The samples of Brachystemma calycinum and Stellaria ovatifolia were collected from silica-dried leaves tissue, and the vouchers were deposited in the herbarium of the College of Agriculture, Guangxi University (GAUA) and the detailed information is shown in Suppl. material 1. The total DNA of the samples were extracted by the CTAB protocol (Maddison and Maddison 2014). The PCR amplification of ITS [5F (White et al. 1990), 4R (White et al. 1990)], matK [390F (Smissen et al. 2002), 1440R (Smissen et al. 2002)], rbcL [1F (Kress and Erickson 2007), 724R (Kress and Erickson 2007)], rps16 [F (Popp and Oxelman 2001), R (Popp and Oxelman 2001)], trnL-F [C (Taberlet et al. 1991), F (Taberlet et al. 1991)] were performed as above cited. The sequencing of PCR products was performed by the Beijing Genomics Institute (BGI). Newly generated sequences are available in GenBank (https://www.ncbi.nlm.nih.gov/), and their accession numbers (in bold) and the sequences of Caryophyllaceae members downloaded from GenBank are listed in Table 1. The absent sequences were coded as missing data.

Sequences alignment were performed with MAFFT v.7.313 (Katoh and Standley 2013). Phylogenetic analyses were conducted separately on the nuclear ribosomal internal transcribed spacer (ITS) and plastid regions (matK, rbcL, trnL-F, and rps16) and then combined; no notable incongruence was found (Fig. 2). The Bayesian Inference (BI) trees were constructed using MRBAYES 3.2.6 (Ronquist and Huelsenbeck 2003), and the maximum likelihood (ML) trees were constructed by RAXML-HPC2 (Stamatakis 2006). ML trees were constructed on CIPRES Science Gateway (Miller et al. 2010) under the GTRGAMMA model with 1,000 bootstrap replicates and default values for the remaining parameters. In Bayesian inference analysis, PARTITIONFINDER v.2.1.1 (Lanfear et al. 2016) was applied to selected models of nucleotide substitution under the Akaike Information Criterion (AIC). Selected models consisted of SYM+I+G for ITS, GTR+G for matK, trnL-F, and rps16, HKY+I+G for rbcL. Each Markov chain Monte Carlo (MCMC) analysis was run for 2,000,000 generations with the tree sampled every 100 generations. The first 25% trees of each run as burn-in were discarded.

Figure 2. 

Phylogenetic relationships among the tribe Alsineae A ITS data B trnL-F, matK, rbcL, rps16 combined data. The numbers on the nodes are Bayesian posterior probabilities (PP > 0.5), maximum likelihood bootstrap percentages (BS > 50%), respectively. “*” indicates that the node is PP = 1.00/BS = 100%, “-” indicates that the node PP < 0.5/ BS < 50%.

Two morphological characters (petal margin and number of seeds) which were diagnostic characters in Brachystemma were selected to reconstruct the ancestral characters in the tribe Alsineae. MESQUITE v.3.6 (Maddison and Maddison 2014) was used to reconstruct the ancestral characters with default parameters, using the ML tree from the combined tree. Morphological characters were coded as the following: (a) the petals are entire or emarginate (coded as 0), apex lobed (less than 1/2 the length of the petals) (1), deeply lobed (longer than 1/2 the length of the petals) (2); (b) the number of seeds in a capsule is 1–3 (0), 4–6 (1), more than 6 (2) (Lu et al. 2001; Arabi et al. 2022).

In the Caryophyllaceae tree, Brachystemma calycinum and Stellaria ovatifolia were nested in the tribe Alsineae with strong support (PP = 1.00, BS = 100) (Fig. 3A). Moreover, in the tree encompassing Alsineae tribe, B. calycinum and S. ovatifolia formed a monophylum (PP = 1.00, BS = 99) with strong support (PP = 1.00, BS = 100) (Fig. 3B), which is sister to the clade composed of Schizotechium, Mesostemma, Lepyrodiclis, Shivparvatia, Odontostemma, and Pseudostellaria in this tree (Fig. 3B). Our results suggested Stellaria ovatifolia was closely related to Brachystemma, instead of either Stellaria s.str. or Schizotechium.

Figure 3. 

Phylogenetic relationships among the Caryophyllaceae (A) and the tribe Alsineae (B). Phylogenetic trees were conducted by ITS, trnL-F, matK, rbcL, rps16 combined sequences. The numbers on the nodes are Bayesian posterior probabilities (PP > 0.5), maximum likelihood bootstrap percentages (BS > 50%), respectively. “*” indicates that the node is PP = 1.00/BS = 100%, “-” indicates that the node PP < 0.5/ BS<50%.

The results of the ancestral character reconstruction indicated that petals with a lobed apex and numerous seeds may be the ancestral characters of the tribe Alsineae (Fig. 4). The presence of entire petals and 1–3 seeds became the diagnostic characters between Brachystemma and related genera. In addition, B. calycinum and S. ovatifolia shared the characters of 1–3 seeds and neither taxa has deeply bifid petals. It suggested a close relationship between B. calycinum and S. ovatifolia.

As currently defined, Brachystemma is a monotypic genus in the tribe Alsineae, which is characterized by annual subscandent life form, lax thyrse with many flowers, petals shorter than 1/2 the length of the sepals with entire margins, two styles, four-valved capsules, and one mature seed (Fig. 1) (Lu and Gilbert 2001). Our phylogenetic results also revealed that Brachystemma formed a single branch with S. ovatifolia (Fig. 2 and Fig. 3) and demonstrated that Brachystemma is an independent genus (S. ovatifolia will be discussed in the following paragraphs), which is consistent with traditional morphological studies (Fenzl 1840; Bentham and Hooker 1862; Pax and Hoffmann 1934; Bittrich 1993; Lu and Gilbert 2001; Takhtajan 2009). Furthermore, the phylogenetic position of Brachystemma was nested in the tribe Alsineae and sister to the clade composed of Schizotechium, Mesostemma, Lepyrodiclis, Shivparvatia, Odontostemma, and Pseudostellaria (Fig. 3). Nevertheless, Brachystemma can be morphologically distinguished from the related genera of this clade. Brachystemma and Lepyrodiclis share characters such as annual life form, lax thyrse, and two styles, but Brachystemma differs from the latter by subscandent life form and four-valved capsules (Lu et al. 2001). It also can be distinguished from Mesostemma, Pseudostellaria, and Schizotechium by annual life form, petals with entire margins, lax thyrse, and two styles (Lu et al. 2001; Arabi et al. 2022). It can be clearly distinguished from Shivparvatia by annual habit, lax thyrse, and two styles (Lu et al. 2001; Keshav and Kumar 2015). Finally, it can be segregated from Odontostemma by lax thyrse, petals with entire margin and wingless seeds (Lu et al. 2001; Sadeghian et al. 2015).

Our results indicated that petals with a lobed apex and numerous seeds may be the ancestral characters of the tribe Alsineae, which was consistent with previous studies (Greenberg and Donoghue 2011; Zhang et al. 2017). Brachystemma has entire petal margins, but it is sister to the clade composed of genera having lobed petals Schizotechium, Mesostemma, Odontostemma, and Pseudostellaria (except Pseudostellaria maximowicziana and Pseudostellaria tibetica) (Fig. 4). Moreover, the tribe Alsineae is defined by a many-seeded (rarely few- or one-seeded) capsule or a rarely indehiscent nutlet (Harbaugh et al. 2010; Greenberg and Donoghue 2011; Arabi et al. 2022), but above genera having lobed petals share the character of fewer seeds (a capsule) (Fig. 4). The tribe Alsineae may have developed in an evolutionary direction toward fewer seeds. In addition, B. calycinum may be a species with diverse petals based on our field observations. B. calycinum may also include long (longer than sepals) and apically lobed petals (Fig. 1H), instead of only short (shorter than 1/2 the sepal length) and entire petals in the protologue (Fig. 1I). While additional observations in the field and specimens are required to confirm the petal condition, the petal condition in Brachystemma is coded here in accordance with the protologue.

Figure 4. 

Evolutionary cladograms of the distribution of two character in Alsineae A petal margin B number of seeds.

Although the placement of Stellaria ovatifolia among Brachystemma, Schizotechium and Stellaria has been uncertain for a long time, S. ovatifolia was considered more similar to B. calycinum in general appearance (Mizushima 1955; Wu and Ke 1996). It was clearly distinguished from the core Stellaria by subscandent life form (vs. non-scandent), lax thyrse (vs. cymes, rarely solitary), two styles (vs. three, rarely four or five), two-lobed (nearly to half of petal length) petals (vs. deeply-bifid petals), four-valved capsules (vs. six-valved capsules), and one mature seed (vs. many mature seeds) (Wu and Ke 1996; Lu and Gilbert 2001; Shilong and Rabeler 2001; Sharples and Tripp 2019). Despite being hypothesized to belong to Schizotechium (Pusalkar and Srivastava 2016), S. ovatifolia shows noticeable differences with Schizotechium, including a lax thyrse (vs. many-flowered compound cymes), two styles (vs. three styles), four-valved capsules (vs. six-valved capsules), and one mature seed (vs. one or two mature seeds) (Wu and Ke 1996; Shilong and Rabeler 2001; Pusalkar and Srivastava 2016). What is more, S. ovatifolia differs from Brachystemma by having two-lobed petals (nearly to half of petal length) and Stellaria type seeds, but they both share the following characters: subscandent life form, lax thyrse, two styles, four-valved capsules, and one mature seed (Fig. 1) (Wu and Ke 1996; Lu and Gilbert 2001; Shilong and Rabeler 2001). Hence, S. ovatifolia is highly similar to Brachystemma, instead of either Stellaria or Schizotechium. In terms of our molecular phylogeny, Stellaria ovatifolia is nested with Brachystemma calycinum in a clade with strong support (PP = 1.00, BS = 100) and not closely related to either Stellaria or Schizotechium in the nrDNA tree, cpDNA tree, and combined tree (Fig. 2 and Fig. 3). We believe that S. ovatifolia should be reclassified as a species of Brachystemma combining the evidence of similar general appearance and close phylogenetic relationship. As a result, the scientific name Brachystemma ovatifolium Mizushima is reinstated here. The main characters of Brachystemma now are: herbs annual or perennial; stems subscandent, branched; leaves opposite, petiolate; leaf ovate-lanceolate to lanceolate; stipules absent; inflorescence a thyrse or numerous in dichotomous, nearly subglobose cymes, terminal or axillary; flowers numerous, 5-merous, pedicellate; sepals free, subscarious, persisting in fruit; petals lanceolate or minute, much shorter than sepals, margin entire or bifid; stamens 5 or 10; styles 2; fruit a capsule, oblate, 4-valved, 1-seeded; seed reniform or globose.