Field investigations on the germplasm resource and morphological observations have been conducted from 2012 to 2016. Field work was done in Hubei, Shanxi, Chongqing and Sichuan Province, China. A total of 120 individuals (30 individuals per population) from four populations of three spurless species, E. ecalcaratum (two populations), E. platypetalum and E. campanulatum, were collected from Sichuan and Shanxi Provinces (Table 1). All populations were investigated and collected during the flowering, as the floral properties are significant for the taxonomy of Epimedium species. To capture variation within populations, 30 individuals per population were observed and sampled. Quantitative measurements on rhizome diameter, height of flowering stem, length of inflorescence, number of flowers, and length and width of the middle leaflet were recorded for each individual. The average data were processed using SPSS 19.0 software. Concurrently, the folloeing discrete morphological characters were observed: the rhizome; pedicel, petiole, underside of leaflet hair characteristics; shape and number of leaflets; number of stem-leaves; inflorescence; leaflets and flowers; shape and color of inner sepals; and shape of petals.

All 120 individuals of the three species were transplanted at the Jiangxi University of Traditional Chinese Medicine, China. Herbarium specimens were examined from the following herbaria: Chinese Academy of Medical Sciences, Peking Union Medical College Institute of Medicinal Plant Development (IMD); Institute of Botany, Chinese Academy of Sciences (PE); Chongqing Academy of Chinese Materia Medica (SM); Virtual Museum System (HX); Nanjing University (N); and Institute of Botany, Jiangsu Province and Chinese Academy of Sciences (NAS).

Based on field investigations and herbarium specimens, E. ecalcaratum, E. platypetalum and E. campanulatum were stenochoric species (Fig. 1). Epimedium platypetalum has previously been collected in Sichuan; however, it was not observed during our field investigations, likely due to habitat destruction. Therefore, the populations that can be collected were very limited.

Figure 1. 

Geographic distribution (black symbols) and sampling populations (white symbols) of E. ecalcaratum (square), E. platypetalum (circle), and E. campanulatum (triangle).

Quantitative morphological data from E. ecalcaratum, E. platypetalum, and E. campanulatum are presented in Table 2. Among the three close allies, E. campanulatum was easily identified by its long flowering stem and inflorescence, the largest number of flowers, and the stoutest rhizome. Epimedium ecalcaratum and E. platypetalum were very similar in terms of quantitative traits, but E. platypetalum has fewer flowers (2–8 flowers per individual; population mean of four flowers per individual). The main quantitative characters of the two populations of E. ecalcaratum were similar, only the length of the flowering stem was found having a slight difference.

The main discrete morphological characters of E. ecalcaratum, E. platypetalum and E. campanulatum are presented in Table 3. The three species all had glandular hairy inflorescence and pedicels; flat, spurless, yellow, pendulous flowers; and obovate petals. Although having much in common with its close allies, E. campanulatum differed by having compound inflorescences and cup-shaped flowers, while E. ecalcaratum differed because of the slightly saccate base of the petals, creating a slightly shouldered flower base (Fig. 2). Among the three species, the most diverse characters are the number of leaflets, the number of stem-leaves, and the arrangement of leaves on the stem (Table 3). The rhizome also presented a clear differentiation among the three species.

The protologue (Zhong 1991) and the subsequent description in Flora of China (Ying 2001) both described the compact rhizome of E. ecalcaratum. We re-examined the holotype and conducted fieldwork in its type locality. However, individuals of E. ecalcaratum in the field all had long creeping rhizomes, slender nodes with numerous fibrous roots, 1–3 mm in diameter, and internodes sometimes to 30 cm. In the genus Epimedium, this situation may not be rare. The form of the rhizome, specifically the degree of elongation and thickness, and also the average size of the terminal winter-bud, is constant for each species, and sometimes offers contrasts of taxonomic value (Stearn 2002). But Stearn (1997) pointed out that the different rhizome forms among some Epimedium species can sometimes be very evident, sometimes more subtle. For example, the major difference between E. leptorrhizum Stearn and E. brachyrrhizum Stearn was that the former had a very slender elongated rhizome while the latter bore a more compact clump-forming rhizome. However, examination of a series of E. leptorrhizum specimens showed that its rhizome was often slender and long-creeping but occasionally thicker and compact (Zhang et al. 2015b). In addition, the protologue of E. lishihchenii Stearn differs from E. franchetii Stearn was in having a long-creeping rhizome (Stearn 1997). However, our field observation based on a population found that E. lishihchenii 20% of individuals had compact rhizomes (Liu et al. 2016).

Due to the slender elongated rhizome and small broadly ovate or almost orbicular leaflets, Guo et al. (1993) described a variety, E. platypetalum var. tenuis B. L. Guo et P. G. Hsiao. Stearn (1995) assessed E. platypetalum var. tenuis and found it differed from E. platypetalum bacause of its long-spurred flowers. Then, he treated E. platypetalum var. tenuis as a synonym of E. pauciflorum K. C. Yen, a new species that was published after the description of E. platypetalum var. tenuis (Yen 1994; Stearn 1995). In a more recent study, Ying et al. (2011) still treated E. platypetalum var. tenuis as a synonym of E. platypetalum. We re-examined the specimens of E. platypetalum var. tenuis, and found the taxon had obviously long-spurred petals (1.7 cm) and long inner sepals (1.4 cm) that were completely different with E. platypetalum. In addition, E. platypetalum var. tenuis has sympatric distribution with E. pauciflorum. Therefore, we agree with Stearn (1995) that E. platypetalum var. tenuis should be revised as a synonym of E. pauciflorum.

The number of stem-leaves was believed to be stable within a species and important for taxonomy, and three informal groups have been divided by the normal number of stem-leaves (Stearn 2002). Stem-leaves, however, are not so unvarying as initially supposed. Stearn (2002), Zhang et al. (2015), and our field observation have recognized and recorded some variation on the number of leaves. For example, usually two opposite or occasionally three whorled leaves were observed in E. sagittatum (Sieb. et Zucc.) Maxim., E. acuminatum Franch., E. myrianthum Stearn and E. franchetii, and one leaf or two leaves in E. epsteinii Stearn, E. flavum Stearn, E. leptorrhizum and E. pauciflorum.

It is significant that the comparatively unstable species occur in western China, where the genus is best represented and where its evolution may still be proceeding (Stearn 2002). Population is the basic unit of evolution (Chen 2016) and the most important unit to study the formation of species (Chen and Wang 1986; Nooteboom 1992; Hong 2016). Thus, morphological differences recorded among individuals in a population should not be ignored and may be more obvious than populations in sometimes (Yang 1991; Jonas et al. 1999). Previous studies on the taxonomy of the genus Epimedium were almost always based on limited samples, or several individuals cultivated abroad (Ying 2001). Our investigations based on populations in their native habitat found that the number of leaves and the habit of the flowering stem presented abundant variation (Fig. 3).

The protologue for E. ecalcaratum described that its stem-leaves are usually opposite with two trifoliolate leaves, occasionally alternate with two trifoliolate leaves or three trifoliolate leaves (Zhong 1991). Our investigation showed that it usually had two alternate or opposite trifoliolate leaves, sometimes three alternate trifoliolate leaves, occasionally two opposite 5-foliolate leaves, or rarely three or four whorled unifoliolate, trifoliolate and/or 5-foliolate leaves (Fig. 3A–G). In the protologue of E. campanulatum (Ogisu 1996), the species was described with one leaf or two usually alternate, rarely opposite leaves, but our investigation showed that it usually had two alternate (mostly two trifoliolate, and occasionally one trifoliolate and one simple) or opposite leaves, sometimes three alternate leaves, and one trifoliolate leaf was also occasionally observed (Fig. 3M–Q). Our study clearly shows that the number and insertion of the leaves and the number of leaflets varies in this species. Upon extensive specimen examination, Zhang et al. (2011, 2015b) observed that the leaves of E. simplicifolium T. S. Ying were mainly unifoliolate, occasionally trifoliolate, and the leaves of E. acuminatum Franch. may be unifoliolate. Subsequently, E. simplicifolium was synonymized with E. acuminatum (Zhang et al. 2011). Hence, as a taxonomist, it is important to study as many collections as possible (Xu 1998).

Although having much in common with E. platypetalum and E. campanulatum, E. ecalcaratum differs in having slightly a saccate petal base, creating a slightly shouldered base to the flower (Stearn 2002), which is in agreement with our field observations (fig. 2A, B, D). According to Stearn’s research, the character could be regarded as moving towards development of a nectar-producing spur, and he published a photo to show the petals are typically without a spur, but may have varying degrees of small spurs (Stearn 2002: 53, fig. 18). This may indicate that E. ecalcaratum represented a transitional stage in floral evolution from series Campanulatae (spurless) to series Davidiance (spur with basal lamina). Spur variations have not been observed in the present study, and are not supported by specimens or other literature.