camus, 1913), was equally inappropriate.

The arrangement of the genera Ampelocalamus, Thamnocalamus, Gaoligongshania, Drepanostachyum, and Fargesia in the Dublin studies is roughly in agreement with the topologies produced later from sequencing of chloroplast and nuclear genes in the US and in Kunming Institute of Botany, China. From the very first Chinese results a group of species with very compressed spathed inflorescences including Fargesia nitida has been well supported, a result that confirms the synonymy of Sinarundinaria within Fargesia.

The usefulness of a variety of molecular techniques in some of the bamboos has been demonstrated, for example in Phyllostachys, (Hodkinson et al. 2000). The relationships between closely related genera and species of bamboo species, however, is often extremely difficult to resolve in a statistically well-supported manner from such studies. From tentative trees such as those produced in Dublin and China, the assistance of morphological characters is still required to produce a workable classification system. The reasons for the difficulty in finding molecular evidence of past evolution and present relationships in the bamboos include rapid recent diversification, long generation times and repeated hybridisation. Despite these difficulties, the adoption of ever-growing amounts of molecular data for larger and larger numbers of bamboos is slowly revealing more and more detail in the relationships. This parallels the need to include a broader range of morphological characters to improve the phenetic classification last century.

It is encouraging to see that the overall and detailed structure of temperate bamboo phylogeny revealed progressively as our phylogenetic knowledge improves is becoming rather consistent with the combined phylogenetic and phenetic system of genera, as applied since the 1990s in the UK & US. The phylogenies based on more gene region sequences and species (Triplett & Clark 2010, Zeng et al. 2010) and RAD sequencing (Wang et al. 2017, Ye et al. 2019) are essentially consistent with our combined analysis tree from 2001 selected for congruence with our morphologically-based classication.

What is clear from molecular information is that the use of floral characters to classify the temperate woody bamboos throughout most of the 20th Century, as championed by grass taxonomists, in contrast to the approach taken by bamboo taxonomists of looking at vegetative characters was neither natural nor helpful, even though this approach is the normal taxonomic procedure in other grasses. Vegetative characters such as rhizome form, buds, and branching would appear to be much better indicators for natural groupings of temperate bamboos.

Below the generic level, distinctions between species and lower ranks cannot be resolved adequately well using the gene sequencing techniques applied at higher ranks. ‘Fingerprinting techniques’ using overall similarities in genes, are of more value at this level, but such studies are not always given much credence, and are not really meant to lead to construction of any hypothetical phylogeny.

Consequently the majority of taxonomic studies down to species level still need to rely on traditional techniques of separating taxa on the basis of differences in morphology, combined where necessary with differences in geographical distribution, in what is known as a morpho-geographical approach. Much more work is still required at this level on the bamboos, undertaking the unglamorous and unfashionable kinds of taxonomy that no-one seems to want to fund any more, and which are unlikely to lead to career advancement, or even any kind of job at all.

This unfortunately still makes any classification system adopted somewhat subjective, and consequently liable to be challenged. There are those who feel that no genera should be recognized unless there is statistical molecular evidence of a clear evolutionary separation (often known as strong bootstrap support for a monophyletic clade). There are others who would recognize more genera, even when there is no molecular evidence to support them, but once the scientific objectivity is lost, the problem again arises of which genera to select, and which parts of the bamboo plants are most important and should be used as criteria for recognition.

The classification system followed on this site uses an underlying structure suggested by DNA analysis, with genera recognized and arranged according to the best currently evidence from DNA, backed up by morphology, the structures and organs of the bamboo plants. For closely related groups of bamboos, and at the species level, where molecular evidence has so far been of little help, a combination of morphological characters along with geographical distribution is applied (morpho-geographical approach). In this way it is a combination of a natural system based on evolutionary pathways (phylogeny), backed up by the practical application of morphological characters (phenetics), not out of choice but out of necessity, because at the end of the day the plants still need to named, described, and recognized. For a detailed explanation of this system see full review (Stapleton et al. 2009) for the Flora of China account. In a nutshell, the philosophy is that evolution is followed where there is evidence of meaningful groupings to follow, practicality and morphology taking over when there is not.

This sort of hybrid approach is now applied in the recognition of genera across the whole grass family (Kellogg 2015).


Further Information

Kellogg, E.A. (2015). In K. Kubtizki, ed., The Families and Genera of Vascular Plants. Volume 13. Flowering Plants: Monocots - Poaceae. Springer  Google Books (with useful ‘Search inside’ facility)


Triplett, J.K., Clark, L.G., (2010). Phylogeny of the temperate bamboos (Poaceae: Bambusoideae: Bambuseae) with an emphasis on Arundinaria and allies. Syst. Bot. 35: 102–120.


Wang, X., Ye, X., Zhao, L., Li, D., Guo, Z., & Zhuang, H. (2017). Genome-wide RAD sequencing data provide unprecedented resolution of the phylogeny of temperate bamboos (Poaceae: Bambusoideae). Scientific Reports, 7(1): 11546. doi: 10.1038/s41598-017-11367-x


Ye, Xia-Ying, Ma, Peng-Fei, Yang, Guo-Qian, Guo, Cen, Zhang, Yu-Xiao, Chen, Yun-Mei, Guo, Zhen-Hua and Li, De-Zhu (2019). Rapid diversification of alpine bamboos associated with the uplift of the Hengduan Mountains. Journal of Biogeography 2019; 00: 1– 12 

see tree on page 10 of Supplementary Information, but note this is not the ML tree they report instead


Zeng, C.X., Zhang Y.X., Triplett, J.K., Yang, J.B., & Li, D.Z. (2010). Large multi-locus plastid phylogeny of the tribe Arundinarieae (Poaceae: Bambusoideae) reveals ten major lineages and low rate of molecular divergence. Molecular Phylogenetics and Evolution, 56: 821–839. DOI: 10.1016/j.ympev.2010.03.041 


[Home] [Introduction] [Background] [Identification] [Cultivated Bamboos] [Morphology] [Classification] [more  Classification] [Nomenclature] [Origins] [Pests] [Author] [Publications] [Bamboos of Bhutan] [Flowers of La Plagne] [ABS Journal]