Taxonomic revision of Sageretia (Rhamnaceae) from China I: identities of S. lucida, S. thea var. cordiformis and S. yunlongensis, with the description of a new species S. ellipsoidea

Abstract A taxonomic revision of Sageretia lucida, S. thea var. cordiformis and S. yunlongensis in China is presented. Sageretia lucida is revised in terms of morphological characters (habit, branchlet color, phyllotaxis and rachis length), distribution, habitat, and phenology; S. thea var. cordiformis is raised to S. cordiformis; and S. yunlongensis is excluded from the genus Sageretia and reduced to the synonym of Rhamnus nigricans. Furthermore, a new species, S. ellipsoidea, is erected based on the paratype collections of S. lucida. The new species morphologically differs from S. lucida in having reddish brown branchlets, opposite or subopposite phyllotaxis, shorter rachises, and flowering in spring or early summer. S. ellipsoidea is factually closest to S. hamosa as they share similar woody-vine habit and larger fruit size, and fruiting in winter, whereas the former can be easily recognized based on its smaller leaf blades, fewer lateral veins, shorter rachises, and ellipsoidal or elliptic-ovoid fruits.


Introduction
Sageretia Brongn., the mock buckthorn genus of Rhamnaceae, contains ca. 35 species (Chen and Schirarend 2007). The genus shows a pan-tropical distribution with most members inhabiting subtropical and tropical Asia, and a few in northeastern Africa and tropical America (Mabberley 2008;Yang et al. 2019). Generally, Sageretia species are shrubs or woody vines, and usually thrive in disturbed habitats which have poorly developed soils. Many members of the genus have branchlets terminating in woody thorns as a defense against herbivores, and some of them such as S. gracilis Drumm. & Sprague and S. thea (Osbeck) Johnst. are popular in bonsai gardening. Besides, the drupes of several species are edible, and the leaves are flavonoid-rich and potential substitutes for tea (Chen and Schirarend 2007;Chung et al. 2009;Hyun et al. 2015).
According to Flora of China (FOC;Chen and Schirarend 2007), a total of 19 species and 3 varieties of Sageretia are found in the regions south of the Qinling Mountains and the Huai River. Only one species, S. paucicostata Maxim., is extensively distributed northward to the Yinshan Mountains. However, two species and one variety are included in a provincial flora but absent in FOC, including S. cordifolia Tardieu in Flora Yunnanica (Fan 2006), S. filiformis (Roth) G.Don in Flora Xizangica (Chen and Zhou 1986) and Flora Yunnanica, and S. thea var. taiwaniana (Masam.) Y.C.Liu & C.M.Wang in Flora of Taiwan (Liu et al. 1993). Moreover, three species (S. gongshanensis G.S.Fan & L.L.Deng, S. latifolia Hand.-Mazz. and S. yunlongensis G.S.Fan & L.L.Deng) published earlier than FOC (Handel-Mazzetti 1933;Fan andDeng 1995, 1997), have not been included in FOC. Hence, together with the recently published S. liuzhouensis Yi Yang & H.Sun (Yang et al. 2017), 25 species and 4 varieties (19 endemic species and 3 varieties) have been recorded in China to date.
We have been studying the taxonomy, molecular phylogeny and biogeography of Sageretia since 2014, especially the members in China. A new species, the tropical Asian origin, and three strongly supported clades matching morphological and distributional divergences, had been reported in Sageretia in our previous studies (Yang et al. 2017(Yang et al. , 2019. In this paper, we present several taxonomic problems in the genus and conduct corresponding revisions. In the process of protologue collation and specimen examination, S. lucida Merr., S. thea var. cordiformis Y.L.Chen & P.K.Chou and S. yunlongensis were found taxonomically problematic. Specifically, the paratypes of S. lucida distinctly differ from the type and actually represent a different species; S. thea var. cordiformis obviously diverges from S. thea var. thea in morphology and molecular phylogeny, and should be raised to a species; S. yunlongensis should be categorized in the genus Rhamnus L. rather than Sageretia. Thus, we here clarify the identities of these and related species.

Rhamnus nigricans
Phenology. Flowering from May to July; ripe fruits from October to December. Distribution and habitat. The species is distributed in southwestern China (Yunnan; Fig. 2). It grows in thickets on dry slope at elevation from 1300 to 2000 m.  Note. Although the genera Rhamnus and Sageretia are similar in morphology, they distinctly differ in characters of inflorescence (fascicled, cymose racemes, or cymose panicles in Rhamnus vs. spikes or spicate panicles in Sageretia) and fruits (basally persistent discoid calyx tube in Rhamnus vs. persistent reflexed calyx or remaining inconspicuous disk in Sageretia). The type collection of S. yunlongensis, Expedition Team 161, has branched cymose panicles and fruits basally covered with discoid calyx tube, suggesting it belongs to Rhamnus rather than Sageretia. In fact, S. yunlongensis (Fig. 1B) extremely resembles R. nigricans (Fig. 1A), they share similar habit, indumentum, leaf blade shape and size, inflorescence, and fruit, and highly overlapped distribution. Thus S. yunlongensis is herein reduced to a synonym of R. nigricans.
Phenology. Flowering in September; ripe fruits from December to January of the following year.
Distribution and habitat. The species is distributed in China (southwestern Yunnan; Fig. 2). It grows in thickets on tropical limestone mountains at elevation from 700 to 1100 m.
Etymology. This species is named for its ellipsoidal or elliptic-ovoid drupes which are different from other Sageretia species (subglobose or globose).
Distribution and habitat. The species is currently found in southern China (Fujian, Guangdong, Guangxi, Hainan; Fig. 2), and probably in northeastern Vietnam. It grows in moist forests along streams on granite mountains below 1200 m.
Note. When he erected the species Sageretia lucida, Merrill (1931) cited four collections, including W.T. Tsang & K.C. Wong 14340, 14579, 14584 and 15121, of which 15121 was selected as type and the other three collections were automatically treated as paratypes. However, the four numbers above belong to herbarium numbers which are ineffective nowadays, and the corresponding field numbers are W.T. Tsang & K.C. Wong 2479, 2718, 2723 and 3260, respectively. Moreover, another problem is that the paratype collections (2479, 2718 and 2723) factually represent an undescribed Sageretia species distinctly differing from S. lucida (3260) based on geological and morphological evidences. Among the four collections of S. lucida, three paratype collections were all collected from "Wan Tong Shan" (Wentang Shan) and type collection from "Chung Tung" (Zhongdong Village, about 10 km apart to the Wentang Shan). Based on field investigations, we find that Wentang Shan has granite landform while Zhongdong Village limestone landform. Furthermore, the three paratype collections are morphologically identical, but noticeably different from the type collection in terms of branchlet color (reddish brown in paratype collections vs. gray to dark gray in type collection), phyllotaxis (opposite or subopposite vs. alternate), rachis length (1-3 (-10) cm vs. 5-10 cm), and phenology (blooming in spring or early summer vs. in autumn) (seen in Table 2). Consequently, the species represented by the paratype collections of S. lucida is erected as a new species, namely S. ellipsoidea Yi Yang, H.Sun & H.Peng.
Besides, the samples of "Sageretia lucida" in Yang et al. (2019) factually also belong to S. ellipsoidea. According to Yang et al. (2019), the new species is sister to S. hamosa and they form an early diverging clade. In morphology and phenology, the new species with S. ellipsoidea rather than the true S. lucida in their study. Nonetheless, we have limited knowledge on S. lucida so far because of the lack of field collections, and so know nothing about the fruits. In order to get more information on the species, we conducted investigations at the type location of S. lucida (Zhongdong Village) during early summer in 2016 and autumn in 2020, respectively, but failed to find any individuals. Consequently, we suggest to suspend the synonymization of S. lucida to S. henryi until more evidence has been obtained.