Bredia bullata, B. enchengensis, and B. nitida (Sonerileae, Melastomataceae), three species occurring in Sino-Vietnamese limestone karst regions, are described as new. Molecular phylogenetic analyses and morphological divergence indicate that these species are well separated from their close relatives in Bredia, justifying their recognition as distinct species. Bredia bullata is unique in its interveinal areas prominently bullate each with an apical seta, a character otherwise never recorded in the genus. Bredia nitida resembles B. malipoensis in habit, leaf shape, and inflorescence morphology, but differs in the glabrescent and nitid adaxial leaf surface (vs. densely pubescent and subvelvety), ovate-elliptic or elliptic calyx lobes (vs. triangular to semiorbicular), and white petals (vs. purplish-red). Bredia enchengensis is closest to B. longiradiosa, but easily recognized by its prostrate habit (vs. erect), the yellowish-green, membranous and fragile leaves (vs. green or dark green, papery), and white anthers (vs. pink to purplish). These new discoveries show that further botanical exploration is warranted in the remote Sino-Vietnamese bordering region.

Bredia, karst, Melastomataceae, phylogeny, taxonomy

Karst is a kind of landscape characterized by a variety of closed surface depressions, a well-developed underground drainage system and a paucity of surface streams (Ford and Williams 2007). The complex terrains and variable climatic conditions on karsts provide numerous ecological niches that harbor a rich biodiversity (Clements et al. 2006). The vast karst terrain stretching across southern China and northern Vietnam connects two global biodiversity hotspots, viz. south-central China and Indo-Burma. It harbors remarkable biodiversity and a high level of endemism (Zhu 2007) and has been considered the model for karst studies (Sweeting 1978). As karst environments in these areas are often remote and under significant threats due to human activity, biodiversity survey and conservation are extremely urgent.

Bredia Blume (Melastomataceae) as currently circumscribed contains 24 species distributed from central and southern mainland China, Taiwan, northern Vietnam, to the Ryukyu Islands and Yakushima, Japan (Zhou et al. 2019a; Wen et al. 2019; Dai et al. 2020; He et al. 2020). Five species of Bredia, namely B. latisepala (C. Chen) R. Zhou & Ying Liu, B. longearistata (C. Chen) R. Zhou & Ying Liu, B. longiradiosa C. Chen ex Govaerts, B. malipoensis D. H. Peng, S. J. Zeng & Z. Y. Wen, and B. reniformis C. M. He, Y. H. Tong & S. J. Zeng, are restricted to limestone karst areas. These species share obvious resemblance in their isomorphic stamens and undulate petal margin ciliate with glandular hairs and thus are easily distinguished from the remaining species of the genus (Fig. 1). The only exception is B. reniformis, which does not have an undulate petal margin (Fig. 1D). Close relationships among the karst species were consistently recovered in previous phylogenetic studies based on nuclear ribosomal internal transcribed spacer (nrITS) and plastome sequences (Zhou et al. 2019a, b, c; Dai et al. 2020). From 2019 to 2021, multiple field expeditions were made to karst areas in southern Guangxi, southeastern Yunnan, and northern Vietnam. In the process, we encountered three species of Bredia with isomorphic stamens and undulate and ciliate petal margin that were morphologically distinct from limestone species. As shown in Fig. 2, the new taxa were found in three localities from Malipo County, Yunnan Province, China and Quan Ba District, Ha Giang Province, Vietnam (B. bullata J. H. Dai & Ying Liu; Figs 3, 4); in one locality from Daxin County, Guangxi Province, China (B. enchengensis J. H. Dai, Yan Liu & S. Y. Nong; Figs 5, 6); and in one locality from Hekou County, Yunnan Province, China (B. nitida J. H. Dai & Ying Liu; Figs 7, 8).

In this study, we inferred the phylogenetic position of the plants in question and then compared them with their close relatives in Bredia to evaluate their specific status. To this end, phylogenetic analyses were performed using sequence data of three nuclear markers (nrITS, Dbr1, and SOS4a) and one chloroplast intergenic spacer (trnV–trnM), sampling all species recorded in Bredia. The results confirmed our suspicion that these plants represented species of Bredia new to science. A key is provided for the karst species.

Figure 1. 

Species of Bredia adapted to limestone karst habitats A B. latisepala, Ying Liu 557 (SYS) B B. longiradiosa var. longiradiosa, Ying Liu 486 (SYS) C B. malipoensis, Yunnan Expedition Team 1073 (PE), photographs by Bing Liu (PE) D B. reniformis, Ying Liu 748 (SYS).

Figure 2. 

Distribution of Bredia bullata (solid circle), B. nitida (triangle), and B. enchengensis (star).

Morphological data for the new species and previous recorded karst species were obtained through field expeditions, herbarium specimens (A, E, GXMI, IBK, IBSC, PE, SYS, VNMN) and literature (Chen 1984; Chen and Renner 2007; Wen et al. 2019; He et al. 2020) surveys as well as by observing living individuals in the facilities of Sun Yat-sen University.

To infer the phylogenetic position of B. bullata, B. nitida, and B. enchengensis, the type species of related genera (Blastus Lour., Fordiophyton Stapf, Phyllagathis Blume, Tashiroea Matsum. ex Ito & Matsum.), and all 24 species so far recorded in Bredia were included in the analyses. Tashiroea yaeyamensis Matsum. was selected as the outgroup according to Zhou et al. (2019a, b, c). The source of the materials and GenBank accession numbers are given in Suppl. material 1.

Total DNA was extracted from fresh leaves using the modified CTAB procedure (Doyle and Doyle 1987). For amplification and sequencing of SOS4a, we used two primers (SOS4a-F: 5´-CAAGAAGGTGAGATCATCCAAA-3´, SOS4a-R: 5´-TAGTTTTGGCCTGCAATGCT-3´) adapted from Reginato and Michelangeli (2016). Primers published in Zhou et al. (2020) were used for Dbr1 and universal primers for nrITS and trnV–trnM (White et al. 1990; Hwang et al. 2000).

Sequences of four genes were aligned using MAFFT v.7.307 (Katoh and Standley 2013) and concatenated. Maximum likelihood (ML) analysis was performed in IQ-TREE v.2.1.4 (Nguyen et al. 2015). The optimal partitioning scheme and best-fitting model for each partition (Suppl. material 2) were selected using ModelFinder (Kalyaanamoorthy et al. 2017) under the Bayesian Information Criterion (BIC). Node support was evaluated by 1000 replicates of ultrafast bootstrap (UFBS) (Minh et al. 2013) and SH-aLRT test. For Bayesian inference (BI) analysis, we used PartitionFinder v.2.1.1 (Lanfear et al. 2017) for partitioning and model selection (Suppl. material 2). BI analysis was conducted in MrBayes v.3.2.6 (Huelsenbeck and Ronquist 2001). Two independent Markov chain Monte Carlo analyses (MCMC) were performed with four simultaneous chains of 2,000,000 generations sampling one tree every 100 generations. We verified that the average deviation of split frequencies had reached a value below 0.01 at the end of MCMC analyses. The first 25% of trees were discarded as burn-in and the remaining were used to construct a majority-rule consensus tree with Bayesian posterior probabilities (PP). Effective sample sizes (ESS) for all parameters and statistics were assessed using Tracer v.1.7.1 (Rambaut et al. 2018). Maximum parsimony (MP) analysis was carried out in PAUP v.4a165 (Swofford 2003). A heuristic search strategy was conducted of 1000 random addition replicates, with the tree-bisection-reconnection (TBR) branch swapping algorithm and MultTrees on. Maxtree was set to 500. We evaluated node support (BSMP) by 1000 bootstrap replicates of 1000 random additions.

The aligned sequence matrix contained 2536 characters. The optimal partitioning scheme contained three partitions, the statistics of which were summarized in Suppl. material 2. Trees from BI, ML, and MP analyses had identical topologies. The tree resulting from ML analysis is shown in Fig. 9, with PP, BSMP, UFBS, and support values from SH-aLRT test indicated at nodes. Bredia bullata, B. nitida, and B. enchengensis formed a strongly supported clade with the other 24 species of Bredia (PP = 1; BSMP = 99%; SH-aLRT test = 99%, UFBS = 98%). Within Bredia, the three new taxa formed a clade with the other karst species (karst clade, Fig. 9), although with low support (PP = 0.67; BSMP = 25%; SH-aLRT test = 0%, UFBS = 66%). Bredia enchengensis was recovered as sister to B. longiradiosa (PP = 1; BSMP = 100%; SH-aLRT test = 100%, UFBS = 97%), while B. nitida and B. bullata constituted a well resolved clade with B. malipoensis (PP = 1; BSMP = 100%; SH-aLRT test = 100%, UFBS = 100%).

Phylogenetic data and morphology confirmed that B. bullata, B. nitida, and B. enchengensis belong in Bredia. All three species have cordate leaf blades, cymose inflorescences, isomorphic stamens, gibbous anthers and enlarged ovary crowns, all of which are typical of Bredia (Figs 3–8). In the present phylogenetic analyses (Fig. 9), the limestone species of Bredia formed a clade containing four subclades, viz. B. reniformis, B. latisepala-B. longearistata, B. longiradiosa-B. enchengensis, and B. malipoensis-B. nitida-B. bullata. Nevertheless, the karst clade is still weakly supported, as well as the relationships among its four subclades. Further molecular sampling is desired to improve these phylogenetic relationships.

Among the three species in question, B. enchengensis was well resolved as sister to B. longiradiosa (Fig. 9). It resembles B. longiradiosa in the somewhat broadly ovate leaf blade, inflorescence often an umbellate cyme, undulate petals with ciliate margin, and isomorphic stamens, but differs markedly from the latter in the prostrate habit (vs. erect), densely pubescent stem (vs. sparsely villous or glabrescent), yellowish-green, membranous and fragile leaves (vs. green or dark green, papery), and white anthers (vs. pink or purplish) (Figs 1B, 6). The remaining two species, namely B. bullata and B. nitida, formed another karst subclade in the genus with B. malipoensis (Fig. 9). Bredia bullata is distinct in its strongly sunken adaxial leaf veins with interveinal areas prominently bullate each with a short apical seta (Fig. 4E), a character otherwise never recorded in the genus. Bredia nitida shares general similarities with B. malipoensis in leaf shape and morphology of the inflorescence, petals, and stamens, but is easily distinguished from the latter in the often glabrescent stem and leaves at maturity (vs. densely pubescent), nitid upper leaf surface (vs. subvelvety), ovate-elliptic or elliptic calyx lobes (vs. triangular to semiorbicular), and white petals (vs. purplish-red) (Figs 1C, 8). Based on the phylogenetic data and morphological divergence, B. bullata, B. nitida, and B. enchengensis should be recognized as distinct species in Bredia.

The Sino-Vietnamese limestone karst region provides a multitude of habitats, such as cliffs, caves, and shaded fissures/crevices (Schindler 1982; Xu 1995; Zhu 2007). For some calciphilous herbaceous plant groups with low vagility, such isolated habitats/microhabitats likely promote allopatric speciation and a steady accumulation of species over time, resulting in a high diversity of narrowly endemic species (Hughes and Hollingsworth 2008; Chung et al. 2014). Aspidistra Ker Gawl. (e.g., Liu et al. 2011), Begonia L. (e.g., Chung et al. 2014), Impatiens L. (e.g., Xue et al. 2020), and Primulina Hance (e.g., Kong et al. 2017) are among the most famous examples. The Sino-Vietnamese limestone areas, where seven species of Bredia have been recorded, is a diversification center for the genus. These species are capsule-fruited and disperse their seeds by raindrops and wind, often within a short distance from the mother plant. Current data indicate that geographic isolation is likely the primary mode of species diversification, in a group with limited distribution range or even site-endemics. The Sino-Vietnamese karst areas are hotspots of species richness and endemism and have been an important source of vascular plant novelties in the past 20 years (Du et al. 2020; Qian et al. 2020). The remote border regions should be further explored to fully unravel the rich biodiversity there.

We thank Ping Yang (IBK) and the staff of the Forestry and Grassland Bureau of Hekou County and Malipo Laoshan Provincial Natural Reserve for their kind assistance during the field survey and Dr. Bing Liu (PE) for providing photos of B. malipoensis. This work was supported by the National Natural Science Foundation of China (grants 32170220, 31770214), Natural Science Foundation of Guangdong Province (grant 2021A1515011214), and partly by the Ministry of Planning and Investment, Vietnam, and the Vietnam Academy of Science and Technology under the project code UQĐTCB.06/22–23 to TVD.