Genetic differentiation between two varieties of Oreocharis benthamii (Gesneriaceae) in sympatric and allopatric regions

Abstract The pattern of genetic differentiation between diverging species receives much attention as one of the key observable features of speciation. It has often been suggested that introgression between closely related species occurs commonly where their distributions overlap, leading to their becoming more morphologically and genetically similar, but there are a few opposite results. However, most of these studies have been carried out with animals and separate species; few have looked at intraspecific cases, especially in plants. Here, we conduct a comparative study on patterns of genetic differentiation among populations of two varieties of Oreocharis benthamii in allopatry and sympatry based on ISSR data for 754 individuals from 26 populations, in order to understand the processes leading to speciation. Contrary to expectations, the facultative xenogamy (mixed mating) species O. benthamii has a relatively low genetic diversity within populations (H = 0.1014, I = 0.1528) and high genetic differentiation among populations (G ST = 0.5867, ФST = 0.659), as is typically found for selfing species. Genetic variance between the two varieties in sympatric populations (44%, ФST = 0.444) is significantly more than that in allopatric populations (14%, ФST = 0.138). Consistent with the taxonomical delimitation of the two varieties, all sampled individuals of O. benthamii clustered into two genetic groups. Moreover, the genetic structures of populations of both varieties are correlated with their different geographical origins. Our studies show that significant divergence between sympatric populations of the two varieties could be attributed primarily to reinforcement by genetic divergent selection in sympatry where secondary contact had occurred. The major proportion of the genetic variation in outcrossing and mixed mating plants may exist among populations when the populations are distributed in fragmented habitats, due to the paucity of suitable habitat combined with inefficient seed dispersal mechanism and limited pollinator foraging area that may limit the gene flow.

Moreover, the genetic structures of populations of both varieties are correlated with their different geographical origins. Our studies show that significant divergence between sympatric populations of the two varieties could be attributed primarily to reinforcement by genetic divergent selection in sympatry where secondary contact had occurred. The major proportion of the genetic variation in outcrossing and mixed mating plants may exist among populations when the populations are distributed in fragmented habitats, due to the paucity of suitable habitat combined with inefficient seed dispersal mechanism and limited pollinator foraging area that may limit the gene flow.

K E Y W O R D S
character displacement, genetic diversity, mating system, population structure, reinforcement selection

| INTRODUC TI ON
For a long time, patterns of genetic differentiation between diverging species and the evolution of the mechanisms of speciation isolation have received considerable attention, and the topic has been widely discussed by evolutionary biologists wishing to understand the processes leading to speciation. Genetic differentiation or speciation has mainly occurred during periods when habitats were fragmented and isolated (Bridle, Pedro, & Butlin, 2004), and demographic processes such as long-distance dispersal are associated with repeated bottlenecks which may have led to increased genetic divergence, due to founder effects and genetic drift (Chen, Liu, Fan, Li, & Liu, 2017;Freedman, Thomassen, Buermann, & Smith, 2010).
Increased spatial isolation and decreased population size may lead to the erosion of genetic variation and increased genetic differentiation among populations through genetic drift, increased inbreeding, and reduced gene flow between populations (Honnay, Jacquemyn, Bossuyt, & Hermy, 2005). These evolutionary processes bring about divergence within species and lead to speciation ultimately. Among many factors affecting speciation, evolutionary theory suggests that natural selection plays a dominant role in speciation.
In parapatric and sympatric regions, resource competition and/or reproductive interference between closely related species can be reduced to a minimum due to selection, increasing interspecific divergence and enabling the species to coexist (Pfennig & Pfennig, 2009. Thus, operational species criteria should reflect different segments of a continuous process of differentiation between evolutionary lineages, resulting in lineage sorting and reciprocal monophyly (Avise, 2000). However, morphologically ambiguous biological species can arise in response to ecological factors (e.g., development of host race specificity by pathogens, adaptation to climatic and geological changes) without observable morphological changes (Kartzinel, Spalink, Waller, & Givnish, 2016). In addition, reproductively isolated cryptic species can arise in populations that are diverging via genetic drift and thus accumulating genetic incompatibilities, independent of natural selection (Nei & Nozawa, 2011). When there is no niche competition in sympatry, reproductive isolation can arise from chromosomal rearrangements in the absence of other barriers (Grant, 1981;Hutchinson, 1957;Levin, 2002). However, cross-breeding, which can lead to genetic exchange, hybrid formation, and introgression, usually occurs between closely related species in areas where they overlap (sympatry or parapatry), with the result that the two species are genetically more similar in parapatry and sympatry than in allopatry (Anderson & Hubricht, 1938;Mckinnon, Smith, & Potts, 2010;Palme, Su, Palsson, & Lascoux, 2004;Wang, Abbott, Ingvarsson, & Liu, 2014). The potential for hybridization can be reduced by prezygotic or postzygotic mechanisms that constitute interspecies reproductive barriers (Costa, Lambert, Borba, & De Queiroz, 2007;Dobzhansky, 1937;Grant, 1981;Stace, 1989). It has been reported that differing morphology and differentiation of flowering time in the areas of overlap (sympatry or parapatry) between related taxa increases prezygotic isolation with reinforcement; for example, changes in corolla color reduce gamete wastage in Phlox Levin, 1985). Selection for reinforcement might have occurred in parapatry or sympatry leading to increased gene flow barriers, while hybrids arose frequently in the past and were less adaptable than their parent species (Hopkins, 2013). Thus, varieties or incipient species may diverge in sympatry and parapatry, because genetic drift accumulates genetic differences and incompatibilities that lead to the evolution of reproductive barriers when secondary contact takes place (Dobzhansky, 1937;Mayr, 1942;Pfennig & Pfennig, 2009;Rundle & Schluter, 2004). Some studies have shown that when two species overlap geographically, their differentiation is more pronounced in sympatry and diminished or lost entirely in allopatry (Díaz Infante, Lara, Arizmendi, Eguiarte, & Ornelas, 2016 and references therein). A pattern of increased interspecific differentiation between closely related species in sympatry (or parapatry) has been reported for character displacement in a number of animals (Grant & Grant, 2006;Kirschel, Blumstein, & Smith, 2009;Pfennig & Martin, 2010;Pfennig & Murphy, 2002) and plants (Gögler et al., 2015;Grossenbacher & Whittall, 2011;Levin, 1971;Smith & Rausher, 2008;Wang et al., 2014). It is suggested that character displacement may be an alternative to competitive exclusion, arising in sympatry or parapatry so as to decrease competition for resources or bring about reproductive interference (Hopkins, Levin, & Rausher, 2012;Kay & Schemske, 2008;Kirschel et al., 2009;Levin, 1978;van der Niet, Johnson, & Linder, 2006;Urbanelli & Porretta, 2008). However, most of these studies have been carried out with animals and separate species; few have looked at intraspecific cases, especially in plants. In this study, we focus on a morphologically ambiguous plant species, Oreocharis benthamii Clarke, which is endemic in South China (Wang, Pan, & Li, 1990;Wang, Pan, Li, Weitzman, & Skog, 1998).
Plants of O. benthamii are perennial herbs comprising two varieties, var. benthamii and var. reticulata Dunn, with no obvious morphological differences (Figure 1). The minor distinctions between them are that the leaf blade of var. reticulata is ovate-orbicular (vs. var. benthamii is oblong to ovate), and its lateral veins and reticulate veinlets are more prominent than those of var. benthamii (Li & Wang, 2004;Wang et al., 1998). However, these distinctions are not always obvious, especially in leaf blade shape and when the lateral veins and reticulate veinlets are covered with the densely woolly on leaves in some populations, resulting in mistaken identification. Our unpublished phylogenetic analysis of the enlarged Oreocharis species basing on trnL-F and ITS sequence variation showed that var.
benthamii and var. reticulata formed a monophyletic clade and were genetically closely related. Both are small herbs and occur on rock walls in valley or moist soil in humid monsoon forests, and their distribution ranges are substantially the same (Li & Wang, 2004;Wang et al., 1998). However, the two varieties rarely overlap at the same site, and no mixed populations have been found. Our field observations showed that flowers of both varieties were generally purple to blue in allopatry (Figure 1a (Guo, 2011). Sympatric and allopatric populations of the two varieties of O. benthamii represent a good system with which to investigate the erosion of genetic variation and increased genetic differentiation among populations through genetic drift, increased inbreeding, and reduced gene flow between populations and thus to understand the processes leading to speciation and the maintenance of species (variety) boundaries. The ISSR (inter simple sequence repeat) is an easy handling, good reproducibility, low cost, quick and highly informative technique, and widely used for plant population genetic studies (Sharma, Sharma, Rana, & Chahota, 2015;Tabina et al., 2016;Zietkiewicz, Rafalski, & Labuda, 1994), in spite of the incapacity to distinguish heterozygous allele states in an individual. Here, we assess the genetic diversity and population structure of O. benthamii using ISSR markers, focusing on the following questions: (1) How genetic diversity and genetic differentiation are distributed throughout all populations of the species; (2) whether there is increased or reduced genetic differentiation between sympatric populations compared with allopatric populations of these two varieties.

| Material sampled
Both O. benthamii var. benthamii and var. reticulata are nonclonal perennial acaulescent herbs endemic to China, and their distribution ranges are partially overlapped, mainly in southern China (Li & Wang, 2004;Wang et al., 1998). However, during our 12 years

| DNA extraction and ISSR PCR reactions
Total genomic DNA was extracted from silica-dried leaf material using a modified cetyl trimethyl ammonium bromide (CTAB) procedure (Doyle & Doyle, 1987). The quality and concentration of the extracted DNA were estimated on a 0.8% agarose gel and a Nano-100 spectrophotometer (Allsheng). Nuclear DNA was then PCR amplified using ISSR primers obtained from the University of British Columbia. Initially, 100 ISSR primers were screened in 18 samples from six populations of the two varieties and ten polymorphic primers (808,834,841,847,857,873,879,881,899, and 900) were eventually selected for generating ISSR profiles. Reactions were performed in a total volume of 20 μl containing The annealing temperature for each primer is given in Table S2. Negative controls, which lacked template DNA in PCR, were included to test for possible contamination. To ensure reproducibility of the amplification, duplicate PCR amplifications were performed and only clear and reproducible bands were scored. Amplification products were electrophoretically separated in 1.8% agarose gels, together with a 100 bp ladder as a size marker, and visualized on a UV transilluminator of the gel documentation system (Bio-Rad Gel Doc XR+, America). The images of the gels were analyzed using Image Lab software (Bio-Rad) to score for the presence or absence of bands and to assign a fragment size to each band. The presence or absence of bands was also visually confirmed.

| Data analysis
All clear and reproducible amplified fragments were scored as binary characters (presence or absence) and converted into a binary data matrix. The resulting presence/absence data matrix was analyzed using POPGENE version 1.32 (Yeh, Yang, & Boyle, 1999)  also computed using the model presented in Nei (Nei, 1972(Nei, , 1973 (2019), and the results across the independent analyses were combined by the program Clumpp (Jakobsson & Rosenberg, 2007) and visualized with Distruct v2.1 (Rosenberg, 2004).
Hierarchical structure of genetic variation and pairwise genetic distance (Φ ST ) (Excoffier, Smouse, & Quattro, 1992;Meirmans, 2006) among the populations was determined by analysis of molecular variance (AMOVA) with GenAlEx ver.6.5. Significance levels of the variance components were based on 999 permutations. Mantel tests were performed to analyze the effects of geographical distance on genetic variation.

| ISSR polymorphism and genetic diversity
Values obtained for ISSR polymorphism and genetic diversity are summarized in Table S2 and Table 1

| Genetic structure and cluster analysis
Bayesian genetic STRUCTURE analyses revealed that the log likeli-    The PCoA analysis ( Figure S2) shows two main groupings, var.
reticulata (ringed in green) and var. benthamii (red). The var. reticulata group was divided into three subsets, which comprised all populations from, respectively, eastern and central Guangdong, northern Guangdong, and western Guangdong and Guangxi. There were also three subsets in the var. benthamii group: DFX, DNJ, and the subset from eastern and central Guangdong. The result of PCoA analysis thus confirmed the partitioning results obtained from UPGMA clustering and the NJ tree, especially the latter.
Because it is pollinated by parasitic bees (Guo, 2011), the short flight ranges of the insects limit pollen dispersal in O. benthamii. This mosaic distribution pattern also restricts seed dispersal by wind and gravity, due to physical barriers and limited seed dispersal ability.

| Genetic differentiation in O. benthamii
The benthamii in processes associated with recent migration in the regions of overlap as well as earlier introgression.

| Increased genetic divergence between the two varieties of O. benthamii in the sympatric region
It has been suggested that closely related taxa may be more similar in sympatry and parapatry than in allopatry, because introgression often occurs between closely related taxa in areas of overlap (e.g., Anderson & Hubricht, 1938;Behm, Ives, & Boughman, 2010;Mckinnon et al., 2010;Mehner et al., 2010;Palme et al., 2004;Rieseberg & Wendel, 1993 entiation between the two varieties in the sympatric region is significantly higher than that in the allopatric region. A high level of differentiation between species in sympatry or parapatry may arise from demographic processes associated with range expansions by the two species (Freedman et al., 2010). Under conditions in which sympatric or parapatric populations of two species originate from allopatric ones, repeated bottleneck events associated with recent range expansion may result in an increase in genetic differentiation between their populations in sympatry or parapatry due to founder events and genetic drift (Wang et al., 2014). In addition, selection can act to minimize resource competition or reproductive interference between closely related species in parapatry or sympatry, thereby increasing interspecific differentiation and thus enabling the species to coexist . These factors may apply in the cases of var. reticulata and var. benthamii. However, strong genetic drift associated with demographic expansion would result in reduced genetic diversity (Allendorf, 1986;Ellstrand & Elam, 1993), which was evident from comparisons of Nei's genetic diversity (H) made between allopatric and sympatric populations of var. reticulata, but not those of var. benthamii (Table 2). This suggests that sympatric populations probably represent zones of secondary contact between the two varieties of O. benthamii, like as that has been shown in Oreocharis × heterandra (Puglisi, Wei, Nishii, & Möller, 2011) and that var. reticulata was subjected to demographic expansion events, but var. benthamii was not. Moreover, only a weak geographical pattern of isolation by distance was found when intervarietal and intravarietal comparisons were made across the full distribution area, while no geographical pattern of isolation was found in sympatry ( Figure 3). Furthermore, no geographical barriers to gene flow between allopatric and sympatric populations of either of the two varieties were found, indicating that geographical distance may has not played a major role, in restricting gene flow within and between var. reticulata and var. benthamii in the sympatry. Genetic drift or demographic processes are therefore unlikely to have caused primarily the increased genetic differentiation between the two varieties in sympatry. Thus, it is likely that divergent selection (Brown & Wilson, 1956;Pfennig & Pfennig, 2009) may have made a major contribution to the increased genetic differentiation between var.
reticulata and var. benthamii in sympatry, just as was found to be the case in two closely related fir species, Abies chensiensis and Abies fargesii (Wang et al., 2014).
Selection may increase ecological adaptation to different habitats, resulting in a decrease in interspecific competition in sympatry or parapatry (Nosil, 2012;Schluter, 2001). Our field observations showed that flowers of both var. reticulata and var. benthamii were generally purple to blue (Figure 1a-3,b-3), but that all flowers of var.
Reinforcement selection is assumed not only to complete speciation between two incipient species, but also to initiate speciation by causing the evolution of prezygotic isolation between populations of the species undergoing reinforcement (Higgie & Blows, 2007;Hoskin, Higgie, McDonald, & Mortiz, 2005;Howard, 1993;Lemmon, 2009;Rice & Pfennig, 2010). When reinforcement causes reproductive character displacement, there is divergent selection within a species for different mating signals or mating preferences in different parts of the range (i.e., in allopatry vs. sympatry) .
This difference in mating traits can lead to further segregation of sympatric heterospecifics and reduce gene flow between sympatric and allopatric populations of conspecifics (Ortiz-Barrientos et al., 2009;Pfennig & Pfennig, 2009). We suggest that the two varieties of O. benthamii may have initially diverged and acquired sterility barriers in allopatry, followed by a period of range expansion causing secondary contact and reinforcing selection (e.g., flower color alteration-reproductive character displacement) leading to the increased genetic divergence observed in sympatry. However, a more extensive survey combining morphological, ecological, and genomic data could make it possible to reconstruct the phylogeographic history of O. benthamii in order to test this hypothesis.

ACK N OWLED G M ENTS
The authors thank Yan-Feng Guo, Bao-Chen Wang, Guang-Bin Tan, Rong Huang, and Hao-Lin Liang for field assistance.

CO N FLI C T O F I NTE R E S T
None declared.