China is one of the richest countries in plant biodiversity, harbouring more than 30,000 seed plants (Liu et al., 2003; López-Pujol et al., 2006). Such “mega-biodiversity” (Mittermeier et al., 1997; Liu et al., 2003; Tang et al., 2006) has been often attributed to the highly complicated ecological and evolutionary history of China and East Asia. For example, China has complex topographies and almost all types of biomes from tropical to boreal, which provides abundant suitable habitats for various plants (Ying, 2001). China is also one of major centres of origin and diversification for vascular plants (Axelrod et al., 1996; Qian, 2002, and particularly the southern part of this vast country is the home for both relict and recently evolved plants (Ying & Zhang, 1984; López-Pujol et al., 2011).

In Southern China, three regions are widely recognized as major “biodiversity hotspots” and they are roughly equal to the country’s “centres of plant endemism” (Ying & Zhang, 1984; Ying et al., 1993; Ying, 1996; López-Pujol et al., 2011). All these three hotspots are located in mountain areas with large rivers (Fig. 1) but they differ significantly in plant composition and origin. The biodiversity hotspot at central China locates at the boundaries of Hubei, Hunan, and Guizhou provinces, and Chongqing City, including the Three Gorges region on the Yangtze River (Fig. 1). Most endemic plants in this region are palaeoendemics, i.e. the relict species that have survived from the Tertiary (Ying et al., 1993; López-Pujol et al., 2011). The Hengduan Mountains on the eastern fringe of the Tibetan Plateau are also widely recognized as a globally important biodiversity hotspot. It consists largely of recently evolved endemic species (neoendemics) that resulted from the uplift of the Himalayas and surrounding mountains (Chapman & Wang, 2002; Qian, 2002). The third hotspot spans the upper reaches of Zhujiang River (the largest river in Southern China), at the confluence of three Chinese provinces (Guizhou, Yunnan, and Guangxi). This biodiversity hotspot (hereafter UZJ) is dominated by limestone landscapes and river valleys (Figs. 1 and 2). UZJ is also distinctive for its nearly equal proportion of neoendemics and palaeoendemics (Ying et al., 1993; López-Pujol et al., 2011) and significantly more tropical species than the other two hotspots. Myers et al. (2000) included it as one part of the globally important biodiversity hotspot “Indo-Burma”. Compared to other hotspots, UZJ hotspot received less attention, and the evolutionary histories and maintenance mechanisms of its unusual biodiversity have not been satisfactorily resolved (Ying et al., 1993; Fang et al., 1995).

In this paper, I will use the distribution map of Hiptage, an Asia-endemic genus of the typically tropical family Malpighiaceae, with the aims of: (1) to determine the diversity and endemism centre(s) of the genus by calculating species richness and endemism for five different geographic regions; (2) to discuss possible determinants for distribution patterns of Hiptage from both environmental and biological factors by comparing widely-recognized hypothesis and updated studies on this genus; (3) to explore the underlying mechanism(s) of origin and maintenance of the tropical plant diversity in UZJ hotspot with an emphasis on the roles of river and associated environmental features.

Basic information of the genus Hiptage

Hiptage is one of the largest Old World genera of Malpighiaceae, with >25 species usually recognized (Anderson et al., 2006–). This genus is originated as a result of inter-continent long-distance dispersals from tropical America to Asia during the late Oligocene, ~29 Ma (Davis et al., 2002).

Hiptage species grows mainly as high-climbing, twining woody lianas in South Asia, Indo-China Peninsula, Indonesia, Philippines, and Southern China including Hainan and Taiwan islands (Chen & Funston, 2008; Ren et al., 2013). Jacobs (1955) also reported one endemic species in Fiji islands, which locates at Southern Pacific. Because of its three-winged peculiar samara, Hiptage is probably the Old World genus easiest to recognize. Its flowers are bilaterally symmetrical and basically white but the posterior petal is often yellow in its centre (Fig. 3A). The number and structure of sepal gland(s) represent diagnostic traits in the family (Jacobs, 1955; Anderson, 1990; Vogel, 1990; Srivastava, 1992; Chen & Funston, 2008). Normally only the posterior sepal bearing one large gland in Hiptage but in some species gland is missing or borne in pairs (Vogel, 1990; Anderson et al., 2006–; Chen & Funston, 2008). Stamens are 10, one of them much longer and bigger than the other nine. All anthers of Hiptage dehisce longitudinally (Ren et al., 2013), contrasting to poricidal anthers in most heteromorphic stamens. Ren et al. (2013) found in H. benghalensis (L.) Kurz, the most widespread species of the genus, that longitudinal anthers probably are an adaptation to its pollinator Apis dorsata, an Asia-endemic pollen-collecting honeybee (Fig. 3B).

Hiptage is also distinctive for having mirror-image flowers (Ren et al., 2013). Mirror-image flowers show a sexual polymorphism in which the style deflects either to the left (left-styled flower) or the right (right-styled flower) side of the floral axis, and the bigger stamen deflects to the opposite side of the style (Fig. 3A). Normally insects enter the flower between the style and the big stamen and consequently contact two sexual organs respectively with their left and right side of abdomens (Fig. 3B; Jesson & Barrett, 2002; Ren et al., 2013). Therefore, mirror-image flowers exhibit a highly specialized insect-pollinating mechanism for promoting cross-pollination between left- and right-styled flowers or reducing pollination between flowers on the same plant (Barrett et al., 2000; Jesson & Barrett, 2002; Ren et al., 2013).

Species identification

Published floras including Flora of China (Chen & Funston, 2008), Flora of Thailand (Sirirugsa, 1991), Flora of Malaysia (Jacobs, 1955), Flora of Bhutan (Grierson, 1991), and Flora of Taiwan (Chen, 1993) were used to identify Hiptage species and their geographic distributions. Hiptage species in India were mainly determined according to Srivastava (1992). All the described species in these floras were checked. The Malpighiaceae website http://herbarium.lsa.umich.edu/malpigh/ (Anderson et al., 2006–) was also used as a guideline. When there was any mismatching information about taxonomy and distribution, I referred to the newest floras. Finally, 29 species were recognized in this study (Appendix).

Distribution regions

According to geographic locations and landscape type, the distribution range of Hiptage was classified into five geographic regions: Islands (Philippines, Indonesia, Fiji, Andaman); South Asia (Pakistan, Sri Lanka, Bhutan, Nepal, Bengal, India); Indo-China Peninsula (including Malay Peninsula); upper reaches of Zhujiang River (UZJ); and Yunnan Province (except the area included into UZJ). Yunnan Province was analysed separately because this province is the major distribution centre of tropical plants in China (Zhu et al., 2003). The area occupied by the different regions varies greatly, ranging from a minimum of 21 × 10⁴ km² in UZJ to a maximum of 410 × 10⁴ km² in South Asia (Table 1).

Table 1.

Excepting the islands region, all the other four regions were characterized by large rivers (Fig. 1). Particularly UZJ is a region with numerous rivers and diverse limestone landscapes (karst) (Fig. 2). The mainstream of UZJ are Nanpanjiang River and Hongshuihe River (Fig. 2). UZJ also includes the upper reaches of the tributaries Youjiang and Zuojiang rivers (Fig. 2). These two rivers are very close to the mainstreams, with a nearest distance of ~20 km and ~10 km to Nanpanjiang and Hongshuihe rivers, respectively (Fig. 2), and their topography and habitats are quite the same to the mainstream (Ying et al., 1993; Fang et al., 1995; Hou et al., 2010).

Determination of diversity and endemism centres

The numbers of total species and endemic species were counted for each geographic region. To detect the diversity centre (distribution centre of species richness), area- and habitat-adjusted species density (SD) was calculated using the formula SD = NT/[ln(A) + ln(E)], where NT is the species richness (the number of total species of the region), A is the area of the region (km²), and E is the habitat heterogeneity estimated by the elevation range (the difference between the highest and the lowest elevation in the region) according to Tang et al. (2006). The data of area and habitat heterogeneity for each geographic region are log-transferred to minimize their impacts on the calculation of species density using a revised method described in Qian (1998) and Tang et al. (2006).

To determine the endemism centre(s), area- and habitat-adjusted endemism index (EI) was calculated using the formula EI = [NE/(NT − NE)]/[ln(A) + ln(E)], where NT is the number of total species, NE is the number of endemic species, A is the area of the region (km²), and E is the habitat heterogeneity.

To explore the correlation of species diversity and main geographic factors, the Pearson correlation coefficient between species diversity (i.e. number of total species, number of endemic species, endemic index) and three factors [region area, distance to the equator, and habitat heterogeneity (elevation range)] was calculated using SPSS software v17.0.

Based on the comprehensive surveys on all the described species, twenty-nine species of Hiptage were identified in this study (see Appendix). Two regions, Indo-China Peninsula and UZJ, were identified as the main diversity centres because they have the highest values of species density, both ≥0.65 (Table 1 and Appendix). There were 16 species in Indo-China Peninsula, in which 10 were endemic, mostly in Thailand at the south of the peninsula. In UZJ, six out the total seven Hiptage species were endemic (Table 1). These two regions were also recognized as endemism centres because both regions had an endemism index >0.1 (Fig. 4 and Table 1). UZJ was especially distinctive for its highest endemism, with a value of endemism index about five times of the second, Indo-China Peninsula (Fig. 4 and Table 1). This result made UZJ the distribution centre of endemic species of Hiptage. All the Hiptage species in UZJ have obviously reduced glands, with only two, one or no sepal gland (Appendix). Such floral trait is found to be a derived character in the family according to a complete generic phylogeny with both morphological and molecular data (Davis & Anderson, 2010).

The endemic species in UZJ were mostly distributed along the rivers and valleys (Fig. 2), with some species occurred at the nearby limestone hills. Three rare endemic species with only one recorded place, i.e. Hiptage luodianensis S. K. Chen, H. fraxinifolia F. N. Wei, and H. multiflora F. N. Wei were found in the valley of Hongshuihe, Zuojiang, and Yongjiang rivers, respectively (Fig. 2). The other three species with two or more recorded places, i.e. H. lanceolata Arènes, H. tianyangensis F. N. Wei, and H. minor Dunn were also distributed along the Hongshuihe and Youjiang valleys (Fig. 2).

Statistical analysis revealed that the distributions of species diversity, measured by total number of total species (species richness), number of endemic species and endemism index, were not correlated with the region area, the distance to the equator, or the elevation range (Table 2).

This study pointed out upper reaches of Zhujiang River in South China is not only the northern distribution edge of the tropical genus Hiptage but also its diversity and endemism centre. The correlation analyses found no associations of species diversity and endemism with geographic factors including region size, distance to the equator, and habitat heterogeneity (elevation range) (Tables 1 and 2). These facts suggest the formation of this diversity and endemism centre of Hiptage cannot be explained by large-scaled geographic factors and mainly is resulted from recent evolution caused by local topographic features or intrinsic traits of the genus, which I discuss in detail below.

Table 2.

Rivers and monsoon climate

Not only UZJ but also Indo-China peninsula (another centre of species diversity and endemism of Hiptage) are characterized by large rivers (Fig. 1). In Indo-China Peninsula, all the rivers such as Mekong River and Salween River are of north–south direction (Fig. 1). This flow direction can promote the northward spread of Hiptage since this tropical genus can only find suitable habitat along the riverside where is far away from typical tropical climate (Anderson et al., 2006–; Ren et al., 2013). The long-distance dispersal of Hiptage is mainly achieved via its three-winged fruits, which can help the fruits to disperse over oceans (Davis et al., 2002). UZJ is the most important diversity and endemism centre of Hiptage (Figs. 2 and 4, and Table 1) and contains numerous mountains and rivers with north-west–south-east direction (Figs. 1 and 2). These rivers are smaller than Mekong, Salween, and Red rivers but their valleys and slopes are also the home to many narrowly-endemic Hiptage species (Ren et al., 2013; Fig. 2).

More importantly, both UZJ and Indo-China Peninsula are close to the Tibetan Plateau and Himalaya (Fig. 1). The Himalayan uplift created tropical monsoon climate in its southern and eastern regions and the role of monsoon climate for species spread and speciation were already widely acknowledged in the nearby Hengduan Mountains (Fig. 1; Chapman & Wang 2002; López-Pujol et al., 2006). Monsoon climate in Indo-China Peninsula and nearby regions is characterized by strong northward wind in the summer (approximately from April to October), which is also rain season (Wu & Zhang, 1998). In UZJ, the mainstream Hongshuihe River and two main branches (Youjiang and Zuojiang rivers) are all north-west–south-east direction (Fig. 2). This flow direction can bring the monsoon climate northward and promote the spread of Hiptage along the river valleys and even to the north bank of Hongshuihe River (Fig. 2). This is why all the Hiptage species in UZJ are distributed along the rivers and, when the river (Hongshuihe and Nanpanjiang sections) changes the flow to west–east direction, no Hiptage species move northward anymore (Fig. 2).

However, a question remains: why the Red River and the upper reaches of Mekong and Salween rivers (Figs. 1 and 2) have not formed a diversity or endemism centre of the genus Hiptage, given that their north–south flow direction also have the potential for species spread? The main reason probably is that the valleys of Mekong and Salween rivers are so wide that the three-winged fruits of Hiptage can disperse over a very long distance along the valleys, resulting in strong and continuous gene flow between the lower and upper reaches. Most mountains in this region are also of north–south direction and in good connectivity, which further maintains gene flow connecting upper and middle or lower reaches of rivers. This is why the only one endemic species on the Salween River, H. yunnanensis Huang ex S. K. Chen, occurs in the extraordinarily complex Hengduan Mountains in the northwestern Yunnan Province (Fig. 1) far away from the middle and lower reaches of the river. The valley of Red River is by far much wider and shorter than Mekong and Salween rivers, and gene flow along the Red River is strong enough to maintain species connectivity. Therefore it is not surprising that there is no endemic species along Red River, even it is only ~60 km away from UZJ (Fig. 2).

Highly-fragmented limestone landscapes

Another reason for UZJ as a diversification centre is its vast fragmentary limestone landscapes. Southwestern China has the largest continuous limestone areas in the world (Clements et al., 2006; Hou et al., 2010), which includes the UZJ and the biodiversity hotspot at borders of Guizhou–Yunnan–Guangxi provinces (Hou et al., 2010). The limestone area here experienced severe erosion due to heavy rainfalls (Zhu et al., 2003; Hou et al., 2010) and is distinctive for its fengcong and fenglin, i.e. partially or completed isolated karst peaks (Fig. 2). These separate peaks, together with numerous slopes and pits with different orientations, may provide isolated microhabitats for plants (Fang et al., 1995; Zhu et al., 2003; Clements et al., 2006; Hou et al., 2010). Clements et al. (2006) attributed the high level of species richness and endemism in Southeast Asian karsts, including Southwestern China, to their high diversity of microhabitats and climatic conditions. This is particularly the case in UZJ and surrounding areas since the limestone landscapes here are highly fragmented and remained relatively stable during the last glacial period of the Pleistocene (Li, 1994; Hou et al., 2010), making this region not only a survival center for relict species, but also an important place for species diversification (Chapman & Wang, 2002; Qian, 2002; López-Pujol et al., 2011).

For Hiptage, many endemic species are found exclusively on limestone rocks in UZJ and Indo-China Peninsula (Sirirugsa, 1991). In UZJ, some species such as H. tianyangensis and H. multiflora occupy the tops or sunny slopes of hills, some species including H. minor, H. lanceolata and H. luodianensis can only grow at the bottom of river valleys, while other species such as H. fraxinifolia and H. benghalensis are found predominantly in dense forests or shrubs (Chen & Funston, 2008; M. X. Ren, pers. obs.). These local separations of species distributions suggest that habitat differentiation is quite possible between species.

UZJ is also the distribution and endemism centre for other tropical families including Gesneriaceae and Begoniaceae (Li & Wang, 2004; Wei et al., 2004; Hou et al., 2010). For example, many recently-evolved genus and species of Gesneriaceae such as Allocheilos W. T. Wang, Thamnocharis W. T. Wang, Tengia Chun, and Lagarosolen W. T. Wang are only found in UZJ and its neighboring areas (Lu et al., 1989; Wei et al., 2004). In Guangxi Province, Wei et al. (2004) found 16 out 38 genus of Gesneriaceae that were distributed exclusively in karst regions, while Lu et al. (1995) and Hou et al. (2010) even reported ~80% of the endemic genera are only distributed in limestone areas. Furthermore, most endemic species of Gesneriaceae are restricted to the highly-fragmented limestone areas between Youjiang and Zuojiang rivers (Fig. 2; Wei et al., 2004). These facts further suggest the evolutionary diversification of these tropical taxa in UZJ mainly resulted from fragmented limestone landscapes (Li, 1994; Fang et al., 1995; Wei et al., 2004).

Highly-specialized pollination mechanism

Hiptage is characterized with mirror-image flowers with heteromorphic stamens and longitudinal anthers (Fig. 3A; Ren et al., 2013). Mirror-image flowers are a sexual polymorphism in which the style deflects away from the floral axis either to the left (left-styled flower) or the right (right-styled flower) (Jesson & Barrett, 2002; Gao et al., 2006; Ren et al., 2013). Normally the pollinators for mirror-image flowers are large-bodied pollen-collecting honeybees such as Apis dorsata (Fig. 3B; Ren et al., 2013). This highly-specialized insect-pollination mechanism can facilitate cross-pollinations between left- and right-styled flowers through two deflected sexual organs touching left and right sides of the pollinator’s abdomen respectively (Jesson & Barrett, 2002; Ren et al., 2013). Therefore, the spatial separation of the deflected stigma and anther (herkogamy; Webb & Lloyd, 1986; Ren & Zhang, 2004) must be under selection to adapt to the pollinator’s body size to ensure successful pollen transfers between two floral types (Jesson & Barrett, 2002; Gao et al., 2006; Ren et al., 2013).

Due to its accurate pollen transfers and high pollination efficiency, mirror-image flowers have the potential to generate floral isolation and facilitate speciation (Grant, 1994; Armbruster & Muchhala, 2009). Hiptage species differ significantly in flower size and length of sexual organs (Chen & Funston, 2008), and it is likely that herkogamy also varies greatly among species. Such differences probably reflect adaptations to different pollinators or different parts of the same pollinator, a phenomenon found in other herkogamous species (Medrano et al., 2005; Kay & Sargent, 2009) or species with similar specialized floral traits (Armbruster & Muchhala, 2009).

Other tropical families with their diversification centre at UZJ also show specialized floral syndromes and pollination mechanisms. For example, Gesneriaceae are well-known for its highly-specialized floral traits and pollination mechanism (Lu et al., 1989; Fang et al., 1995; Li & Wang, 2004) and also contain species with mirror-image flowers in this region (Gao et al., 2006). Begoniaceae are also distinctive for its unisexual flowers and this trait is probably an adaptation to stressful habitats on limestones (Hou et al., 2010). Species of both families in UZJ region are mostly neoendemics (Lu et al., 1989; Fang et al., 1995; Li & Wang 2004; Hou et al., 2010). Nevertheless, this hypothesis about the association of species diversity and floral specialization in UZJ is in need of further experimental tests.

In conclusion, upper reaches of Zhujiang River can be seen as the “evolutionary front” of tropical plants in China due to its predominant proportion of narrowly-endemic and recently-evolved species of Hiptage (Malpighiaceae) and other tropical groups such as Gesneriaceae and Begoniaceae. Environmental factors (rivers with monsoon climate and limestone landscapes) and intrinsic traits of the plants such as highly-specialized pollination systems are responsible for the origin and evolution of the biodiversity in this “evolutionary front”, with the environmental factors more likely being the most significant driving factors. Tropical plants are the main composition of UZJ and its neighbouring regions and should be paid enough attention in the studies of the origin and evolution of this unusual biodiversity hotspot, which is one of globally-important biodiversity hotspots.

Appendix