Volume 234, Issue 2 p. 392-404
Tansley review
Free Access

Species divergence with gene flow and hybrid speciation on the Qinghai–Tibet Plateau

Shengdan Wu

Shengdan Wu

State Key Laboratory of Grassland Agro-Ecosystems and College of Ecology, Lanzhou University, Lanzhou, 730000 China

These authors contributed equally to this work.

Search for more papers by this author
Yi Wang

Yi Wang

Key Laboratory for Bio-Resource and Eco-Environment of Ministry of Education, College of Life Sciences, Sichuan University, Chengdu, 610065 China

These authors contributed equally to this work.

Search for more papers by this author
Zefu Wang

Zefu Wang

Key Laboratory for Bio-Resource and Eco-Environment of Ministry of Education, College of Life Sciences, Sichuan University, Chengdu, 610065 China

Search for more papers by this author
Nawal Shrestha

Nawal Shrestha

State Key Laboratory of Grassland Agro-Ecosystems and College of Ecology, Lanzhou University, Lanzhou, 730000 China

Search for more papers by this author
Jianquan Liu

Corresponding Author

Jianquan Liu

State Key Laboratory of Grassland Agro-Ecosystems and College of Ecology, Lanzhou University, Lanzhou, 730000 China

Key Laboratory for Bio-Resource and Eco-Environment of Ministry of Education, College of Life Sciences, Sichuan University, Chengdu, 610065 China

Author for correspondence:

Jianquan Liu

Email:[email protected]

Search for more papers by this author
First published: 12 January 2022
Citations: 37

Summary

The Qinghai–Tibet Plateau (QTP) sensu lato (sl), comprising the platform, the Himalaya and the Hengduan Mountains, is characterized by a large number of endemic plant species. This evolutionary cradle may have arisen from explosive species diversification because of geographic isolation. However, gene flow has been widely detected during the speciation processes of all groups examined, suggesting that natural selection may have also played an important role during species divergence in this region. In addition, natural hybrids have been recovered in almost all species-rich genera. This suggests that numerous species in this region are still ‘on the speciation pathway to complete reproductive isolation (RI)’. Such hybrids could directly develop into new species through hybrid polyploidization and homoploid hybrid speciation (HHS). HHS may take place more easily than previously thought through alternate inheritance of alleles of parents at multiple RI loci. Therefore, isolation, selection and hybridization could together have promoted species diversification of numerous plant genera on the QTP sl. We emphasize the need for identification and functional analysis of alleles of major genes for speciation, and especially encourage investigations of parallel adaptive divergence causing RI across different lineages within similar but specific habitats in this region.

Author-Provided Video

Species divergence with gene flow and hybrid speciation on the Qinghai–Tibet Plateau

by Wu et al.

Contents
Summary 392
I. Introduction 392
II. Species divergence with gene flow 394
III. Hybrids and hybrid polyploidization 395
IV. Homoploid hybrid speciation 397
V. Concluding remarks 399
Acknowledgements 400
References 402

I. Introduction

The Qinghai–Tibet Plateau (QTP) sensu lato (sl) comprises the platform, the Himalaya and the Hengduan Mountains (Fig. 1a) (Liang et al., 2018; K. S. Mao et al., 2021). Known as the ‘Roof of the World’, the QTP sl region has the highest (average elevation > 4 km) and the most extensive range (> 2.5 million km2) on Earth (Zhang et al., 2014). It extends from the southern edge of the Himalaya to the northern edge of the Kunlun and Qilian Mountains, and from the Pamirs and Karakorum Mountains in the west to the Hengduan Mountains in the east (Zhang et al., 2014). At present, the QTP sl is home to at least 12 000 species of vascular plants in > 1500 genera, and > 20% of the total number of the species are endemic to this region (Wu & Wu, 1996; Wen et al., 2014). In particular, the Himalaya and Hengduan Mountains have the richest species diversity in the Northern Hemisphere and both exhibit a high level of endemism (Sun et al., 2017). This region is a diversity center for many species-rich genera, such as Rhododendron (Ericaceae), Pedicularis (Orobanchaceae), Corydalis (Papaveraceae), Gentiana (Gentianaceae), Kobresia (Cyperaceae), Stipa (Poaceae), Primula (Primulaceae) and Saussurea (Asteraceae), all of which contain numerous endemic species (Wu & Wu, 1996; Sun et al., 2017). In addition, the floristic composition of the QTP sl is highly correlated with environmental gradients in this region (Q. Li et al., 2021).

Details are in the caption following the image
The circumscription and uplift histories of the Qinghai–Tibet Plateau sensu lato (QTP sl). (a) Maps of the QTP sl, including the ‘Platform’ (also called the Qinghai–Tibet Plateau sensu stricto), the Himalaya and the Hengduan Mountains. Insets show representative plant groups on the QTP sl. (b) Topographic evolution models of the Himalaya (proposed by Ding et al., 2017), the platform (proposed by Su et al., 2019a) and the Hengduan Mountains (proposed by Sun et al., 2011; Li et al., 2015; Su et al., 2019b; Spicer et al., 2020). The abscissa represents the era, and the ordinate represents the elevation. Ma, million years ago. Paleo, Paleocene; Plio, Pliocene; Q, Quaternary.

In contrast to monolithic uplift as a coherent entity (Rowley & Currie, 2006), growing evidence suggests that the QTP in fact evolved diachronously (Fig. 1b; Ding et al., 2014; Deng & Ding, 2015; Su et al., 2019a; Spicer et al., 2020, 2021a,b). The present platform or the QTP sensu stricto (ss), as the largest part of the QTP sl, comprised at least two separate high mountain ranges during the Paleocene: the Tanghula upland in the northeast and the Gangdese mountain range in the southwest (Spicer et al., 2021b). Several parts of each of the two ranges were raised much earlier; for example, the Gonjo Basin in the eastern Tangula at c. 3.8 km was subjected to uplift > 40 million years ago (Ma) (Xiong et al., 2020). However, between these two east–west mountain ranges, some lines of evidence suggest that a single deep valley of comparatively low altitude (< 2.3 km) existed until the early Neogene according to recently published data pertaining to the tropical palm flora dated to c. 25 Ma (Fig. 1b; Su et al., 2019a; Spicer et al., 2020). Furthermore, a monsoon climate may still have persisted on the central plateau during the mid-Eocene (Su et al., 2020). It is highly likely that it was only in the Neogene that this subregion reached a ‘platform-like’ plateau through mountainous compression and sediment infilling (Su et al., 2019a; Spicer et al., 2020, 2021a,b). The northeastern Qaidam area of the platform seemed to reach the current elevation much later, at nearly the end of the Miocene (Fang et al., 2020). These findings refute the proto-Tibetan Plateau model based on stable isotope records, which assumes that the entire platform had reached its modern elevation as early as 40 Ma (Rowley & Currie, 2006). It is generally agreed that the Himalaya grew slowly from an elevation of c. 1 km in the late Paleocene (c. 56 Ma) to 2.3 km in the early Miocene (21–19 Ma), and then rapidly uplifted to a high elevation (c. 5.5 km), exceeding the Gangdese Mountains, in the middle Miocene (c. 15 Ma). It reached the current elevation (c. 6 km) more recently (Ding et al., 2017). The Hengduan Mountains, located on the southeastern margin of the present QTP sl, comprise multiple north–south-oriented mountain ridges. The early uplift of its southern part can be traced back to the late Paleocene and even Eocene (Gourbet et al., 2017; Su et al., 2019b; S. Li et al., 2020; Xiong et al., 2020). Some mountainous peaks in this subregion may represent one of the early highlands to arise across the QTP sl, not long after the ancient Gangdese uplands began to approach their current height around 56 Ma (Ding et al., 2014). However, further orogenic activities, river incisions and sediment infilling of the Hengduan Mountains might have still occurred within the Miocene and more recently (Sun et al., 2011; Wang et al., 2012; Cook et al., 2013; Li et al., 2015; Meng et al., 2016; S. H. Li et al., 2020; Spicer et al., 2020).

In addition to diachronous geological activities (Fig. 1), the associated climate changes prevailed from the Paleocene to the Quaternary in the QTP sl (Zachos et al., 2001; Spicer, 2017; Spicer et al., 2020, 2021a). For example, in the early Miocene, the East Asian Monsoon was established in this region due to uplift of the northern QTP sl and temperature decreases in central Asia (Zachos et al., 2001; Spicer, 2017). In the Quaternary, climatic oscillations with cycles of warming and cooling were apparent in the high-altitude regions of the QTP sl. (K. S. Mao et al., 2021). Both geological and climatic changes across the QTP sl should have continuously created extremely diverse habitats, thus providing speciation opportunities (Liu et al., 2014; Wen et al., 2014; Favre et al., 2015; Spicer, 2017; Antonelli et al., 2018; Mosbrugger et al., 2018; Muellner-Riehl, 2019; Rahbek et al., 2019; Ding et al., 2020). On the one hand, geographic isolation as a result of heterogeneous orogenic activities and climatic oscillation across temporal and spatial scales might have greatly favored ‘allopatric speciation’ in such mountainous ecosystems (Mosbrugger et al., 2018). On the other hand, strong divergent selection imposed by contrasting habitats and climates along both environmental and elevational gradients may have particularly promoted ‘parapatric speciation’ in the QTP sl (Liu et al., 2014). However, under both circumstances, hybridization and gene flow could not have been totally impeded by isolating environments and/or divergent habitats during speciation processes in the QTP sl for two reasons: second contacts after the initial isolation of two species may have frequently occurred because of climatic oscillations and range movements (Liu et al., 2014; Yang et al., 2019); and both seed and pollen could have traversed the short geographic barriers in this region, such as between two adjacent mountains isolated by deep valleys (Flantua & Hooghiemstra, 2018). Therefore, this ‘flickering connectivity’ may have played an important role in increasing mountain biodiversity through hybridization (Mosbrugger et al., 2018).

In recent decades, due to the availability of genetic and genomic data, our understanding of the speciation processes leading to the current species diversity on the QTP sl has greatly improved. In this review, we summarize general findings relating to speciation with gene flow in numerous genera occurring in this region, mainly based on studies of population genomics. Specifically, the available evidence suggests that hybrid speciation occurred in numerous groups. The newly identified genetic mechanism and the simple model established for homoploid hybrid speciation (HHS) indicate that hybrid speciation should have occurred more frequently and easily than previously thought on the QTP sl. We finally note that parallel adaptive divergence, causing reproductive isolation and speciation in response to similar environmental stresses, may have frequently taken place on the QTP sl. Our review addresses major areas of progress and key questions pertaining to plant speciation on the QTP sl and could serve as a useful guide for future speciation studies in this region.

II. Species divergence with gene flow

Orogenic activities and accompanying environmental changes created extensive and complex landscapes across the QTP sl (Spicer et al., 2020, 2021a,b). In both the Himalaya and the Hengduan Mountains, high mountains alternate with deep valleys across very short straight distances, creating distinct ‘Sky islands’ (He & Jiang, 2014; Hughes & Atchison, 2015; Luo et al., 2016). The platform also includes diverse habitats from the Gobi desert to alpine screes, although grass steppe and alpine Kobresia meadows dominate in most regions (Liu et al., 2014). Numerous closely related endemic species are distributed on different mountains, valleys and diverse habitats along the altitudinal and spatial gradients of the extensive area of the QTP sl, particularly in the Himalaya and Hengduan Mountains (Sun et al., 2017). Therefore, strong geographic isolation has always been assumed to have driven species divergence in this region (Liu et al., 2014; Wen et al., 2014; Favre et al., 2015; Muellner-Riehl, 2019). However, ecological selection may also have played an important role in causing reproductive isolation (RI) and probably ‘parapatric speciation’ due to high habitat diversity in the region (Liu et al., 2014).

Despite the distinct geographic divides within and between species, gene flow was found to have occurred frequently between the isolated lineages. Zhang et al. (2019) investigated the speciation processes in five diploid Eutrema (Brassicaceae) species based on population genomics data; they found that geographic isolation did contribute to the divergence of these species. However, gene flow still occurred during the speciation process and one species was found to have originated probably through hybridization. This occasional gene flow was also discovered for the isolated populations of one widely distributed Salix species on the ‘Sky islands’ (Chen et al., 2019). Like the species on the isolated mountains of the ‘Sky islands’, those in valleys also show strong genetic divergence due to geographic isolation, and this could have resulted from the historical reorganization of drainage systems (Sun et al., 2017). River dissection can change the original continuous distribution of a plant species into discontinuous patterns, thus promoting genetic differentiation. In turn, secondary contact and gene exchange of isolated populations might be triggered by river capture events (Sun et al., 2021). For example, phylogeographic analyses of an endemic shrub, Terminalia franchetii (Combretaceae), in these deep valleys revealed distinct distributions of the river-specific haplotypes, but historical gene flow via secondary contact by river reorganization and seed-mediated gene flow were also detected between populations presently separated by high mountains (T. C. Zhang et al., 2011). Similar cases have also been reported for Buddleja crispa (Scrophulariaceae) (Yue et al., 2012) and Excoecaria acerifolia (Euphorbiaceae) (Z. W. Wang et al., 2019). Across the entire QTP sl, geographic isolation contributed greatly to the population divergence and formation of the distinct geographic divide in several species, such as Marmoritis complanatum (Lamiaceae) and the relic Circaeaster agrestis (Circaeasteraceae) (D. Luo et al., 2017; Zhang et al., 2020), although gene flow was still found to occur within each geographic lineage. Based on whole-genome population data, Ma et al. (2021) further examined intraspecific divergence within the butterfly bush Buddleja alternifolia (Scrophulariaceae), which has a disjunct distribution in the Himalaya and Hengduan Mountains of the QTP sl and the Loess Plateau in China. Populations from three areas were well delimited, but a low level of secondary gene flow was detected between them after their initial divergence. Further adaptive genes linked to local adaptation for each area were also detected. Therefore, in addition to the important role of geographic isolation for ‘allopatric’ divergence, ecological selection also played an important role in the intraspecific divergence of this taxon.

Within Allium subg. Cyathophora (Amaryllidaceae), two diploid (A. farreri and A. cyathophorum) and one tetraploid (A. tetraploideum) species are allopatrically distributed in the QTP sl (M. J. Li et al., 2021). Based on population genomic data from specific-locus amplified fragment sequencing (SLAF-seq), gene flow was frequently detected during historical divergence of the two diploid species, and a complex hybridization history involved in the allopolyploid origin of A. tetraploideum (see also Section IV). The current species of Genetiana sect. Cruciata on the QTP sl seem to have been derived from extensive diversification due to rapid and strong geographic isolation (Zhang et al., 2009). Using transcriptome population data, gene flow was found to be distinct between G. siphonantha and G. straminea after their initial divergence (C. Chen et al., 2021). This gene flow, due to secondary contact, led to plastome introgression from G. straminea into G. siphonantha. Species divergence along altitudinal gradients is particularly interesting on the QTP sl because the ecological shift therein is more obvious than in any other regions of the world (Liu et al., 2014). Populus rotundifolia (Salicaceae) occurs in high-altitude areas, while its sister species P. davidiana is distributed in the low-altitude region. Based on population genomics data, J. L. Li et al. (2021) found continuous gene flow during the speciation process of these two species. In addition, numerous highly positively evolved genes for local adaptation were found to be fixed in the high-altitude species P. rotundifolia.

Population transcriptome data have revealed continuous gene flow during the speciation of a spruce (Picea, Pinaceae) species complex widely distributed on the QTP sl (Sun et al., 2018). For three closely related species of the genus Primula distributed across the QTP sl, although geographic isolation contributed greatly to species divergence, gene flow was still detected between them based on restriction-site associated DNA sequencing (RAD-seq) data (Ren et al., 2018). In addition, ecological selection was found to have played an important role in maintaining their species cohesion. Based on genomic phylogeny, gene flow was found to be widespread in the species diversification of many genera on the QTP sl. For example, a genomic phylogeny of the genus Picea revealed trans-lineage polymorphism as well as nonbifurcating diversification due to continuous gene flow, especially in the species distributed on the QTP sl (Feng et al., 2019). Similarly, whole-genome sequencing of multiple individuals of each species in the genus Populus recovered widely shared genetic variations and continuous gene flow during speciation in this region (Wang et al., 2020). The QTP sl is the diversification center for the subtribe Gentianinae (Gentianaceae). Phylotranscriptomics of this group have also suggested frequent gene flow and hybrid speciation during the diversification of this subtribe (C. L. Chen et al., 2021). Genomic studies based on single nucleotide polymorphisms (SNPs) called during genome-resequencing, RAD-seq, SLAF-seq and transcriptome analysis have become predominant methods to detect the occurrence of historical gene flow among lineages within a phylogeny (Payseur & Rieseberg, 2016; Árnason et al., 2018; Edelman et al., 2019). More such studies are needed for other groups present on the QTP sl to examine the importance of hybridization for past species diversification.

III. Hybrids and hybrid polyploidization

In addition to historical hybridization and gene flow during diversification on the QTP sl, natural interspecific hybrids and hybrid swarms can be identified for almost all species-rich genera (Yang et al., 2019), including Cupressus (Cupressaceae) (Xu et al., 2010), Gentiana (Zhang et al., 2006), Primula (Zhu et al., 2009), Rhododendron (Milne et al., 2010; Zha et al., 2010; J. Wang et al., 2019; X. X. Mao et al., 2021) and Ligularia (Asteraceae) (Yang et al., 2019). Numerous species of these speciose genera on the QTP sl seem to be still ‘on the speciation [path]way to the final complete RI’ (Liu, 2016). For example, two Rhododendron species, R. irroratum and R. delavayi, in the Hengduan Mountains were found to produce hybrid swarms comprising mainly F1s in areas where their distributions overlapped (Zha et al., 2010). These hybrids were mistakenly classified as a ‘distinct species’ (R. agastum) although no signature of independent evolution was observed. In fact, such hybrids from closely related species of the same section or even from ‘distantly related species’ of different sections have been widely recovered for the species-rich genus Rhododendron (J. Wang et al., 2019; X. X. Mao et al., 2021). Some rare individuals even contain an admixed genetic composition of the three different species in this genus (J. Wang et al., 2019). Further similar explorations through population-level genetic and morphological statistical analyses are urgently needed to rectify the taxonomic placements of such ‘incorrect’ species, which may have been established based on natural hybrids in the QTP sl. This is needed in particular for taxonomic revisions of the above-mentioned and other genera for the Flora of China (Liu, 2016).

These interspecific hybrids may evolve with several outcomes (Abbott et al., 2013). The first-generation hybrids may be sterile because of the high genetic divergence between parents. Otherwise, such F1 hybrids are fertile within themselves and/or with parent(s), producing recombinant or backcrossing hybrids (Abbott et al., 2013). If two parent species with restricted distributions accumulate too few RIs, repeated hybridization will produce widespread hybrids, which admix two parents together to evolve as a single lineage. Under such a scenario, hybridization leads to speciation reversal and a further loss of species diversity (Taylor et al., 2006). To date, no speciation collapse has been directly observed on the QTP sl, although one current allotetraploid Allium was recently revealed to have probably originated from hybridization between two extinct allotetraploid species (M. J. Li et al., 2021). If two parents with distinct distributions have developed enough RIs, stable hybrid swarms with various recombinant types usually appear in the intermediate niches of the two parents, as mentioned before for the genera Rhododendron, Ligulara, Picea and others (Zha et al., 2010; Du et al., 2011; Yang et al., 2019). The backcrossing hybrids can develop into widespread introgressed populations, producing adaptive introgressions from one species to the other (Karrenberg et al., 2019). Such introgressions may enlarge the distributional range of the recipient species and further initiate new hybrid speciation processes and even diversification (Givnish, 2015; Edelman et al., 2019). In addition, recombinant offspring through interspecific F1 and backcross hybrids can evolve into a distinct lineage to directly increase species diversity through independent evolution because of the development of RI during such hybridization processes (Abbott et al., 2013; Wang et al., 2021). Therefore, the frequent occurrences of interspecific hybridization and hybrids provide enough opportunities for hybrid speciation on the QTP sl.

Hybrid speciation comprises both hybrid polyploidization (allopolyploidy) and HHS (Rieseberg & Willis, 2007; Abbott et al., 2013). Allopolyploidy was initially defined as involving the hybridization of two well-delimited diploid species with further polyploidization (Stebbins, 1971). However, this type of hybrid speciation may be overlooked under the two following scenarios (Soltis & Soltis, 1995). First, two diploid genetic groups fail to develop morphological differentiation but maintain cryptic divergence as independently evolved lineages. Second, one diploid ancestor is no longer alive or not found upon investigations of the past. Both scenarios could be present in the QTP sl because diverse niches may lead to and maintain cryptic lineages within a single ‘morphological species’, the repeated mountainous uplifts and climatic oscillations could exterminate the ancestral diploid species, and the newly produced allopolyploid species may replace the ancestral diploids due to its strong competition and superior adaptation in arid habitats (Stebbins, 1971).

Both overlooked allopolyploidy scenarios have been found for plant speciation on the QTP sl. The first concerns a diploid–tetraploid species, Allium przewalskianum, which is widely distributed across the diverse habitats of the QTP sl. Multiple sets of population genetic data have revealed distinct differentiation not only between diploids and tetraploids, but also between diploid groups in diverse niches (Wu et al., 2010; Liang et al., 2015). These divergences were accompanied by occasional gene flow. Based on shared nuclear and chloroplast genetic composition, three allotetraploid lineages were identified and these had originated independently through hybridization between the well-differentiated diploid lineages (Liang et al., 2015). Such repeated cryptic allopolyploid speciation may also account for the origins of multiple allopolyploid species if many closely related and divergent diploid species are able to hybridize with one another, facilitated by geological activities. The other example involves the genus Oxyria (Polygonaceae), which comprises only two extant species. The diploid O. digyna (2n = 14) is found in the alpine habitat of the QTP sl and the arctic region, while the polyploid O. sinensis (2n = 40) occurs only in the dry valleys of the QTP sl. Combined genomic in situ hybridization (GISH), population transcriptomes and phylogenetic analyses suggested that O. sinensis is an allopolyploid species and, in addition to O. digyna, at least one more extinct diploid and one more extinct tetraploid species must have been involved in its origin. Furthermore, two polyploidization events involved in the formation of the extinct tetraploid species and the extant O. sinensis occurred at c. 12 and c. 6 Ma, respectively (X. Luo et al., 2017). However, the reason for the extinction of the unknown diploid and tetraploid species remains unknown, although it might have resulted from superior adaptation of the lately generated tetraploid species, geological activities or climatic changes in the region.

It is interesting that a recent study based on phylotranscriptomics, population genomics and genomic/fluorescence in situ hybridization revealed that three extinct species, one diploid and two tetraploids, may have been involved in the origin of one alpine tetraploid, Allium tetraploideum, on the QTP sl (M. J. Li et al., 2021). This tetraploid arose via HHS from two extinct tetraploid parent species: one was an allopolyploid originating from hybridization of two extant diploids, A. farreri and A. cyaphophorum, and the other an allotetraploid derived from hybridization of an extinct diploid and one of the two extant diploids (M. J. Li et al., 2021). These two case studies suggest that similar analyses may shed light on other allopolyploid species with repeated polyploidization events on the QTP sl. Such studies might further provide the likely means, besides fossils, for generating more useful evidence for the existence of now extinct lineages. In addition to these detailed case studies, allopolyploid speciation has been inferred in more groups mainly occurring on the QTP sl, such as Aconitum subgenus Lycoctonum (Ranunculaceae) (Kong et al., 2017), Buddleja (Scrophulariaceae) (Chen et al., 2007), Ephedra (Ephedraceae) (Wu et al., 2016), Leontopodium (Asteraceae) (Meng et al., 2012) and Saxifraga (Saxifragaceae) (Ebersbach et al., 2020). In contrast to an earlier summary based on a limited number of the available studies (Nie et al., 2005), growing evidence suggests that the contributions of allopolyploidy events in plant diversification may have been underestimated in the QTP sl. Further case studies involving more groups are needed to gain a comprehensive understanding of the relative contribution of allopolyploidy in producing species diversity across the QTP sl.

IV. Homoploid hybrid speciation

As a special type of hybrid speciation, HHS was thought to occur rarely in plants because RI was assumed to be difficult to establish through hybridization (Schumer et al., 2014). However, the origins of at least five diploid species on the QTP sl are assumed to have involved HHS – more than in any other region of the world. The first identified HHS on the QTP sl involves the origin of the alpine pine, Pinus densata (Pinaceae). This hybrid species is exclusively distributed in the high-altitude mountains of the QTP sl mainly in the Himalaya, but extending to the Hengduan Mountains. Its two parent species, however, occur mainly in the low-altitude Yunnan province and northern China with distinct niches (Gao et al., 2012). Compared to its two parents, it can tolerate higher drought stress and is better adapted to the high-altitude habitat (Ma et al., 2010). Another HHS was revealed for the genetic origin of Hippophae gyantsensis (Elaeagnaceae) from its two parent species. This species occurs in both the Himalaya and the platform region, while the two parents are present on the northeastern platform and Hengduan Mountains and have contrasting niches (Jia et al., 2016). The hybrid species Picea purpurea and its two parents occur together in the Hengduan Mountains, but still inhabit different niches with respect to both altitude and latitude (Ru et al., 2018). It is interesting that the three Cupressus lineages involved in the HHS event are distributed in different valleys along three different rivers in the Hengduan Mountains (J. L. Li et al., 2020). Although the diploid hybrid species Ostryopsis intermedia (Betulaceae) and one of the two parent species, O. nobilis, occur together in the southern red-soil region of the Hengduan Mountains of the QTP sl, they occupy contrasting niches (Liu et al., 2014). In addition, the five hybrid species have distinct distributional ranges from their parents (Fig. 2a–e). Geographic isolation can accelerate hybrid speciation, especially HHS in plants (Kadereit, 2015). Therefore, the diverse niches of the QTP sl seem to promote HHS through hybridization between two ancestral species, which have accumulated genetic divergence but still without absolute RI. Note that, theoretically, HHS could occur at the polyploid level. To date, only Paeonia officinalis (Paeoniaceae) has been confirmed to be derived by HHS from extant allopolyploid parents, while one of the diploid ancestors seems to be extinct (Ferguson & Sang, 2001). In the QTP sl, as mentioned previously, only the formation of the allotetraploid Allium tetraploideum has been inferred to have occurred through HHS between two extinct tetraploid species (M. J. Li et al., 2021). It is expected that more cases of HHS at the polyploid level will be recovered from this region given the great species diversity and geographic complexity.

Details are in the caption following the image
Distribution of the five diploid hybrid species and their parent species (a–e) and reproductive isolation between Ostryopsis intermedia and its two parent species, O. nobilis and O. davidiana (f). In each subplot of (a–e), the relationship between hybrid species and its parent species is shown by a phylogenetic tree. The names and distributions of hybrid species are marked with a red color, and their parental species are outlined with sky blue, dark green, light green, purple, and orange for Pinus, Hippophae, Picea, Cupressus and Ostryopsis species, respectively.

The key step for the origin of a novel hybrid species through HHS is for hybrid offspring to develop RI with their parents in sympatry when hybridization occurs between two parents (Rieseberg, 1997) and such RI should arise directly from hybridization (Schumer et al., 2014, 2015). Wang et al. (2021) used three Ostryopsis species as a model system to examine how hybridization leads to the development of RI. They measured RIs through a common garden study and both in vitro and in vivo experiments; they demonstrated that two parent species were isolated by habitat type (soil iron content) and flowering time, while the hybrid species was isolated from each parent through either of the two barriers (Fig. 2f). Based on assembled genomes and population genomic data, they used the newly developed pipeline to identify the candidate major genes underlying these RI traits. The hybrid species inherited iron tolerance alleles from one parent but flowering alleles from the other. Biochemical and transgenic tests of all alleles from the three species confirmed consistent RI phenotypes with those observed in the field. The inheritance of alternate alleles at genes underlying prezygotic parental RI can obviously lead to direct RI of the hybrid offspring from two parents even with sympatry. Further independent evolution and establishment of such a hybrid lineage may occur when it colonizes novel habitats, which may reduce potential niche competition with its parents. This was the first report of HHS genes, and the authors further developed a new simple two-locus model for HHS through prezygotic barriers (Fig. 3a). This model and the method developed for identifying HHS genes was applied to one hybrid bird lineage and the expected candidate genes were recovered, although further functional tests are needed (Wang et al., 2021). In addition, under this two-locus model, two hybrid species may theoretically evolve through RI from each other and from both parents (Fig. 3b). Although both method and model need to be tested in more hybrid speciation systems, HHS seems to occur more easily and quickly, and produce more species to directly increase species diversity than previously thought. Such a study also highlights the importance of prezygotic RIs in HHS. In fact, the prezygotic RIs may be more important in plants than the postzygotic RIs during speciation (Rieseberg & Willis, 2007). All prezygotic RIs can arise directly from natural (or sexual) selection while postzygotic RIs seem to appear mainly as byproducts of the divergent fixtures of allelic variations in at least two loci (Coyne, 1992) due to both selection and genetic drift. However, during HHS, such postzygotic RIs may also arise directly from the alternate fixtures of numerous genetic divergences (both SNPs and indels) between parents across the total genome (Schumer et al., 2015; Sun et al., 2020). Most of these divergent loci between parental species and also between hybrid species and each parent are neutral (Schumer et al., 2015), and identification of the effective loci and functional tests of such postzygotic RIs remain critically difficult. In addition, even under the simplest scenario, at least four effective postzygotic RI genes were needed to create the new RI and one hybrid species during HHS (Schumer et al., 2015). Despite this, both pre- and postzygotic RIs may therefore together arise in the recombinant hybrid lineage by alternate inheritance of numerous divergent SNPs and indels between parents.

Details are in the caption following the image
A simple two-gene genetic model for homoploid hybrid speciation (HHS). (a) Multiple prezygotic reproductive isolation (RI) barriers between two diverging species. The RI barriers are controlled by multiple loci. In the simplest scenario (Wang et al., 2021), only two prezygotic RI genes determining two corresponding but different RI traits are needed to produce a new species through HHS. (b) The ideal recombination of the alleles of two prezygotic RI genes theoretically produces two hybrid species at the same time, with RI between them and also between them and the two parent species.

HHS defined here comprises two types. One type consists of nearly similar genomic contributions from both parents directly through independent evolution from recombination offspring of the first generation (F1) hybrids. The other originates from independent evolution of backcross hybrids (BC1 or BC2) with one parent (Wang et al., 2021) or unequally selective evolution of two parental genomes and, therefore, contains different genomic contributions from two parents. However, for both types of HHS, pre- and postzygotic RIs could arise directly through hybridization by alternate fixtures of parental genetic divergences (alleles). By contrast, pure ancient introgression should only comprise genetic contributions from the other species without RI effect and evolution of an independent lineage. However, the introgressed populations in the recent past suggest early HHS, which was also found to occur frequently in the QTP sl plants. For example, based on population genetic analyses from whole transcriptome sequencing, Ma et al. (2019) examined the divergence process of two closely related cypress species in the high- and low-altitude regions of the QTP sl. These two species are highly divergent because of strong geographic isolation and ecological selection. However, hybridization clearly occurred between them, with introgression from the high-altitude species (Cupressus gigantea) to the low-altitude one (C. duclouxiana). The introgressed genes are functionally related to UV-B protection and survival at low temperatures, which may have helped the low-altitude species to colonize the high-altitude niche during past climatic oscillations. These populations with introgressed genes from the other high-altitude species occur distinctly in the northern and high-altitude area. Although sharing most genomic information and experiencing strong gene flow with the southern and low-altitude populations, they have developed clear genetic divergence through these introgressions. Such introgressed populations are evolving as an independent lineage distinct from the original species through prezygotic RI due to environmental adaptation and selection. A similar introgressed lineage was also found for the genus Rhododendron on the QTP sl (X. X. Mao et al., 2021). This lineage, with multiple populations, is distributed in a niche not overlapping with the two parents and has similarly started independent evolution due to geographic isolation. Therefore, these two introgression lineages seem to suggest HHS at the early stage as mentioned previously. Such introgressions have been reported for multiple genera on the QTP sl, such as in Gentiana (Fu et al., 2020), where the introgressed populations appear also to comprise an independent lineage, although further confirmation based on population genomic evidence is needed. In addition, note that these intraspecific introgressions may disappear through repeated backcrossing and merge with the original species when selection pressures and geographic isolation gradually decrease.

V. Concluding remarks

Across the QTP sl, orogenic activities and climatic changes since the early Paleocene have created extremely diverse environments. Such habitat diversity may have favored speciation through both geographic isolation and divergent selection. In addition, gene flow has been widely detected in the speciation of many genera. Numerous species are still ‘on the speciation [path]way to complete RI’ and therefore produce many natural hybrids in the field. The past and current hybridization between these speciating taxa and diverse niches could have further promoted allopolyploid speciation, HHS and adaptive radiation to increase species biodiversity. Therefore, the extensive species diversification on the QTP sl may have arisen from the combined effects of isolation, selection and hybridization (Fig. 4). Such synthetic drivers may together promote explosive species radiation of a few species-rich genera (Givnish, 2015; Hughes & Atchison, 2015), including Ligularia (Liu et al., 2006), Rhododendron (Milne et al., 2010) and Gentiana (C. Chen et al., 2021; C. L. Chen et al., 2021). For example, hybrid speciation between the same parents may occur repeatedly across different localities and timescales to produce more than one hybrid species (Rieseberg, 1997; Hegarty & Hiscock, 2005). In addition, geographic segregation between parental and hybrid lineages may accelerate hybrid speciation through ‘allopatric isolation’ (Kadereit, 2015). Such repeated isolation and mixture through hybridization can further increase the total species diversity more greatly and rapidly than pure ‘allopatry’ (He et al., 2019). All these hypotheses need further tests based on wide sampling and population genomic data for species-rich genera.

Details are in the caption following the image
The probable drivers for species diversification on the Qinghai–Tibet Plateau sensu lato (QTP sl). The center of the figure is a flat Earth, the color scale represents elevation and the part of the map circled in red is the ‘roof of the world’. The two color blocks (inner circle) represent two major speciation mechanisms: hybridization (pink), and geographic isolation and divergent selection (blue). The outer circle represents respective events associated with each driver.

Selection may have played a more important role than previously thought in plant speciation processes on the QTP sl because gene flow that counteracts divergence has been found in numerous speciation case studies. Therefore, major genes for such adaptive selection need to be recovered to understand their roles in speciation. It is expected that such selection occurrences are first associated with habitat adaptation, such as alpine habitats, dry and hot valleys and red-soils with high iron content. For example, one wild peach species, Prunus mira (Rosaceae), endemic to Tibet, was found to develop a genetic signature for high-altitude adaptation and the detected genes included those related to flavonoid biosynthesis and stomatal development in order to survive the alpine habitat with strong UV-B radiation and drought, compared with low-altitude closely related species (Y. Li et al., 2021). In addition, many convergent phenotypes are found in the arid environments of the QTP sl, for example ‘glasshouse’ plants in the scree habitat and ‘cushion’ morphotypes in the dry alpine habitats. Note that such adaptations may have targeted multiple genes in similar or different pathways. Direct comparison of the genomes of two closely related species from different niches with or without such a phenotype may provide insights into these adaptations (Guo et al., 2018). However, we suggest designing studies at the population level for pairs of species and multiple populations of the widespread species in different niches (Fig. 5). In such studies, genome-wide association studies, genetic mapping and functional tests can be complementarily used to develop an understanding of how natural selection has affected speciation and associated phenotypes. These likely ‘speciation genes’ and allelic divergences can be further compared with other genomic features to understand how much selection and drift have each contributed to the speciation process.

Details are in the caption following the image
Suggested approaches to study parallel adaptive divergence causing reproductive isolation and identify candidate speciation genes. Schematic diagram showing that two populations in diverse niches (high elevation vs low elevation) under divergent selection gradually evolve into two new species. The use of comparative genomics and population genomics is encouraged to identify adaptive variants and selective genomic signatures at both population and species level.

In addition, many endemic species from different genera or families are found in the same niche with particular phenotypes. The convergent phenotypes may result in the phylogenetic and taxonomic position of such species being unclear (Liu et al., 2014). For example, Parasyncalathium souliei (Asteraceae) was recently confirmed as a member of a new genus in the Lactucinae, but had long been misidentified as a member of the genus Syncalathium (Crepidinae) because of the very similar adaptive morphology (J. W. Zhang et al., 2011). However, these endemic species (genera) provide an excellent opportunity to examine the parallel origins of similar phenotypes and adaptive divergence causing speciation if they are compared with their sister lineages without such phenotypes. Although they share similar selective pressures, it remains unknown whether such adaptations and the development of particular phenotypes resulted independently from new mutations or as a result of selection from ancestral standing genetic variation, or through introgression. In addition, because of frequent gene duplication and whole genome duplication (WGD) in plants, such genetic convergences may result from paralogs of the same gene family rather than orthologs as found for animals, which rarely exhibit WGD. Certainly, such convergence may also arise from genetic changes in different genes of the same pathway. Further demographic studies of these adaptive variants and nonadaptive variants will improve our understanding of how parallel adaptive divergence causes RI to occur and which factor is the most important for such adaptive speciation. Despite strong challenges, such evolutionary studies are necessary: they will not only provide new insights into the speciation processes in the diverse niches of the QTP sl at multiple timescales, but also offer useful genetic knowledge for alleviating the effects of severe climate change in the future, which may threaten the survival of these species and populations in arid environments.

Acknowledgements

We would like to thank New Phytologist for the invitation to write this contribution, and several reviewers for their excellent and highly useful comments and suggestions. This work was supported equally by the Strategic Priority Research Program of Chinese Academy of Sciences (XDB31000000) and the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2019QZKK0502), and further by the National Natural Science Foundation of China (grant nos. 32030006, 31590821 and 91731301), the Fundamental Research Funds for the Central Universities (grant no. lzujbky-2019) and International Collaboration 111 Programme (BP0719040). The authors declare no competing interests.

    Author contributions

    JL conceived the concept of the manuscript. JL, SW, YW, ZW and NS designed the figures, and wrote the manuscript. All authors contributed to the final version of the manuscript. SW and YW contributed equally to this work.