Taxonomic revision of the genus Glochidion (Phyllanthaceae) in Taiwan, China

Abstract A comprehensive taxonomic revision of the genus Glochidion J.R. Forst. & G. Forst. from Taiwan in China was carried out based on the examination of herbarium specimens and filed investigations. Eight species and three varieties are recognized, including a new species endemic to Taiwan, G. lanyuense Gang Yao & S.X. Luo, which is described and illustrated. Three names, viz. G. chademenosocarpum Hayata, G. kusukusense Hayata, and G. ovalifolium F.Y. Lu & Y.S. Hsu, are reduced to the new synonyms of G. rubrum Blume, G. lanceolatum Hayata, and G. ellipticum Wight, respectively. Two names, viz. G. lanceolatum Hayata and G. suishaense Hayata, are lectotypified here. A key to the Glochidion species in Taiwan is provided.


Introduction
Glochidion J.R. Forst. & G. Forst. is the second largest genus within the tribe Phyllantheae Dumort. (Phyllanthaceae Martinov) (Govaerts et al. 2000;Webster 2014;Duocet Group 2016 onwards). It is represented by more than 300 species of shrubs or trees distributed primarily in the Indo-Pacific, east to southeast Polynesia and south Hayata, G. kusukusense, and G. suishaense Hayata] have long been treated as dubious taxa because relevant specimens, especially the types, were unavailable (Kanehira 1936;Keng 1955;Deng 1992;Deng and Wang 1993;Hsu et al. 2006). Although the latter three species were accepted in Flora Reipublicae Popularis Sinicae (Li 1994) and Flora of China (Li and Gilbert 2008), their morphological descriptions were merely derived from their protologues without further observations. Hsu et al. (2006) accepted the species G. kusukusense and synonymized the name G. suishaense under G. rubrum, but types of the two species were still not referred to in their study.
In our taxonomic study of the genus Glochidion, types of aforementioned enigmatic taxa were found in the herbaria A (G. assamicum var. magnicapsulum, Fig. 1A) and TI (G. chademenosocarpum, Fig. 1D; G. kusukusense, Fig. 1B; G. suishaense, Fig. 1F), and a collection of the genus from Lanyu island of Taiwan, China, was found to be very different from all of the other Glochidion species recorded from Taiwan and adjacent regions. Thus, a comprehensive taxonomic revision of Glochidion in Taiwan was conducted in this study.

Material and methods
Specimens of Glochidion deposited in the herbaria A, HAST, K, KUN, IBSC, LINN, MA, NAS, NCAI, NY, P, PE, PH, TAI and TI, were studied carefully in the present study. Field investigations of Taiwanese Phyllantheae species were also conducted from 2015 to 2019. Additionally, most materials of Taiwanese Glochidion, which were obtained by Dr. A. Kawakita from Kyoto University, Japan, in his recent field studies of the co-evolutionary system involving Glochidion plants and Epicephala moths (Kawakita et al. 2004;Kato and Kawakita 2017), were generously provided for the present study. Morphology of leaves, styles and capsules, as well as the number of ovaries and stamens, were all studied carefully. Herbarium abbreviations cited here are based on the Index Herbarium of Thiers (2013 continuously updated).

results
In total, over 800 specimens were examined in the present study. Morphological studies based on the careful examination of herbarium specimens and extensive filed investigations revealed that eight species and three varieties of Glochidion should be recognized in Taiwan, China, viz. G. acuminatum var. acuminatum, G. ellipticum Wight, G. lanceolatum, G. lanyuense Gang Yao & S.X. Luo, G. philippicum, G. puber, G. rubrum, G. zeylanicum var. zeylanicum, G. zeylanicum var. tomentosum Trim., among which the species G. lanyuense is new to science. Additionally, the three names G. chademenosocarpum, G. kusukusense and G. ovalifolium should be reduced to the new synonyms of G. rubrum, G. lanceolatum and G. ellipticum, respectively. Because other species were morphologically described in detail by previous authors (e.g. Hsieh 1977;Deng and Wang 1993;Hsu et al. 2006;Li and Gilbert 2008), we only provide a morphological description for the new species G. lanyuense, but a key to all of the Glochidion species in Taiwan is presented. siamense Airy Shaw. The species is represented in Taiwan by the typical variety, and another variety is distributed in Thailand and Yunnan province of China (Li and Gilbert 2008;Yao and Zhang 2015b). A detailed morphological comparison between the two taxa can be referenced in Yao and Zhang (2015b). The typical variety can be distinguished from all other Taiwanese Glochidion species by its small (5-7 mm in diameter) and deeply 6-or 8-grooved capsules ( Figure 2C).

Glochidion ellipticum
Taxonomic discussion. Croizat and Hara (1940) described the variety G. assamicum var. magnicapsulum from Taiwan, and considered that it differed from the typical variety G. assamicum var. assamicum by its large female flowers (ca. 2.5 mm in diameter) and the pubescent and large capsules (8-10 mm in diameter). However, this variety had long been treated as a dubious taxa, or even not referred to in latter taxonomic treatments since its publication (e.g. Keng 1955;Hsieh 1977;Deng and Wang 1993;Hsu et al. 2006), until it was reduced to be a synonym of G. ellipticum (G. assamicum was cited as one of its synonyms) in Flora of China (Li and Gilbert 2008). After observing the types and many non-type specimens of G. ellipticum, it was found that the species showed much variation in morphology, such as its leaves are elliptic, lanceolate, oblong or ovoid in shapes; ovary usually 3-4-locular, and rarely 5-locular; fruits subglabrous or sparsely pubescent, usually 6-8 mm in diameter, and sometimes could be up to 10 mm in diameter. So the treatment of Li and Gilbert (2008) is accepted and the distribution of G. ellipticum in Taiwan is confirmed here. More Taiwanese specimens of G. ellipticum were also found and examined in the present study.
Glochidion ovalifolium was described from Chiayi, Taiwan, China, and it was suggested to be similar to G. lanceolatum in morphology, but differs in having hairy ovaries and fruits (Lu and Hsu 2003). However, results from checking the types of G. ovalifolium (Figure 3) revealed that the species is actually conspecific with G. ellipticum ( Figure 1A), which is very different from G. lanceolatum by its female flowers and fruits cluster in axillary (vs. usually pedunculated supra-axillary cymes), ovary and fruits sub-glabrous or sparsely pubescent (vs. glabrous), stamens 3 (vs. 4-6). The pedunculated supra-axillary cymes of G. ovalifolium as described in its protologue, and observed in its line drawing provided in Lu and Hsu (2003), are incongruous with its types (Figure 3). Hence, based on a careful study of its type and non-type specimens, G. ovalifolium is here reduced to a new synonym of G. ellipticum.
Representative specimens examined. China. Taiwan Hsu & J.C. Liaw 198,199,201,202,203,204,207,208,209 (NCAI)  Taxonomic discussion. The morphological description of G. kusukusense provided by Hayata (1920) is brief and short, and it is in accordance with its type that has only a short branch and several leaves ( Figure 1B). In the protologue, G. kusukusense was compared morphologically with G. wrightii Benth, a species widely distributed in southern and southwestern China. After observing the type of G. kusukusense deposited in herbarium TI ( Figure 1B), we found that its glabrous habit and lanceolate leaves were identical to that of the species G. lanceolatum ( Figure 1C), which is widely distributed in Taiwan. Thus, we considered that G. kusukusense is conspecific with G. lanceolatum ( Figure 1C) and reduced it to be a new synonym of the latter.
In Hsu et al. (2006) study, some specimens collected from Nantou and Taoyuan of Taiwan (out of the type locality of G. kusukusense) cannot be identified as Glochidion species usually known to Taiwan, while their lanceolate leaves and glabrous pedicel of male flowers observed seemed to be consistent with the diagnostic traits of G. kusukusense, when compared with another dubious species G. chademenosocarpum also described in Taiwan, as suggested by Li (1994) based on the protologues of the two species. Thus the distribution of G. kusukusense in Taiwan was accepted by Hsu et al. (2006), although the type of the species was unavailable in their study. However, the detailed morphological description and line drawing of G. kusukusense provided in Hsu et al. (2006) are very different from the type of the species but well identical with G. ellipticum, a species distributed in Taiwan but omitted in most literature of Taiwanese Glochidion, including Hsu et al. (2006). The result from rechecking the specimens cited as G. kusukusense in Hsu et al. (2006) further confirmed our conclusion. More specimens of G. ellipticum collected from Chiayi, Kaosiung, Keelung, Pingtung, Nantou, Taichung and Tainan of Taiwan were also found and studied in the current study (see 'Representative specimens examined' under the species G. ellipticum).
Distribution and habitat. The species is known only from its type locality, Lanyu island of Taiwan, China. It grows on roadsides of broadleaf forest at low altitude. Etymology. Glochidion lanyuense is named after its type locality, Lanyu island of Taiwan.
Taxonomic discussion. Glochidion lanyuense is quite different from all of the other Glochidion species recorded from China, Japan, and the Philippines by its special characters of female flowers. It is similar to G. lanceolatum in habit, but differs (Table 1) in its female flowers which are usually solitary or rarely two in axillary ( Figure 5J, K), pedicel of female flowers and ovaries are usually densely strigose ( Figure 5B, D, F), style connate into a short cylindrical column ( Figure 5D, H), stamens 3, capsules ca. 10 mm in diameter and 5-6-grooved ( Figure 5M). In contrast, the species G. lanceolatum has multiple female flowers (usually 6-15) in supra-axillary cymes ( Figures 2F, 5I) or rarely axillary, pedicel of female flowers and ovaries are glabrous ( Figure 5A, C, E), styles sub-conical Figure 5C, G), stamens 4-6, fruits 6-7 mm in diameter and shallowly 4-or 6-grooved or obscurely grooved ( Figures 2G, 5L). Additionally, as revealed in our previous palynological study (Yao and Zhang 2016), pollen grains of G. lanyuense (recorded as Glochidion sp.3 in Yao and Zhang 2016) were much larger in size (P = 21.01 μm, E = 21.12 μm), 4-colporate in aperture system, and rugulate in exine ornamentation. While pollen grains of G. lanceolatum were smaller in size (P = 17.25 μm, E = 15.47 μm), 3-4-colporate, and regular reticulate in exine ornamentation, all of these characters are also consistent with the observation of Deng (1992). The rugulate ornamentation observed in pollen grains of the new species seems to be very different from those of other Taiwanese Glochidion species observed previously (Deng 1992;Yao and Zhang 2016). Distribution and habitat. Glochidion philippicum is distributed in China (only in Taiwan), Malaysia, and the Philippines. The species usually occurs beside a gravelly road, forest margins, and roadsides, at low and medium altitudes. In Taiwan, the species is widely distributed from Ilan, Taichung, Changhua and Chiayi, to Hualian, Kaohsiung, Pingtung, Tainan and Taitung.

Glochidion rubrum
Taxonomic discussion. Hayata (1920) described G. chademenosocarpum based on one of his collections (B. Hayata s.n., TI, photo!; Figure 1D) from Taiwan, and he suggested that the species differed from G. rubrum (recorded as G. fortunei in his study) by its much more densely clustered and sessile female flowers. After checking the protologue and observing the type of G. chademenosocarpum (Figure 1D), it was concluded that the species is well conspecific with G. rubrum ( Figure 1E) and should be treated as a new synonym of the latter. Based on morphological description (Hayata 1920) as well as our observation of the type, we found that the type of G. chademenosocarpum might represent an unripe flower branch of G. rubrum because the morphology of branch, leaves, ovaries and styles observed are all identical with those of G. rubrum, except the ovaries and styles are smaller in size compared with those of the ripe female flowers described by Hsu et al. (2006) as well as those observed in living plants. Additionally, most male flowers observed from the type of G. chademenosocarpum also seem to be unripe because sepals of most male flowers were unopen ( Figure 1D).
The taxonomic history of G. suishaense was similar to that of G. chademenosocarpum and G. kusukusense, except Hsu et al. (2006) treated it as a synonym of G. rubrum based on its protologue. In the present study, the result from observing the type of G. suishaense ( Figure 1F) further confirmed its taxonomic status.
Morphologically, G. rubrum could be distinguished from all other Taiwanese Glochidion species by its styles which are cylindrical in shape and 1-3 mm in length ( Figure 2M).  Figure 2Q-S Distribution and habitat. The typical variety G. zeylanicum var. zeylanicum is widely distributed from India, Sri Lanka, through Myanmar, Thailand, Vietnam, to China, south Japan, Indonesia, and the Pacific islands. In China, it occurs widely from the southwest of the mainland to Taiwan island. It usually grows in sparse forests, margins of woods, humid valleys, scrub on stream banks, roadsides, and at low and medium altitudes. In Taiwan, the variety is widely distributed from Hsinchu, Ilan, Taoyuan and Taipei, to Nantou, Taichung and Pingtung.

a. var. zeylanicum
Taxonomic discussion. The typical variety G. zeylanicum var. zeylanicum is similar to G. lanceolatum in habit, and morphological differences between them have been discussed under the latter species.
Taxonomic discussion. Morphologically, the variety G. zeylanicum var. tomentosum differs from the typical variety G. zeylanicum var. zeylanicum by its hairy habit. For the hairy taxon, the name G. zeylanicum var. tomentosum was accepted by several authors in their treatment of Taiwanese Glochidion (Hsieh 1977;Deng and Wang 1993) and further appreciated recently (TPL 2013 continuously updated, Yao and Zhang 2015a), but in some other treatments the name G. hirsutum (Roxb.) Voigt was adopted (Li 1994;Hsu et al. 2006;Li and Gilbert 2008). In our taxonomic study of the genus Glochidion, it was found that the hairy variety ( Figure 2T