Boehmeria Jacq., as currently circumscribed (Wilmot-Dear & Friis 2013), comprises 52 species and has a pantropical distribution (Plants of the World Online 2020). Whilst the greatest diversity of species occurs in Asia, 14 species are known from the Americas (Wilmot-Dear & Friis 1996, 2013) including the type of the genus, B. ramiflora Jacq., which is native to Central America and north-western South America (Plants of the World Online 2020).

Boehmeria includes the species of greatest economic value within the nettle family (Urticaceae), B. nivea, known as ‘ramie’. Ramie is presumed native to China where it has been in cultivation for over 3000 years (Chen et al. 2003). Boehmeria nivea was originally described in the nettle genus Urtica by Linnaeus (1753) before being transferred to the genus Boehmeria by Gaudichaud-Baupré (1830). In (1891) Kuntze transferred the species to the genus Ramium but in doing so generated an illegitimate homonym and in the most recent revision of the genus (Wilmot-Dear & Friis 2013), it remains a species of Boehmeria.

Within a systematic framework, subgeneric classifications, such as sections or subgenera, provide a means of recognizing major evolutionary lineages that do not warrant generic rank. Identifying such lineages provides a framework for testing evolutionary hypotheses and in designing plant breeding programmes (Zhang et al. 1998). Indeed, Liang et al. (2020), as justification for their study, state that, ‘Boehmeria is the most important natural germplasm bank for deriving commercially viable ramie cultivars’. It is therefore crucial to their study that Boehmeria nivea can be assigned to the genus Boehmeria. If not, then in designing a breeding programme potentially important sources of germplasm may be overlooked.

Within systematic biology, it is implicit that the genus and sections should represent monophyletic groups. Previous phylogenetic studies based on the analysis of chloroplast, nuclear ribosomal and low-copy nuclear DNA have strongly suggested that Boehmeria does not form a monophyletic group (Fig. 1 in Wu et al. 2013; Fig S1 in Wu et al. 2018). Within these studies, Boehmeria is recovered as two or four monophyletic groupings most closely related to other genera in the Boehmerieae tribe of the Urticaceae. These include Boehmeria nivea, which is recovered as most closely related to Archiboehmeria, with strong support, whilst the core of Chinese species were recovered as sister to Cypholophus, with moderate to strong support and the Latin American species as sister to Cypholophus + Asian Boehmeria with strong support, and Boehmeria depauperata as sister to Latin American Boehmeria + Cypholophus + core of Asian Boehmeria (Fig S1 in Wu et al. 2018). This strongly suggests that Boehmeria is paraphyletic with respect to Asian Boehmeria, Cypholophus, American Boehmeria and B. depauperata, and polyphyletic with respect to B. nivea.

The selection of a single outgroup by Liang et al. (2020), comprising the most basal member of the Boehmerieae tribe, Oreocnide, meant that no effective test of monophyly was undertaken. Their resultant sectional classification combines distantly related taxa that only form a monophyletic group, with very low support, because of the application of an even more distantly related outgroup. In addition, in the presentation of their analyses of SNP data Liang et al. (2020) omit key details. For example, the filtering of homologous loci, a crucial step in the generation of phylogenetic trees from SNP data. In addition, the phylogenetic tree presented is based on neighbour-joining, rather than a maximum likelihood approach, and there seems no clear justification for this. Maximum likelihood approaches are accepted as more reliable for estimating relationships between taxa, than neighbour-joining (Holder & Lewis 2003). Finally, the phylogenetic tree that they support has very low support values on most branches, less than 50% in some cases, suggesting that they are poor predictions of relationships. Such low support values suggest strong incongruence or conflict within their data and that they have in fact recovered a large polytomy, rather than a resolved tree. That is, that many other groupings and relationships between groupings may be equally plausible, not a basis for a stable classification.

The most obvious risk of using this classification would be with respect to designing a crop breeding programme for ramie. The classification proposed by Liang et al. (2020) would suggest that, outside of Boehmeria nivea, the nearest relatives of ramie suitable for inclusion in a crop breeding programme would be from their sections, Siamensis, Duretia, Phyllostachys or Silvestri, whereas research published by Wu et al. (2013, 2018) suggests that populations of Archiboehmeria atrata are the nearest relatives of B. nivea. Using the classification of Liang et al. (2020) would, however, exclude this possibility.

We conclude that poor taxon sampling, a failure to consult the relevant molecular biology and taxonomic literature, internal conflict within their data and the over interpretation of low support values has resulted in the erroneous conclusion that Boehmeria represents a monophyletic or ‘natural’ genus and the presentation of an inaccurate estimate of relationships within Boehmeria, especially with respect to Boehmeria nivea, ‘ramie’.