Taxonomic notes on Scutellaria taipeiensis (Lamiaceae) from morphological and molecular data

Abstract The genus Scutellaria comprises eight species distributed from 50 to 2000 m in Taiwan. Amongst them, S. barbata and S. taipeiensis are very similar on the basis of morphological and plastid DNA sequence information. Therefore, a comprehensive study of the taxonomic status of S. taipeiensis is necessary. We reviewed the herbarium sheets, related literature and protologues and compared morphologies of these two species, as well as their phylogenetic relationships. All evidence, including the diagnostic characters between S. taipeiensis and S. barbata, suggest that they belonged to a single species rather than two. As a result, S. taipeiensis is treated as a synonym of S. barbata.


Introduction
The genus Scutellaria L. is composed of approximately 360 species worldwide (Paton 1990;Li and Hedge 1994;Harley et al. 2004). This genus is characterised by being non-aromatic, having simple leaves with entire to pinnatifid margins, a terminal or axillary raceme-like thrysoid inflorescence with single-flowered cymes, a two-lobed calyx with a scutellum on the upper lobe and a two-lobed corolla with an often saccate or spurred base, anterior anthers dimidiate due to aborted development of upper thecae and the ovary being borne on a peg-like gynophore (Paton 1990;Li and Hedge 1994).
The Taiwanese Scutellaria were revised in the 1990s, based on morphology and palynology and five species were recognised (Hsieh and Huang 1995). Later, a new species, S. austrotaiwanensis T. H. Hsieh & T. C. Huang was described (Hsieh and Huang 1997), resulting in a total of six species recorded in the second edition of the Flora of Taiwan (Huang et al. 1998). Two new species, S. taipeiensis T. C. Huang, A. Hsiao & M. J. Wu and S. hsiehii T. H. Hsieh, were described subsequently (Huang et al. 2003;Hsieh 2013). A genetic study of S. barbata D. Don and S. taipeiensis was conducted by Hsiung et al. (2017) and the data showed no remarkable divergence between these two species. These results attracted our attention to verify their findings. Therefore, we revised the taxonomic status of S. taipeiensis after re-evaluating morphological and plastid DNA sequence evidence in this study.

Morphological comparison
Study materials were obtained from herbarium sheets of the HAST, TAI and TAIF herbaria and from living plants (herbarium acronyms follow Index Herbariorum (Thiers 2019, continuously updated). Type specimens of S. taipeiensis, deposited in the herbarium of the National Taiwan University (TAI), were also examined. Voucher specimens were deposited in the herbarium of the Taiwan Forestry Research Institute (TAIF). We examined leaf, floral and fruit morphology from dried and living materials. For living materials, we observed four populations of S. barbata and two of S. taipeiensis, including the type locality. Scutellaria barbata is widespread in Taiwan. Hence, herbarium sheets complement the fresh material gathered so that the variation, present in Taiwan, was represented in the study. For the population of S. taipeiensis, only few populations, including the type, have been recorded. All of these populations were located in Taipei City. Thus, our observation covered all populations in Taiwan. Observation of the nutlet sculpture of Hsiung et al. (2017) was applied here as a reference. The identification of S. barbata, S. taipeiensis and other Taiwanese species was according to the protologues of Huang et al. (2003) and other related literature (Hsieh and Huang 1995;Huang et al. 1998).

Molecular analysis
In order to revise the taxonomic state of S. taipeiensis, phylogenetic trees were reconstructed. The species, selected for analysis, were from Chiang et al. (2012) and Holmskioldia sanguinea was applied as outgroup, since it was closely related to Scutellaria (Bendiksby et al. 2011). Two nuclear (CAD, CHS) and three chloroplast DNA fragments (matK, ndhF-rpl32 and rpl32-trnL) were used by Chiang et al. (2012), amongst them, ndhF-rpl32 and rpl32-trnL being also applied in the study of Hsiung et al. (2017). Two chloroplast regions (ndhF-rpl32 and rpl32-trnL) were applied in the phylogenetic analysis of this study. In addition to the sequences from Chiang et al. (2012) and Hsiung et al. (2017), Table 1. Sequences and accession number of sequences applied in this study. Sequences generated for this study are marked *. Other sequences were sourced from Genbank.
we sequenced the chloroplast DNA fragments of ndhF-rpl32 spacer from S. galericulata, S. incana and H. sanguinea and rpl32-trnL spacer from S. galericulata, S. incana, S. taiwanensis and H. sanguinea. These newly generated sequences were amplified following the procedure of Hsiung et al. (2017). All sequences, applied in this study, are listed in Table 1. These sequences were used for phylogeny reconstruction by Bayesian Inference (BI), Maximum Likelihood (ML) and Neighbour-Joining (NJ) approaches. The variable sites, parsimony-informative sites and substitution model were checked and selected by MEGA 7 (Kumar et al. 2016). The optimal model with the highest BIC and AIC values was selected for BI and ML analyses (Kumar et al. 2016) ( Table 2). The BI reconstruction was conducted using Mr. Bayes 3.2.6 (Ronquist et al. 2012). Two independent runs were conducted with 10,000,000 generations, sampled every 1000 generations and a 10% dataset was discarded as burn-in. ML analysis was performed by PhyML 3.1 (Guindon et al. 2010). The substitution model of the two loci was the same as for BI analysis and the gamma distribution parameter was fixed at 1.52 and 0.35, respectively, according to the results of model selection. The tree topology search operation was set as the best of NNI and SPR (Guindon et al. 2010). The approximate likelihood ratio test non-parametric branch support was based on a Shimodaira-Hasegawa-like procedure (Guindon et al. 2010). NJ analysis was conducted using MEGA 7, with 1000 bootstrap resamplings. All of the phylogenetic trees were summarised and output by FigTree 1.4.4 (Rambaut 2012).

Leaves
Leaf morphology had been regarded as a diagnostic character for distinguishing S. barbata from S. taipeiensis (Huang et al. 2003). Leaf shape of S. barbata varies from suborbicular to narrowly lanceolate ( Fig. 1A-D); in contrast, the leaf shape of S. taipeiensis varies from ovate to broadly ovate (Fig. 1E, F). The leaves of both species had sparse pubescence on the abaxial surface. The leaves of S. barbata are 1.1−2.8 cm long and 0.9−1.4 cm wide, the length-width-ratio from 1.1 to 2.0, while S. taipeiensis leaves are 1.0−1.7 cm long and 0.5−1.1 cm wide, the length-width-ratio from 1.5 to 2.0. The shapes and sizes of the leaves overlapped between the two species and thus were difficult for use as a diagnostic character to distinguish species.

Inflorescence and flowers
The floral morphology of S. barbata ( Fig. 2A, C) was very similar to that of S. taipeiensis (Fig. 2B, D). They both had terminal inflorescences and bilabiate flowers that were only slightly curved near base (Fig. 3), while other species of Taiwan have geniculate (e.g. S. austrotaiwanensis, S. indica etc.) or a strongly curved corolla (e.g. S. tashiroi). The corolla was bluish-purple, 0.8−1.3 cm long and pubescent on the outer surface.

Nutlets
According to the observations of Hsiung et al. (2017), the sculpture of S. barbata and S. taipeiensis are rounded concentric type and no other difference is found between them.

Molecular phylogeny
The best substitution model for both fragments, ndhF-rpl32 and rpl32-trnL, was HKY+G. The phylogenetic trees, reconstructed by ML, NJ and BI, revealed similar topologies with slight differences in the arrangement of non Taiwanese species (Figs 4, 5; Suppl. materials 1−4). In all analyses, all Taiwanese species formed a clade with moderate to high support (PP = 0.99, bootstrap = 0.65-0.91). Amongst them, S. barbata and S. taipeiensis formed a highly supported monophyletic group (PP = 1.0, bootstrap = 0.87-0.95), but neither S. barbata nor S. taipeiensis formed a single clade. Instead, S. taipeiensis was nested with S. barbata in the phylogenetic tree, i.e. neither species being monophyletic as currently delimited.

Distribution and habitat
Scutellaria barbata grows throughout low altitudes, ca. 50-500 m, in Taiwan, but lower in southern parts. Scutellaria taipeiensis is found in Taipei City and New Taipei City.
The two species were growing in similar habitats, such as grassland, roadside, riverbank or forest margin and often with high humidity. Some references regarded S. barbata as an aquatic plant in a broad sense (Chen 1990) due to its wetland habitat. This could also be true for S. taipeiensis, according to the field investigation. No apparent differentiation was observed in the distribution and habitat types between these two species.

Discussion
The taxonomic status of S. taipeiensis Scutellaria taipeiensis was first described by Huang et al. (2003), based on the morphology of the leaves and the nutlets. According to the original description, the leaves of S. taipeiensis were triangular-ovate to broadly ovate and less than twice as long as the width. To confirm the similarity of the leaves, we collected and compared leaf morphology of the two species amongst several populations. The results showed that the variation within S. barbata was larger than the difference between S. barbata and S. taipeiensis. Additionally, the length-width-ratio of leaves is the same in both species. Therefore, leaf morphology could not be regarded as a diagnostic character for these two species.
With regard to the nutlets, the coat had also been used to distinguish S. barbata from S. taipeiensis. The former had a radiating, umbrella-like shape, while the latter was a rounded concentric type (Huang et al. 2003). However, Hsiung et al. (2017) reviewed this character on a population level and found no remarkable difference between the two species. The mature nutlets appear to the rounded concentric type in both species, such state was stable amongst populations (Hsiung et al. 2017). The umbrella-like appendage was found in immature nutlets only, which meant that it was a transitional state during nutlet development and could not provide a valid taxonomic value.
We further looked at other characters to separate them, such as floral morphology and DNA sequence data. Different sequence data revealed some phylogenetic incongruence amongst lineages S. alpina, S. altissima, S. bicalensis, S. diffusa, S. galericulata, S. salviifolia and S. zhongdianensis. Such incongruence may be due to uneven sampling, but the relationship of these species was not a concern in this study. Therefore, we will not discuss the evolutionary relationship between this group of species here. All Taiwanese species formed a highly supported clade in all trees. Scutellaria barbata and S. taipeiensis, which had very similar inflorescences and floral morphology, are phylogenetically nested within a monophyletic clade. Based on this evidence, S. taipeiensis was treated as a synonym of S. barbata, rather than a distinct species or on an intraspecific level.

Taxonomic treatment
According to the results and discussions, we established the following taxonomy: Taipei City, growing on exposed rocks or soils adjacent rocks along sunny roadside between Maukong to Chihnan Temple T. C. Huang and A. Hsiao 18104 (Holotype: TAI!, Isotype: TAI!) syn. nov.

Scutellaria barbata D. Don in Prodromus Florae
Distribution. Scutellaria barbata is widely distributed in southern and eastern Asia (Li and Hedge 1994). In Taiwan, this species is found in low altitude from 50 to 500 m, in wet grasslands, riverside and margins of forest. Specimens examined. Specimens marked with an asterisk (*) denote material S. taipeiensis following.the concept of Huang et al. (2003)