Current biogeographical roles of the Kunlun Mountains

Abstract Large‐scale patterns of biodiversity and formation have garnered increasing attention in biogeography and macroecology. The Qinghai‐Tibet Plateau (QTP) is an ideal area for exploring these issues. However, the QTP consists of multiple geographic subunits, which are understudied. The Kunlun Mountains is a geographical subunit situated in the northern edge of the QTP, in northwest China. The diversity pattern, community phylogenetic structures, and biogeographical roles of the current flora of the Kunlun Mountains were analyzed by collecting and integrating plant distribution, regional geological evolution, and phylogeography. A total of 1911 species, 397 genera, and 75 families present on the Kunlun Mountains, of which 29.8% of the seed plants were endemic to China. The mean divergence time (MDT) of the Kunlun Mountains flora was in the early Miocene (19.40 Ma). Analysis of plant diversity and MDT indicated that the eastern regions of the Kunlun Mountains were the center of species richness, endemic taxa, and ancient taxa. Geographical origins analysis showed that the Kunlun Mountains flora was diverse and that numerous clades were from East Asia and Tethyan. Analysis of geographical origins and geological history together highlighted that the extant biodiversity on the Kunlun Mountains appeared through species recolonization after climatic fluctuations and glaciations during the Quaternary. The nearest taxon index speculated that habitat filtering was the most important driving force for biodiversity patterns. These results suggest that the biogeographical roles of the Kunlun Mountains are corridor and sink, and the corresponding key processes are species extinction and immigration. The Kunlun Mountains also form a barrier, representing a boundary among multiple floras, and convert the Qinghai‐Tibet Plateau into a relatively closed geographical unit.

theory has integrated the relative effect of contemporary environments and historical processes on biodiversity patterns (Hawkins & Porter, 2003;Montoya et al., 2007;Svenning & Skov, 2005;. In addition, the potential linear correlation between contemporary environments and historical processes is difficult to distinguish from their respective value. Therefore, it is particularly important to independently explore contemporary environments and historical processes. Integration analysis of taxonomy, phylogeny, ecology, biogeography, phylogeography, and paleontology may offer an insightful perspective on the biodiversity patterns at different scales (Li, Qian, Sun, 2018). Ecologists and biologists have analyzed multiple regions to independently explore historical processes of biodiversity patterns. For example, the formation time of the Andes flora was 6.40 Ma Särkinen et al., 2012); they present biogeographical roles are cradles of alpine flora, a dispersal barrier to lowland species, and a dispersal corridor for South and North American species (Antonellia et al., 2009;Luebert & Weigend, 2014).
The Andean uplift has been vital for the evolution of the Amazonian flora, with a formation time of 8.30 Ma Hoorn et al., 2010). The formation time of the Australian flora was estimated to be 18.80 Ma Crisp & Cook, 2013). Their biogeographical roles are reservoirs of ancient taxa and sinks of recent lineages (Crisp & Cook, 2013). The formation time of the South African flora was 18.70 Ma Linder & Verboom, 2015). Speciation and dispersal played a leading role in biodiversity in the area (Linder & Verboom, 2015). According to the analysis on the angiosperm flora of China, the eastern China act as museums and cradles of woody species, and the western China act as cradles of herbaceous species .
Studies have widely recognized that the abiotic environment, contemporary biotic interactions, and evolutionary history jointly explain community assembly at different scales (Cavender-Bares et al., 2009;Ricklefs, 2006;Vellend, 2010). The phylogenetic community structure can be used to explore ecological and evolutionary processes of community assembly at different scales (Kraft et al., 2007;Webb et al., 2002). Evolutionary processes such as rapid in situ speciation, niche conservatism, and dispersal limitation can lead to phylogenetic clustering . In comparison, evolutionary processes such as niche evolution, convergent evolution, and colonization may lead to phylogenetic overdispersion within communities (Allen & Gillooly, 2006). Ecological processes, habitat filtering, and competitive exclusion can result in nonrandom community phylogenetic structures (Kraft et al., 2007;Webb et al., 2002). Habitat filtering can lead to phylogenetic clustering, the process select species with similar functional traits into a community (Wiens & Graham, 2005), whereas a community dominated by competitive exclusion might show phylogenetic dispersion (Burns & Strauss, 2011).
With topographically complex mountains, the biodiversity and ecosystem processes between mountains and adjacent lowlands are influenced by biotic interchange, regional climate, and nutrient runoff (Rahbek, Borregaard, Antonelli, et al., 2019). In addition, mountains reportedly disproportionately influence the global terrestrial biodiversity, especially in the tropics, where they harbor extraordinarily rich species. Generally, the mountains of the arctic and temperate regions have few endemic species and low species diversity; biodiversity of these mountains barely exceeds that of the adjacent lowlands (Rahbek, Borregaard, Colwell, et al., 2019). In addition, at a large spatial and temporal scale, geological history and abiotic environment jointly regulate four key processes that determine the biodiversity worldwide: speciation, dispersal, persistence, and extinction (Rahbek, Borregaard, Antonelli, et al., 2019). Consequently, mountains are ideal regions for exploring the mechanisms that govern biodiversity patterns at different scales. Based on the different processes, mountains are classified as having different biogeographical roles (Rahbek, Borregaard, Antonelli, et al., 2019).
The high mountains of China are mainly located in the Qinghai-Tibet Plateau (QTP) and adjacent regions (Wang et al., 2004). The QTP is the plateau itself, which is the largest and highest plateau in the world, occupying an area of 2.5 million km 2 , with an average elevation of over 4000 m (Zhang et al., 2002). The datasets thereby accumulated from studies that have been conducted on the QTP offer opportunities to investigate the biodiversity patterns and plant communities in the regions (Favre et al., 2015). According to data in published monographs and literature, the QTP possesses ~10,000 species of vascular plants (APGIV, 2016;Wu, 2008), of which ~20% are endemic to the region (Wu, 2008;Yan et al., 2013;Yu, Zhang, et al., 2018). Further, species richness varies considerably across the region (Mao et al., 2013;Yan et al., 2013), with the southern regions having especially high species richness (Mao et al., 2013). Rapid speciation and habitat filtering have been reported to dominate the biodiversity and community assembly processes on the QTP, and the phylogenetic structure of vascular species is clustered in most regions of the QTP (Yan et al., 2013). The geological history and uplifts of the QTP are still being debated because the QTP consists of multiple geographical subunits that have experienced different geological events and uplifts Renner, 2016;Spicer et al., 2020;Su et al., 2019;Sun & Zheng, 1998).
However, previous studies indicate that the QTP has risen to its current elevation only in the late Neogene (23.3 Ma-2.6 Ma) Spicer et al., 2020;Su et al., 2019). There is a consensus that the QTP has undergone strong climatic fluctuations and four major glacial events during the Quaternary (Owen & Dortch, 2014;Renner, 2016;Shi et al., 1998). These geological processes at the QTP have promoted radiation and species diversification in various plants taxa , and caused mass plant extinction.
Owing to major advancements in phylogeographic studies and tools, the numerous plant speciation and adaptations in the QTP and adjacent regions, such as Saussurea (Wang, Susanna, et al., 2009), Rheum (Sun et al., 2012), Gentiana (Favre et al., 2016), Rhodiola , Saxifraga (Ebersbach et al., 2017), and Syncalathium , among others (Liu et al., 2017;Qiu et al., 2011), have been increasingly reported (Liu et al., 2014). These datasets provide the opportunity to explore the biodiversity formation and maintenance mechanisms in these areas. Datasets from different subunits have driven further exploration of the plant diversity on the QTP. For example, the Hengduan Mountains have acted as cradles, refugia, and independent biogeographic sources since the Neogene (Ding et al., 2020;Liu et al., 2017;Muellner-Riehl, 2019;Sun et al., 2017;Xing & Ree, 2017). The alpine flora of the Hengduan Mountains is the largest source of species dispersal for the Himalayas and the QTP (Ding et al., 2020). Recent studies have shown that the main phylogeographic patterns of seed plants include contraction/ recolonization, platform refugia/local expansion, and microrefugia in the Tibeto-Himalayan region (Muellner-Riehl, 2019).
Most previous researches have focused on the QTP as a whole, and there has been little research on the geographical subunits in the region. The Kunlun Mountains are a geographical subunit with a relatively clear geographical range and available plant distribution data; however, they are not considered as a biodiversity hotspot and seem to harbor few species (Pan, 2000;Su, 1998;Sun et al., 2015;Wu, 2012Zachos & Habel, 2011;Zheng, 1999;). With respect to the phytogeographical regions of the Chinese flora, the Kunlun Mountains form the border between the Tethyan region and the QTP (Ye et al., 2019(Ye et al., , 2020. In addition, they present the richest contemporary glaciers in China (Liu et al., 2015). Therefore, The annual precipitation and average annual temperature of the regions vary from ~100 to 500 mm and below 0°C, respectively. The annual precipitation is characterized by a decrease from the east to the west. The climate varies also on the slopes of the mountains, with a steep climate gradient leading to a dramatic change in vegetation cover. The dominating vegetation types are alpine meadow and alpine steppe, a few alpine scrubs and coniferous forests distribute in the east and west of the Kunlun Mountains (Wu, 2012Zheng, 1999).
The uplift of the Kunlun Mountains coincided with the Himalayan movement, and its geological history and uplifts are still unknown (Duvall et al., 2013;Jiang et al., 2013;Wang et al., 2003;Wang & Chang, 2012;Yin et al., 2008). However, it is certain that the climatic fluctuations and glaciations of the Quaternary also occurred in the Kunlun Mountains (Owen et al., 2008;Owen & Dortch, 2014;Renner, 2016).
To accurately analyze the regions, the Kunlun Mountains was  Table 1).

| Species distribution
The species distribution was derived from Flora Kunlunica (four volumes) by Wu (2012, with references to relevant local floras, specimens' information, and other literature, including Tibet Autonomous Region (Wu, 1983(Wu, -1987, Xinjiang (Shen, 1993(Shen, -2011, Qinghai (Liu, 1996(Liu, -1999, the Qinghai-Tibet Plateau (Wu, 2008), and the National Specimen Information Infrastructure antli st.org). When species names differed between these databases, we followed The Plant List. The information formed a comprehensive checklist, which only recorded wild seed plants, and preserved infraspecific taxa. To reveal spatial patterns, a species checklist of each county-level geographical unit was also created using species distribution data.

| Geographical origin and divergence time of floras
To reveal the geographical origin and divergence time of the flora, we collected data from published phylogeography of clades, following two principles of data collection: taxa from the Kunlun Mountains flora had to include geographical origin or divergence time of these clades, with the divergence time of these clades being crown age (Table S1). Based on the corresponding data, mean divergence times (MDTs) were calculated as: where AGE i is the age of the genus i (i = 1, …, n) in a sample, and S i is the species number of the genus i in the sample. The MDTs of these clades may be used to explore spatial divergence patterns in a region . The unit of MDT is Ma, which stands for million years.

| Phylogenetic structure
The nearest taxon index (NTI) was calculated to reveal the community phylogenetic structure, and to explore possible ecological and evolutionary processes of community assembly (Webb et al., 2002).
The NTI was based on the mean nearest taxon distance (MNTD), which show the total of the mean phylogenetic relatedness between each taxon and its nearest relative in a sample. The NTI indicates the structure in the shallower parts of a sample (Webb et al., 2002). The positive NTI values indicate that the community phylogenetic structure is phylogenetically clustered, whereas negative values indicate that the community phylogenetic structure is phylogenetically dis- Phylogenetic analyses require a phylogenetic tree of seed plants, and the phylogenetic tree used in our study was constructed using Phylomatic (http://phylo diver sity.net/phylo matic/) with the stored tree data from Zanne et al. (2014). Phylomatic standardizes the species names according to The Plant List (Qian & Jin, 2016). The phylogenetic tree was obtained using the Phylomatic dependent on the Angiosperm Phylogeny Group Ⅳ and standardized species names.  (Table 1).
Overall, the Kunlun Mountains flora was spatially varied (Table 1).
The species and genera richness indicated that biodiversity in the eastern region was higher than those in the western and central regions (Table 1; Appendix S3). Similar results also characterized the endemic taxa on the Kunlun Mountains (Table 1). Consequently, the eastern region was the center of the species richness, genera richness, and endemic taxa in the Kunlun Mountains flora.

| Geographical origin and divergence time of floras
In this study, 126 clades of seed plants (species or genus level) were collected, accounting for 126 genera, 55 families, and 30 orders of greater at both ends of the areas, and the eastern flora was found to be older than the western flora ( Figure 5a). However, the SES-MDTs of the 24 counties did not show significant differences.
Geographical origin analysis of 126 clades on seed plants indicated that they were primarily from the Laurasian flora, such as Eastern Asia (40 clades), Tethyan (18 clades), and Northern Hemisphere unknown (28 clades), while only three clades were from the Gondwanan flora (Table 2; Table S1), which were only distributed in the eastern region.

| Phylogenetic structure
The NTIs were calculated by the phylogenetic tree of angiosperms, and indicated that the counties had different phylogenetic structures ( Figure 5c). In addition, the phylogenetic tree of gymnosperms could not calculate the index ( Figure S1). The 27 NTIs were positive in county-level communities, and 23 of these NTIs were statistically significant (p < .05). Furthermore, the only one NTI was negative in the Banma community and showed significant differences (p < .05; Chinese endemic species accounted for approximately 20% of the total species in the QTP (Wu, 2008;Yan et al., 2013;Yu, Zhang, et al., 2018), and 32.4% of the total in the Hengduan Mountains (Zhang et al., 2009). The distribution pattern of taxa was that these taxa were mainly distributed in the eastern region of the Kunlun Mountains ( Figure 4).

The results of MDT analysis indicate that the Kunlun
Mountains flora is ancient (19.40 Ma) compared with the flora of western China (15.29-18.86 Ma;Lu et al., 2018). In addition, the and Unknown (23 clades), while only one clade was from the QTP (Table 2). However, the MDT only represents temporal patterns of these clades in the Kunlun Mountains flora. Geographical origin of these clades revealed only their spatial patterns. The results of MDT indicated that the origin time of the Kunlun Mountains flora was determined to be later than the early Miocene (19.40 Ma) because temporal patterns of these clades could be assembled since 19.40 Ma. In addition, the geographical origin of these clades also indicates that spatial patterns of these clades are not in situ assemblies in the Kunlun Mountains. Therefore, the formation processes of the Kunlun Mountains flora could include at least ex situ speciation and dispersal since 19.40 Ma.
Recent studies have indicated that the formation of the QTP only occurred in the late Neogene  and that the formation of the Asian monsoon system also emerged in the Neogene Xie et al., 2021). Some studies have recently demonstrated that the ecosystem of the QTP experienced a significant shift at the Paleogene/Neogene boundary  and that the Kunlun Mountains have reached their present height over the last 17 Ma (Pan, 2000;Sun et al., 2015).   (Huang, 1994). Therefore,  Owen et al., 2008;Owen & Dortch, 2014;Renner, 2016;Su, 1998), such as the Largest Glaciation (1.2-0.6 Ma) and the Last Glacial Maximum (Liu et al., 2014;Shi et al., 1997). Almost all species went extinct in the Kunlun Mountains because of the numerous glaciations. A recent study has highlighted that the main phylogeographical patterns of seed plant species in the Tibeto-Himalayan region are contraction/recolonization, platform refugia/local expansion, and microrefugia (Ding et al., 2020;Muellner-Riehl, 2019). However, some previous reports also indicated that few species, no Chinese endemic species included, were harbored in the platform refugia and microrefugia (López-Pujol et al., 2011;Muellner-Riehl, 2019). In addition, a recent study has suggested that no platform refugia existed on the Kunlun Mountains (Yu, Favre, et al., 2018). These studies in-

| The phylogenetic structure of the Kunlun Mountains flora
The extant biodiversity on the Kunlun Mountains occurred by species recolonization. A complex species recolonization was likely the most important evolutionary process to affect the deeper phylogenetic community structure. The evolutionary history of taxa was significant, particularly for the net relatedness index (Webb et al., 2002). NTI analyses can help reveal the phylogenetic structure in a community; NTIs primarily reveal the shallower parts of community phylogenetic structure. The complex sources of species colonization had little influence over NTIs. When NTI was closer to 0, the neutral theory could explain the community assembly. In contrast, niche theory may be used to reveal community assembly.
Based on the results of NTI, the community phylogenetic structure was dispersed in Banma, whereas the community phylogenetic structures were clustered in the other counties. Twenty-four counties showed statistically significant NTIs (p < .05). The three positive NTI, namely, Jiuzhi, Maqin, and Minfeng, were greater than 1, and the NTI in Yutian was 0.97; however, they were not significantly different ( Figure 5c).
Previous studies have revealed that abiotic determinism tends to increase with spatial scale, while biotic determinism tends to decrease with spatial scale. The abiotic determinism dominates biodiversity maintenance mechanisms at the regional scale; the biotic interactions had little effect on biodiversity (Cardillo, 2011;Charles et al., 2010;Niu et al., 2011;Villalobos et al., 2013;Yang et al., 2014). Therefore, the abiotic environment and evolutionary history of biodiversity patterns greatly influence the community phylogenetic structure at the regional scale (Kraft et al., 2007).
In addition, in the QTP, the responses of species diversity to climate depend on the biotype. The diversity of woody plants was more strongly associated with climate than that of herbaceous plants. Energy and water availability jointly influence the diversity of woody plants, whereas water availability alone predominantly regulates the diversity of herbaceous plants (Yan et al., 2013 (Wu, 2012Zheng, 1999), while there are numerous rivers in the eastern and western regions ( Figure 2). Therefore, the positive NTIs indicate that habitat filtering was primarily the driving force behind community assemblies.
We speculate that water availability plays an important role in the current biodiversity pattern, particularly in the western and central regions of the Kunlun Mountains. Based on the vegetation type, forests were concentrated in Banma. The MDT of the flora in Banma was greater than 21 Ma, which was the most ancient flora.
The combination of species from multiple floras and adequate hydrothermal conditions may explain the dispersion of phylogenetic structures in Banma.

| CON CLUS IONS
The biodiversity patterns indicate that the eastern region of the Kunlun Mountains is a center of species richness and endemic taxa.
However, compared with the flora in the southeastern part of the QTP, the Kunlun Mountains flora has relatively low biodiversity, which is consistent with the findings of previous studies Mao et al., 2013;Yan et al., 2013).
The results of MDT analyses indicate that the divergence time of the Kunlun Mountains flora was in the early Miocene (19.40 Ma), and the eastern region was the most ancient. However, the extant biodiversity on the Kunlun Mountains appeared in the early Pliocene

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.