Effects of the Tanaka Line on the genetic structure of Bombax ceiba (Malvaceae) in dry‐hot valley areas of southwest China

Abstract Southwest China is an important biodiversity hotspot. The interactions among the complex topography, climate change, and ecological factors in the dry‐hot valley areas in southwest China may have profoundly affected the genetic structure of plant species in this region. In this study, we determined the effects of the Tanaka Line on genetic variation in the wild Bombax ceiba tree in southwest China. We sampled 224 individuals from 17 populations throughout the dry‐hot valley regions. Six polymorphic expressed sequence tag–simple sequence repeat primers were employed to sequence the PCR products using the first‐generation Sanger technique. The analysis based on population genetics suggested that B. ceiba exhibited a high level of gene diversity (H E: 0.2377–0.4775; I: 0.3997–0.7848). The 17 populations were divided into two groups by cluster analysis, which corresponded to geographic characters on each side of the Tanaka Line. In addition, a Mantel test indicated that the phylogeographic structure among the populations could be fitted to the isolation‐by‐distance model (r 2 = .2553, p < .001). A barrier test indicated that there were obstacles among populations and between the two groups due to complex terrain isolation and geographic heterogeneity. We inferred that the Tanaka Line might have promoted the intraspecific phylogeographic subdivision and divergence of B. ceiba. These results provide new insights into the effects of the Tanaka Line on genetic isolation and population differentiation of plant species in southwest China.

network region (He & Jiang, 2014). Studies have also shown that species distribution patterns and structures are also affected by many external ecological factors, including climate, predators, and competitors (Montgomery, 1987).
Southwest China is one of the most important biodiversity hotspots, and it is characterized by extremely complex geographically isolated habitats (He & Jiang, 2014). The altitudes in this region differ greatly where they range from 300 m in Nujiang valley to Mount Gongga at more than 7,556 m above sea level (Zhao & Yang, 1997).
Most of these parallel mountain chains are oriented in a north-south direction, and they are divided by very deep river canyons. The highly complex terrains may provide a relatively stable model of ecologically diverse habitats and glacial refugia because the vegetation and habitats only shifted vertically by a few hundred meters during the Pleistocene climate fluctuation (He & Jiang, 2014). Thus, this particular geographic environment is a natural cradle that maintains species richness. A well-known biogeographic boundary exists in southwest China known as the Tanaka Line (Tanaka, 1954;Zhu & Yan, 2002). The Tanaka Line is considered to be a straight line between approximately 28°N, 98°E and 18°45′N, and 108°E, which divides the two floristic subkingdoms of East Asia, with the Sino-Japanese to the east and the Sino-Himalayan to the west (Li & Li, 1997). The genetic diversity and population subdivisions are markedly different on either side of the Tanaka Line (Fan et al., 2013;Tian et al., 2015), which makes this an ideal region to study the effects of different factors on species diversification and evolution.
Bombax ceiba Linn. (Malvaceae), known as the red silk cotton tree, is a tall, drought-tolerant, and arbor tree species with a wide distribution on both sides of the Tanaka Line (Chaudhary & Khadabadi, 2012). Natural populations of this tree species are widespread in South-East Asian countries at altitudes below 1,400-1,700 m (Li, 1984). In China, B. ceiba occurs naturally in subtropical regions, such as the dry-hot valleys of Yunnan and adjacent provinces (Jin, Yang, & Tao, 1995). The specific habitat range of wild B. ceiba provides an opportunity to verify whether the Tanaka Line has acted as a geological or climatic barrier to affect population structure formation.
In our previous study , we analyzed phylogeographic patterns based on three chloroplast DNA regions (psbB-psbF, trnL-rpl32, and psbI-psbK) in 17 natural B. ceiba populations (201 individuals), where the results showed that the main reasons for differences in the genetic structure of B. ceiba either side of the Tanaka Line are historical climate change and complex topographical conditions . However, it is not clear whether there is an intraspecific divergence pattern in this species where apparent gene flow occurs across the Tanaka Line.
Genetic diversity determines population diversity and the genetic variation among populations or species (Meng et al., 2015;Zhang, Chen, Zhang, Chen, & Fang, 2011). The long-term survival of wild species requires a rich gene pool with sufficient genetic diversity to adapt to continual environmental changes, thereby increasing the likelihood of survival or recovery (Cruz et al., 2012). In the study, we aimed to determine the population structure and genetic variation in wild B. ceiba resources in order to facilitate conservation strategies. We used six pairs of expressed sequence tag-simple sequence repeat (EST-SSR) primers to determine the population structure and diversity of wild B. cieba populations on both sides of the Tanaka Line. Moreover, the PCR products were subjected to Sanger sequencing to make the results more accurate and reliable (Hutchison, 2007).

| Plant materials
We collected 224 samples from 17 populations of B. ceiba in the dry-hot valleys of southwest China (Figure 1). The fresh leaf sample was placed onto silica gel and dried immediately. The latitude and longitude were recorded for each sampled population using GPS system (Garmin, Taiwan), and the locations are listed in Appendix 1.
Voucher specimens were preserved and archived in the herbarium of Southwest Forestry University, China.

| DNA extraction, PCR amplification, and sequencing
Total genomic DNA was extracted from leaf tissues using DNA Extraction Kits (TIANGEN, Beijing, China) according to the manufacturer's protocol. The concentration and quality of the DNA were detected using a spectrophotometer. Six highly variable pairs of EST-SSR primers (Appendix 2) were used to detected polymorphisms in B. ceiba. PCR amplification was performed according to the procedure described previously by Ju, Ma, Xin, Zhou, and Tian (2015). All of the high-quality PCR products were sequenced using the amplified forward and reverse primers with an ABI 3730xl Sequence Analyzer (Life Technologies, Carlsbad, CA, USA).

| Data analysis
The sequences obtained were aligned using MUSCLE (Edgar, 2004) and revised manually in MEGA 7 (Tamura et al., 2011). Hardy-Weinberg equilibrium and linkage disequilibrium were assessed for each population and microsatellite locus pair with PopGen version 32 (Yeh, Yang, & Boyle, 1999). Neutral microsatellite loci were used for the population genetic analyses. Genetic diversity parameters comprising the allele size (A), effective number of alleles (N e ), observed heterozygosity (H o ), expected heterozygosity (H e ), and polymorphism information content (PIC) were calculated for each locus using GenAlEx version 6.501 (Peakall & Smouse, 2012) and PIC_CALC version 0.6. Correlation analyses of the genetic similarity and geographic distances among the 17 populations were calculated using PopGen version 32 and based on a Mantel test (Mantel, 1967) with 999 matrix randomizations using GenAlEx version 6.501. According to the genetic distance matrix calculated among the 224 samples with GenAlEx version 6.501, the similarity matrix was subjected to cluster analysis using the UPGMA algorithm with NTSYS-PC version 2.0 and a dendrogram was generated (Rohlf, 2000). Interpopulation and intrapopulation genetic differentiation were partitioned by analysis of molecular variance (AMOVA) using ARLEQUIN version 3.5.2.1 (Excoffier & Lischer, 2010) with 1,000 random permutation tests. The population genetic structure was determined with the Bayesian clustering approach implemented in STRUCTURE version 2.3.1 (Evanno, Regnaut, & Goudet, 2005). An admixture ancestry model was applied, and 10 independent runs were conducted for each K (1-9) with 50,000 burn-in and 100,000 Markov Chain Monte Carlo iterations. A suitable number of clusters (K) were selected as the largest rate of change in the log probability of data between successive K values (Pritchard, Stephens, & Donnelly, 2000), as implemented in STRUCTURE HARVESTER (available online at: http:// taylor0.biology.ucla.edu/structureHarvester/). Admixture proportions obtained from replicate simulations at the optimal K were averaged using CLUMPP version 1.1.2 (Jakobsson & Rosenberg, 2007). We then employed MIGRATE-N v3.6 (Beerli, 2006) to explore the direction of historical gene flow among the 17 populations based on the Bayesian clustering results. The geographic locations of genetic discontinuities among populations were determined with BARRIER version 2.2 (Manni, Guerard, & Heyer, 2004).

| Genetic diversity
In total, 27 alleles were identified in the six SSR loci among the 224 individuals from 17 B. cieba populations. All of the loci conformed to Hardy-Weinberg equilibrium, and they were polymorphic among populations. The number of alleles (A) ranged from two to seven (   (Table 3). The results indicated a relatively low level of genetic differentiation among groups (2.69%), where 5.89% of the diversity was attributed to the population level and 91.42% was due to the genotypes within the populations. In addition, the population genetic differentiation within the northeast group was F ST = 0.12069 and that among populations in the southwest group was F ST = 0.06810.

| Population genetic structure
According to the six neutral polymorphic markers employed in the population genetic structure analyses using K values ranging from 1 to 10, the STRUCTURE simulation obtained the highest peak at K = 2 (Figure 4). The two B. ceiba subpopulations at K = 2 were attributed to those in the southwest and northeast regions with respect to the Tanaka Line. The membership results inferred that the   Table 4).
We defined two clusters based on the STRUCTURE results in order to evaluate the direction of historical gene flow among the 17 populations. The gene pool in cluster 1 was mostly attributed to the northeast group and that in cluster 2 was attributed to the southwest group. We performed maximum likelihood analyses with MIGRATE-N using 10 short chains (5,000 trees) and three long chains (50,000 trees), where 10,000 trees were discarded as a burn-in. Interesting patterns of historical gene flow were determined between the two groups where these patterns were relatively symmetrical with slight differences (Table 5). However, all of the slightly asymmetrical patterns were related to a population migration direction from the northeast group to the southwest group (m 12 > m 21 ).
BARRIER analysis suggested that the largest genetic breaks in many cases agreed with mountainous areas and rivers ( Figure 5).   (Booy, Hendriks, Smulders, Van Groenendael, & Vosman, 2000). Hence, it is important for plant species to retain as much genetic variation as possible to enhance its likelihood of recovery (Cruz et al., 2012). The natural B. ceiba populations had high genetic diversity in this study, possibly because this species is a perennial species with a high potential for outcrossing via entomophilous flowers (Aluri, Srungavarapu, & Kone, 2005). Previously, Nybom (2004) showed that perennial, outcrossing, and widely distributed species exhibit higher levels of genetic variability within populations. Furthermore, this high diversity may be the main factor that allowed B. ceiba to adapt to harsh environments and become the dominant species (Li, 1984) in these dry-hot valleys through a long evolutionary process.

| Genetic variation
F I G U R E 4 STRUCTURE clustering analysis results for B. ceiba populations based on their geographic distribution. Colors represent the population's probability of populations belonging to either of the two clusters, where blue represents cluster 1 and yellow represents cluster 2.
The right-hand figure shows the number of clusters (K) determined for the B. ceiba populations analyzed was the highest peak that was at K = 2. Each vertical bar in the histogram represents a population

| Population structure
Clustering analysis based on UPGMA and Bayesian methods suggested that the 17 natural B. ceiba populations could be divided into two genetically divergent clusters (Figures 2 and 4) located on either side of the Tanaka Line. This result is similar to that obtained based on chloroplast DNA data in a previous study by Tian et al. (2015). However, in the present study, we detected the gene flow among populations based on SSR markers. These two types of molecular markers differ in terms of genetic diversity and genetic differentiation, and they have been detected in various plants (Kurokawa, Kobayashi, & Ikeda, 2010;Zeinalabedini, Khayamnekoui, Grigorian, Gradziel, & Martinezgomez, 2010). Combining analyses based on nuclear and chloroplast markers can help to elucidate the evolutionary history of species with different inherited patterns (Mariana & Juan, 2016). Thus, in contrast to the chloroplast fragments, the DNA microsatellites could be used to determine contemporary pollen and seed dispersal (Wolfe, Li, & Sharp, 1987). Variation is influenced by the parental heredity and a high level of mutation rate, which reflects the current genetic structure and distribution of genetic variation (Mariana & Juan, 2016).
The B. ceiba flowers are red and cup-shaped with rich nectar, and they could emit a mild fetid smell to attract a wide range of insects and animals, such as bees, birds, bats, and even monkeys (Aluri et al., 2005). While exploring the B. ceiba flowers, animals contact the stigma and stamens so the pollen can adhere to their head and body to facilitate dispersals. Some bees only collect nectar and move between conspecific trees nearby, thereby facilitating pollination (Aluri et al., 2005). This foraging behavior is considered to affect cross-pollination, and it might weaken the genetic structure in the natural populations.
In addition, Ashoke (1999) found that the highest number of pollen grains generated per flower by B. ceiba was about 8,863,000 and the maximum atmospheric incidence was 156/m 3 at 10 hr.
In addition, the F ST analysis showed that the proportion of genetic differentiation among populations accounted for about 0.1749 of the total genetic diversity (Table 4). According to Wright (1978), the differentiation among populations is relatively large (0.15-0.25). Clearly, a positive correlation between the genetic and geographic distances was detected among the populations (r 2 = .2553, p < .001) (Figure 3), and thus, topography may be one of the most important factors that have led to differentiation. Natural adaptation probably explains the first level of differentiation within the progenitor B. ceiba population, while habitat fragmentation may have been responsible for the second level of hierarchical variation. The isolation between populations is due to physical barriers in the form of complex terrain with mountains and rivers in southwest China. The genetic structure is expected to be congruent with the geographic arrangement of the mountains and river systems. The genetic distance was relatively large even with a close geographic distance, and genetic discontinuities between the two nearby territories were also identified by BARRIER ( Figure 5).
BARRIER analysis based on microsatellite data showed that, in recent times, variations in the topography and climate have contributed to the high endemic biodiversity in southwest China (Myers, Mittermeier, Mittermeier, Da, & Kent, 2000 (Tanaka, 1954;Zhu & Yan, 2002). The heterogeneous environmental conditions on the Tanaka Line have significantly affected the development and evolution of plant species, that is, a genetic diversity study of Sophora davidii found obvious differences in the population structure on both sides of the Tanaka-Kaiyong Line (Fan et al., 2013). Hence, the Tanaka Line may be responsible for maintaining the major southwest and northeast split in the B. ceiba populations associated with an ecological transition. This major form of isolation may hinder the gene exchange via birds but not pollen dispersal. Hence, this pattern may weaken the specific population structure of B. ceiba on either side of the Tanaka Line. Natural adaptation and physical barriers could explain the divergence among the two subpopulations. Overall, our findings support a hypothesis that the Tanaka Line has contributed to the intraspecific divergence pattern in this species, thereby facilitating the protection and exploitation of wild B. ceiba population resources.

ACKNOWLEDGMENTS
This study is supported by the National Natural Science Foundation of China (NSFC: 31260050) and the National Key R&D Program of China (2017YFC0505200). We are grateful to Dr. Duncan E. Jackson for his help in the manuscript modification.

CONFLICT OF INTEREST
None declared.

AUTHOR'S CONTRIBUTIONS
BT contributed to the conception of the study. BT and YF collected the materials. ZHL and MMJ contributed significantly to analysis and manuscript preparation. MMJ performed the data analyses and wrote the manuscript. CZH contributed the reagents/materials/analysis tools. BT, ZHL, and GFZ helped perform the analysis with constructive discussions. All authors contributed critically to the drafts and gave final approval for publication.