During the past two decades, field investigations by both botanists and plant hobbyists discovered numerous new gesneriads in China, providing surprising insights into species diversity of Gesneriaceae. The number of species in Gesneriaceae in China has had a marked increase from 1990 to now, evident in the 413 species recorded in the Flora Reipublicae Popularis Sinicae (Wang et al. 1990) and the ca. 690 species compiled in the latest list of Chinese Gesneriaceae (Wen et al. 2021). There are more than 60 authors who have recently contributed to the species discovery in China (Möller 2019). However, compared to the strong enthusiasm of searching and describing new species, few people focus their interest on older and problematic species (but see Hong et al. 2020; Lu et al. 2020; Yang et al. 2020). In fact, several species of Gesneriaceae from China described earlier need to be reassessed due to a lack of key morphological characters. For example, Oreocharis rhytidophylla C.Y.Wu ex H.W.Li was a doubtful species described on the basis of a single collection with only fruiting material, and its specific status was not confirmed until the rediscovery of flowering plants in the type locality recently (Zhang et al. 2019). An accurate estimate of biological species is critical for fully understanding biodiversity. Therefore, in the present study, we focus on another unknown species, i.e. Chirita umbrophila C.Y.Wu ex H.W.Li.

Chirita umbrophila was originally described by Wu (in Li 1983) based on one collection (Ping-Hua Yu 1189; Fig. 1A) without flowers. The type locality of C. umbrophila is Zhenxiong county, Yunnan province, China (Fig. 2E), and up until now no other specimens were collected after the gathering of the type specimens. Li (1983) placed C. umbrophila in Chirita sect. Chirita (most species in this section had been transferred to the genus Henckelia Spreng. by Weber et al. (2011)), and allied it to Chirita pumila D.Don [≡Henckelia pumila (D.Don) A.Dietr.] and Chirita forrestii J.Anthony [≡Henckelia forrestii (J.Anthony) D.J.Middleton & Mich.Möller]. The recognition of Chirita umbrophila as a species belonging to Chirita Buch.Ham. ex Don was mainly based on its taller stem, green-spotted leaves, smaller sepals, and stubby capsules compared to its allies (Li 1983). But the most important diagnostic feature of Chirita (now a synonym of Henckelia Spreng.) was the presence of the chiritoid stigmas (Brown 1840; Burtt 1965; Wood 1974; Wang 1985a) characterized by a barely developed upper stigma and enlarged lower stigma with usually emarginate to bifid apex (see Results regarding the description of stigma; Weber et al. 2020). However, it was not possible to examine the chiritoid stigmas of Chirita umbrophila in the fruiting type specimens. Therefore, Wang (1985b) treated C. umbrophila as a species with uncertain generic affiliation. The generic affiliation of C. umbrophila was not addressed in several following important publications, including the Flora Reipublicae Popularis Sinicae (Wang et al. 1990), Flora of China (Wang et al. 1998), and Plants of Gesneriaceae in China (Li and Wang 2004). Then, Chirita was shown to be highly polyphyletic and disintegrated based on both molecular and morphological evidence (Wang et al. 2011; Weber et al. 2011), and Weber et al. (2011) refrained from placing the species into Henckelia and cited “Chirita umbrophila C.Y.Wu ex H.W.Li: generic placement uncertain (see Wang et al. 1998).” Up to now, Chirita umbrophila remains a species with unknown generic affiliation.

Here we examine the phylogenetic relationships of C. umbrophila compared to morphologically similar species and establish species boundaries based on monophyletic groups that can be characterized with phenotypic characters. The aims of this research were to: 1) investigate the floral morphology of Chirita umbrophila, especially the stigma; and to 2) provide a taxonomic treatment based on the results of molecular phylogenetic reconstruction and comparative morphology.

Fig. 1.

Type specimens of Chirita umbrophila (A) and Loxostigma kurzii (B–D). A. Holotype of Chirita umbrophila (China, Yunnan, Zhenxiong, P.H. Yu 1189, KUN [barcode] KUN1219130). B. Holotype of Didymocarpus kurzii (India, Sikkim, Kurz s.n., CAL [barcode] CAL0000019236). C. Lectotype of Didissandra amabilis (China, Yunnan, Lijiang, G. Forrest 2689, E [barcode] E00387561). D. Lectotype of Briggsia amabilis var. taliensis (China, Yunnan, Dali, G. Forrest 4385, E [barcode] E00387582). Note: B–D, copyright © The director of CAL and E, respectively. Used with permission.

Fig. 2.

A–D. Variation of corolla color of Loxostigma kurzii. E. The known geographical distribution of L. kurzii and Chirita umbrophila. The type localities and the samples included in the phylogenetic analysis (Fig. 3) are labeled on the map.

Materials and Methods

Sampling for Phylogenetic Study, Amplification, and Sequencing—To infer the possible phylogenetic position of Chirita umbrophila, we included 61 accessions representing 19 out of 35 genera of subtribe Didymocarpinae (GRC 2021). We included three individuals of Loxostigma kurzii (C.B.Clarke) B.L.Burtt, which has a wide distribution from eastern Nepal to southwestern China (Fig. 2E; Wang et al. 1998; Sun and Zhou 2002; Möller et al. 2014; Sinha and Datta 2016). Two species of Microchirita (C.B.Clarke) Yin Z.Wang were chosen as outgroups according to the phylogeny inferred by Möller et al. (2011a). In addition, we selected representative samples of the straight-fruited advanced Asiatic and Malesian genera (Möller et al. 2009) as ingroups based on main clades recovered in Möller et al. (2011a). Sequences of approximately half of the included samples were retrieved from several preceding phylogenetic studies (Möller et al. 2009, 2011a; Qiu et al. 2015; Xu et al. 2021) and 141 sequences of the remaining samples were newly acquired. The full list of sampled species and corresponding NCBI accession numbers are provided in Appendix 1.

Total genomic DNA was extracted from silica-dried leaves using a modified CTAB protocol (Doyle and Doyle 1987) and used as the template in the polymerase chain reaction (PCR). The nuclear ribosomal external transcribed spacer (ETS), internal transcribed spacer (ITS, including ITS1, 5.8S subunit, and ITS2), and three chloroplast DNA regions (rpl16, rps16, trnL-F) were amplified using primers ETS-B/18S-IGS (Beardsley and Olmstead 2002), ITS-p5/u4 (Cheng et al. 2016), rpl16-F71/R1661 (Jordan et al. 1996; Kelchner and Clark 1997), rps16–2F/R3 (Bremer et al. 2002), and trnL-F-c/f (Taberlet et al. 1991), respectively. The PCR conditions for all DNA regions included initial denaturation at 94°C for 3 min, followed by 35 cycles of 94°C for 30 s, 54°C for 30 s, and 72°C for 50 s, with a final extension at 72°C for 5 min. The PCR products were purified using Tian quick Midi Purification Kit (Tiangen Biotech, Beijing, China) following the manufacturer's protocol, and then were directly sequenced with the same primers used for PCR amplification. All the sequences generated in the present study have been deposited in GenBank (Appendix 1).

Sequence Alignment and Phylogenetic Analyses—Editing and contig assembly were performed in Geneious v. 7.1.8 (Kearse et al. 2012). The newly obtained sequences together with those retrieved from GenBank were aligned with Clustal W (Larkin et al. 2007), followed by manual adjustment in Geneious v. 7.1.8. Three data matrices, i.e. nuclear (concatenated ETS and ITS), chloroplast (concatenated rpl16, rps16, and trnL-F), and the combined dataset (concatenated nuclear and chloroplast datasets), were generated and used for downstream phylogenetic analyses. The combinability of nuclear and chloroplast was determined by the incongruence length difference (ILD) test (Farris et al.1994, 1995) implemented in PAUP^* 4.0b10 (Swofford 2003). Because the systematic positions for five species, i.e. Didymocarpus glandulosus var. minor (W.T.Wang) W.T.Wang, Allostigma guangxiense W.T.Wang, Allocheilos guangxiensis H.Q.Wen, Y.G.Wei & S.H.Zhong, Petrocosmea minor Hemsl., and Didymostigma obtusum (C.B.Clarke) W.T.Wang, were strongly conflicting between the nuclear and chloroplast phylogenies (see Results), we removed these species in the combined dataset. Both maximum likelihood (ML) and Bayesian inference (BI) were selected to infer the phylogenetic relationships. The ML analysis was performed in IQtree v. 1.7 (Nguyen et al. 2015), and the best-fit model for each gene was automatically determined using the implementation of ModelFinder (Kalyaanamoorthy et al. 2017). Branch support was assessed with 1000 standard nonparametric bootstrap (BS) replicates. Bayesian inference analyses were conducted using MrBayes v. 3.2.7 (Ronquist et al. 2012). For the BI analyses, we determined the best-fit model of evolution for each gene using MrModeltest v. 2.3 (Nylander 2004) according to the Akaike information criterion (AIC; Akaike 1974). The BI analyses consisted of two independent Markov chain Monte Carlo (MCMC) runs each with one cold chain and three heated chains for 10,000,000 generations. For all Bayesian analyses, we ensured that the standard deviation of split frequencies was less than 0.001 and the potential scale reduction factor approached 1 for all parameters. The MrBayes outputs were checked in Tracer v. 1.7 (Rambaut et al. 2018) to confirm that all effective sample size (ESS) values were more than 200. The first 25% of posterior trees were discarded as burn-in and the 50% majority rule consensus trees were generated from the remaining trees. The obtained trees were visualized in FigTree v. 1.4.4 (A. Rambaut, University of Edinburgh,  http://tree.bio.ed.ac.uk/software/figtree) and modified in Adobe Illustrator (Adobe Systems, Inc. San Jose, California).

Morphological Observations—Molecular phylogenetic analyses recovered Chirita umbrophila imbedded in the three accessions of Loxostigma kurzii (see results). Therefore, firstly, we consulted publications related to L. kurzii and C. umbrophila (Clarke 1874, 1876, 1883; Kuntze 1891; Diels 1912; Craib 1919; Evans 1928; Burtt 1975; Li 1983; Wang et al. 1990, 1998; Vitek et al. 1997, 2000; Li and Wang 2004; Weber et al. 2011; Möller et al. 2014). Then, we studied herbarium specimens of C. umbrophila and its allies, especially L. kurzii (≡ Didymocarpus kurzii C.B.Clarke = Didissandra amabilis Diels), in CAL, CUH, IBSC, KUN, and PE. We also examined high-resolution digital images and photographs of specimens through the online resources of Chinese Virtual Herbarium ( https://www.cvh.ac.cn/), herbarium of Royal Botanic Garden Edinburgh ( https://data.rbge.org.uk/search/herbarium/), GBIF ( https://www.gbif.org/), and JSTOR ( https://plants.jstor.org/). Additionally, living materials of C. umbrophila were observed in the type locality (Yunnan province, Zhenxiong county) and L. kurzii was observed in Sichuan (Xide, Muli, and Miyi counties) and Yunnan (Dali city) provinces. Voucher specimens of C. umbrophila were collected in the type locality and deposited in IBSC for future reference. Herbarium acronyms follow Thiers (2021).

Results and Discussion

Phylogenetic Analyses—The general statistics of all datasets were summarized in Table 1. The alignment length of nuclear dataset was 1353 base pairs (bp), of which 660 bp (48.78%) were parsimony-informative. The chloroplast alignment contained 2894 bp, including 10.57% parsimony-informative characters (306 bp). The total alignment length of the combined dataset was 4228 bp, of which 922 bp (21.81%) were parsimony-informative. Results of MrModeltest show that the best fit models were GTR+I+G for ETS, ITS, rpl16, and rps16 data matrices and GTR +G for trnL-F dataset according to the AIC (Table 1). The ILD test gave a value of p = 0.11 after removing the five conflicting taxa, indicating that the remaining data did not contain significant incongruence.

Table 1.

Summary of the alignments employed in the phylogenetic analysis.

Phylogenetic results inferred from different sequence matrices (nuclear, chloroplast, and combined datasets) and methods (ML and BI) recovered a strongly supported clade consisting of C. umbrophila and three samples of L. kurzii (Fig. 3; Figs. S1–S5, Li et al. 2021). Henckelia, which is sister to all the other ingroup taxa with maximum support, is phylogenetically isolated from the aforementioned clade (Fig. 3). The phylogenies inferred from both nuclear and combined datasets clearly revealed that C. umbrophila is clustered with one accession of L. kurzii (LPW2012029) with moderate support and the other two accessions of L. kurzii (MMO08–1249 and SCMG03) are grouped together with strong support (Fig. 3; Figs. S1–S2, S5). However, the chloroplast dataset failed to recover a bifurcating phylogeny for these four accessions (Figs. S3–S4) likely due to insufficient parsimony-informative characters. Supplemental figures (Figs. S1–S5) are available from the Dryad Digital Repository (Li et al. 2021).

Morphological Observations—The type locality of Chirita umbrophila can be unambiguously traced according to the record (China: Yunnan province, Zhenxiong county, Moduo town [now Changba town, Guochang village, Moduo branch], Bojiwan) of type specimens (Fig. 1A). Therefore, we successfully rediscovered C. umbrophila in the type locality in April 2021, but unfortunately, the aerial part of the plants was just emerging from the slender underground rhizome (Fig. 4H). Our second visit to the type locality of C. umbrophila was carried out in late August 2021, and C. umbrophila entered its flowering stage then. We found that the dense hairs on the stem of the young plants (Fig. 4A–B) gradually disappeared on mature plants, and the more or less unequal herbaceous opposite leaves usually cluster at the top of the stem (Fig. 4C). Most importantly, we observed the floral characters of C. umbrophila for the first time, and our detailed observations revealed that the corolla of this species is white with yellow-brown spots on the interior, the tube is distinctly ventricose on the lower side, four fertile stamens are coherent in pairs, the thecae are divaricate with confluent apex, and the two stigmas are equal or subequal (Fig. 4D–G).

Fig. 3.

Maximum likelihood (ML) tree inferred from combined nuclear (ETS, ITS) and chloroplast (rpl16, rps16, trnL-F) datasets. ML bootstrap values and Bayesian inference (BI) posterior probabilities are indicated on the left and right sides of the slashes, respectively. Asterisks indicate maximum support, and dashes represent no support. Colored dots and red triangle correspond to those in Fig. 2.

Fig. 4.

Morphological characters of Chirita umbrophila. A. Elongated shoots arising from the underground rhizomes. B. Young shoots, showing dense eglandular hairs on the stems. C. Habit, indicating the leaves are mainly arranged at the top of stem. D. Pistil, showing the calyx divided to base and the two sub-equal stigmas. E. Side view of the white and distinctly ventricose corolla. F. Front view of corolla. G. Dissection of the corolla, showing the yellow-brown spots and four fertile stamens with each pair coherent and thecae confluent. H. Slender underground rhizome. Scale bars: B, C, H = 10 cm; D–G = 2 cm.

The description of the stigma as “bilobed” in Gesneriaceae is often blurred as the two stigma lobes may either represent the lobes of the upper and lower carpel or fused carpel halves (left- and right-position of stigma lobes) (Burtt 1954; Weber et al. 2020). Since the two ovary carpels may each produce a stigma (Wang et al. 1998), we would like to treat the upper and lower “stigma lobes” as two stigmas in the present study to avoid confusion with the left- and right-position of stigma lobes frequently seen in species of Oreocharis Benth. and Petrocodon Hance as well as many Gesnerioideae (Weber et al. 2020). Our morphological observation found that the two well-developed stigmas of C. umbrophila (Fig. 4D) clearly demonstrate that it is not a member of the genus Chirita, which possess chiritoid stigmas (Wang et al. 2011; Weber et al. 2011, 2020). Chirita umbrophila is characterized by a broad ventricose white corolla with yellow-brown spots on the interior and four anthers cohering in pairs as well as two equally or sub-equally developed lamellate stigmas, which fit well into the concept of the genus Loxostigma (sensu Möller et al. 2014). In addition, its slender underground rhizomes, cylindrical stems, short internodes, often entire sepals, straight capsules, and ellipsoid seeds without appendages indicate its affinity with L. kurzii (Fig. 1B–D).

The main morphological difference between C. umbrophila and L. kurzii is that the former has a white corolla while the latter usually has a yellow to orange one (Wang et al. 1998). Actually, we didn't observe the orange corollas of L. kurzii possibly due to our insufficient field investigations. However, both white and pale-yellow corollas of L. kurzii could be easily found in the field (Fig. 2A–D), and populations with white corollas are rather common across the distribution range of L. kurzii and can be observed in Yunnan [Lijiang: Yu-Zhang Zhao 21414 (KUN); Xianggelila: Zhen-Dong Fang et al. DJDC-688 (SABG)] and Sichuan [Emeishan: Xing-Lin Jiang & Xiu-Shi Zhang 31823 (PE); Meigu: Anonymous 1930 (SM); Miyi: Peng-Wei Li LPW2012029 (PE); Zhaojue: Zhaojue Expedition 0909 (SM)] provinces. Besides, records in the Plant Photo Bank of China (PPBC;  http://ppbc.iplant.cn/) showed that individuals with white or light-yellow flowers even occur in the same population at Emeishan, Sichuan province. Clearly, Chirita umbrophila lies within the morphological range of variation for Loxostigma kurzii.

Our molecular phylogenetic analyses showed that Chirita umbrophila is imbedded in a strongly supported clade comprising otherwise three accessions of Loxostigma kurzii, and the comparative morphological analysis revealed that the two species are indistinguishable. Therefore, Chirita umbrophila should be reduced to synonymy of Loxostigma kurzii.

Taxonomic Treatment