P. brevistyla, P. delavayi and P. leptophylla. belong to Celastraceae, Parnassia L., Sect. Nectarotrilobos Drude, Subsect. Xiphosandra (Franch.) Ku. Due to lack of material, the reliability of their taxonomic characteristics remained unknown and all three species have been retained. Following extensive field investigations, population sampling and examination of specimens, we determined that the previously used characteristics to distinguish these three species, i.e. the shape of basal leaves and the depth of staminode lobes, vary continuously within populations and should not be used to distinguish separate species. Therefore, P. brevistyla and P. leptophylla are hereby designated as synonyms of P. delavayi.

Subsect. Xiphosandra, Celastraceae , new synonym, taxonomy

Parnassia L. (Linnaeus 1753) is a genus that is predominantly distributed in the arctic and temperate zones of the Northern Hemisphere; it is most speciose in China and the Himalayas (Simmons 2004). All species of Parnassia are perennial herbs, glabrous; flowers solitary, borne on unbranched scape, actinomorphic, hermaphrodite; staminodes 5, inserted opposite to petals, terete and entire or flat and divided into lobes or filiform rays. The genus was traditionally placed in the family Saxifragaceae (Hooker and Thompson 1858, Engler 1930, Cronquist 1981, Ku 1995, Ku and Hultgård 2001); however, it was recently transferred to Celastraceae (APG IV 2016, Ball 2016) based on molecular evidence (Soltis et al. 1997, Soltis et al. 2000, Zhang and Simmons 2006, APG IV 2016). The latest updated checklist of the genus recognised 61 species, 2 subspecies, 11 varieties and 1 forma (Shu et al. 2017).

Subsect. Xiphosandra (Franch.) Ku (1987) has been recognised as consisting of those species in which the stamen connectives apically project as lanceolate appendages (Fig. 1g). It includes three species: P. brevistyla, P. delavayi and P. leptophylla. Based on unique morphological characteristics (i.e. the stamen connectives apically project as lanceolate appendages), the three species were placed in Sect. Xiphosandra Franch. (Franchet 1897) with the support of Wu et al. (2003). Alternatively, most researchers placed them into Subsect. Xiphosandra of Sect. Nectarotrilobos Drude (Ku 1987, Wu 2005) or directly into Sect. Nectarotrilobos due to their flat and 3-lobed staminodes (Nekrassova 1927, Engler 1930, Handel-Mazzetti 1941).

Figure 1. 

Parnassia delavayi: habitat (a), individual (b), flower (c), fruits (d, e), staminodes (f) and stamens (g). Scale bar: 1 mm.

Parnassia delavayi Franch. was published based on three collections (Delavay 217, Delavay s.n. and Delavay Parnassia n. 2) (Franchet 1896). Then, Parnassia delavayi Franch. var. brevistyla Brieger was proposed in 1922 based on specimens collected from Tibet, China, 3750 m and Schenhsi, China, 2600 m, s.n. (Brieger 1922). Handel-Mazzetti (1931) promoted Parnassia delavayi var. brevistyla into the new species Parnassia brevistyla (Brieg.) Hand.-Mazz. Handel-Mazzetti (1941) published 10 new taxa, including Parnassia leptophylla Hand.-Mazz., on the basis of a specimen collected from Baoxing, Sichuan, China—KL Chu 3231. Since the authors did not specify the holotype of Parnassia delavayi or P. leptophylla in their publications, P00709352 (Fig. 2) was designated as the lectotype of P. delavayi and PE00866146 (Fig. 3) was designated as the lectotype of P. leptophylla by Shu et al. (2017). However, according to the information provided by the original description of P. brevistyla, these types of P. brevistyla were not found after reviewing the specimens and high-resolution images of specimens in major herbaria, which was also mentioned by Stafleu and Mennega (1995).

Figure 2. 

Parnassia delavayi, lectotype (P05556394).

Figure 3. 

Parnassia leptophylla, lectotype (PE00866146).

Based on the descriptions of P. brevistyla, P. delavayi and P. leptophylla, the differences amongst them mainly consist of two characteristics: the shape of leaf blades and the depth of staminode lobes (Ku 1995, Table 1). With regard to the shape of blades: P. brevistyla has ovate-cordate basal leaves, while the other two species have reniform or occasionally suborbicular basal leaves. With regard to the depth of staminode lobes: P. brevistyla staminodes are shallowly 3-lobed and the length of lobes is 1/6 or less of staminodes’ length; P. delavayi staminodes are shallowly 3-lobed or to half their length; P. leptophylla staminodes are 3-lobed for ca. 3/4 of their length.

Through observation of herbarium specimens, we found that the three species of Subsect. Xiphosandra are extremely similar in morphology and there is obvious overlap amongst P. brevistyla, P. delavayi and P. leptophylla in the shape of leaf blades and the depth of staminode lobes. Moreover, during extensive field work, we found that the shape of basal leaves and the depth of staminode lobes varied significantly within populations, even sometimes within individuals. Therefore, the aim of this study was to clarify the classification relationship amongst P. brevistyla, P. delavayi and P. leptophylla by re-evaluating their morphological characteristics, particularly the shape of blades and the depth of staminode lobes.

Specimens from BJFC, CDBI, IMC, KUN, QFNU, P, PE, SM, SZ and WCNU (Suppl. material 1) were examined during this study. High-resolution photographs were also captured. The measurements provided herein were mostly obtained from dried herbarium specimens. Since the shape of both the leaf base and the apex are basically identical, the leaf shape is defined by the aspect ratio of leaves in this study. The primary examined populations were in Chongqing, Gansu, Guizhou, Hebei, Henan, Hubei, Hunan, Shanxi, Sichuan, Tibet and Yunnan. These populations include the type localities of P. brevistyla, P. delavayi and P. leptophylla (Taibai, Shanxi; Baoxing, Sichuan; Heqing, Dali). To ensure the data are fully comprehensive, we have recorded all measurable traits, such as sepals and petals, in each population.

An Olympus SZX16 dissecting microscope was used for observations and an Olympus DP72 cooled digital colour camera was used to photograph leaves and staminodes. TpsDig software: Version 2.17 (http://life.bio.sunysb.edu/ee/rohlf/software.html) was used to measure the lengths and widths of leaves, thus improving the accuracy of experimental results.

Overlapping or non-overlapping standard deviations can be used to determine if there is continuous or discontinuous variation of a quantitative trait between two or more groups, thus providing evidence for species differentiation or merging. This approach was applied to the aspect ratio of leaves (the width of leaves / the length of leaves) and the depth of staminode lobes (the length of central lobes / the length of staminodes) (Fig. 4). A total of 100 populations from 11 provinces were studied. To ensure the universality and comprehensive nature of the data, we measured all available individuals.

Figure 4. 

Measurements of leaves and staminodes: W: the width of leaves; L: the length of leaves / staminodes; L1: the length of central lobes.

The morphological observations showed that the depth of staminode lobes varies considerably in the sampled populations, ranging from shallowly 3-lobed to 3-lobed for ca. 2/3 of their length (Fig. 5). Furthermore, the standard deviation analysis of the depth of staminode lobes also showed statistical continuity amongst populations of P. brevistyla, P. delavayi and P. leptophylla. As shown in Fig. 6, the variation of the depth of staminodes in the specimens collected from Baoji County, Heqing County and Baoxing County, where the types of P. brevistyla, P. delavayi and P. leptophylla were collected, is within the variation range of the population of P. brevistyla. There is also significant overlap of the depth of staminode lobes amongst the three species. Thus, P. brevistyla, P. delavayi and P. leptophylla show no differences in the depth of staminode lobes.

Figure 5. 

Variation of staminodes: a, b represent variation of the depth of staminode lobes in population of Kangding, Sichuan c represents variation of the depth of staminode lobes in population of Baoxing, Sichuan (the type locality of Parnassia leptophylla) d represents variation of the depth of staminode lobes in population of Cangshan, Dali (the type locality of P. delavayi) e represents variation of the depth of staminode lobes in population of Mabian, Sichuan. Scale bar: 1 mm.

Figure 6. 

The standard deviation analysis of the depth of staminode lobes.

The shape of basal leaves has also been considered to be a distinguishing characteristic in previous treatments (Ku 1987, 1995). However, we found that the shape of basal leaves varies greatly within populations. Based on field experience, we found that their leaf shape varies dramatically, not only amongst different regions, but also in the same population. Shapes range from reniform, broadly cordate, to cordate and ovate-cordate. As shown in Fig. 7, the variation of the blade aspect ratio in the specimens collected from type localities of P. brevistyla, P. delavayi and P. leptophylla is within the variation range of P. brevistyla populations. Furthermore, a significant overlap was found for the blade aspect ratio amongst these three species with respect to the observed ranges as well as the standard deviations (Fig. 7). Based on this continuous variation, it is difficult to find a dividing line to distinguish different leaf shapes.

In summary, P. brevistyla, P. delavayi and P. leptophylla show no differences in the shape of blades and the depth of staminode lobes. Thus, we treated P. brevistyla and P. leptophylla as new synonyms of P. delavayi.

Figure 7. 

The standard deviation analysis of the blade aspect ratio.

Hong (2016) proposed a new operational species concept called the morphological-biological species concept which emphasises that two or more qualitative traits or statistically discontinuous quantitative characters should exist between two species (Liu 2016). We applied this species concept in this study.

Taxonomically delimited species can be regarded as an assemblage of populations or individuals that have common morphological features and show distinct morphological discontinuity with other such assemblages in a number of characteristics (Singh 2004). That is, individuals of a given species show continuous variation in some characteristics. However, due to limitations of herbarium material, researchers can only observe a certain state of an individual within a population, rather than its variation within populations. Therefore, field investigation and random sampling within populations are important components of taxonomic studies (Xu et al. 2017). In this study, through both random sampling and statistical analysis within populations, the variability of morphological characteristics used in previous taxonomic treatments were analysed and reconsidered to avoid taxonomic artifacts that may be caused by undersampling.

We determined that both major characteristics that were previously used to distinguish P. brevistyla, P. delavayi and P. leptophylla, i.e. the shape of blades and the depth of staminode lobes, vary remarkably and continuously. Therefore, these characteristics should not be used to justify the recognition of three species. Furthermore, based on our examination of herbarium specimens as well as field observations, we found no other distinct characteristics to separate P. brevistyla, P. delavayi and P. leptophylla at the species level.

Furthermore, the distributions of these three species are continuous and overlap. Therefore, since these three species lack clear morphologic distinction and overlap in their distribution range, we hereby reduce P. brevistyla and P. leptophylla to synonyms of P. delavayi.

We are grateful to Ce Shang, Biao Xiong and Lei Wang for their valuable comments on the manuscript. We thank the curators of BJFC, CDBI, IMC, KUN, QFNU, P, PE, SM, SZ and WCNU for allowing us to examine specimens or use their images of specimens. This work was supported by the National Natural Science Foundation of China (grant no. 31110103911 & J1310002) and the Science & Technology Basic Work (grant no. 2017FY100100).