Genetic diversity and its conservation implications of Vitex rotundifolia (Lamiaceae) populations in East Asia

Population sampling

A total of 157 individuals of V. rotundifolia and V. trifolia from 25 sampling sites for cpDNA and 177 individuals of V. rotundifolia and V. trifolia from 27 sampling sites for ITS were sampled from distribution areas in East Asia (Table 1 and Fig. 1). Leaf samples were collected from individuals separated by more than 20 m to avoid collecting the same clones (genotypes). The leaf samples were dried in silica gel and then stored in a freezer (−20 °C) for use. A sample of V. negundo from Shanghai was also collected as the outgroup. Voucher specimens were deposited in the herbarium of the Second Military Medical University (Shanghai, China).

Sequencing of cpDNA and nrDNA

The noncoding regions DNA sequences of cpDNA trnG (encoding tRNA glycine)-trnS (encoding tRNA serine) and trnH (encoding tRNA Histidine)-psbA (encoding photosystem II protein D1) and ITS intergenic spacer region DNA sequences were examined to analyze the genetic variation in V. rotundifolia and V. trifolia from different geographical locations (Hamilton, 1999; Shaw et al., 2007; Zhang et al., 2014). Total DNA was extracted using a Hi-DNA secure Plant Kit (Tiangen Biotech (Beijing) Co., Ltd, Beijing, China). PCR amplification was conducted in a total reaction volume of 50 µl, containing 10–20 ng total DNA, with 1 µl of each primer, 21 µl of ddH₂O, and 25 µl of 2 × Trans Taq High Fidelity (HiFi) PCR SuperMix (Transgen Biotech, Beijing, China). Double-stranded DNA was amplified after 3-min incubation at 94 °C; followed by 30 cycles at 94 °C for 30 s, 54/56 °C for 30 s, and 72 °C for 30 s; with a final extension at 72 °C for 15 min. To examine the geographical distribution of cpDNA and nrDNA in V. rotundifolia and V. trifolia, we amplified two noncoding regions of cpDNA trnG-trnS and trnH-psbA and ITS intergenic spacer regions using primers that successfully amplified the expected DNA fragments and exhibited some variations in our preliminary experiments (Hamilton, 1999; Shaw et al., 2007; Zhang et al., 2014). Primers used for PCR amplification were also used as sequencing primers. The PCR products were assessed using electrophoresis through 1.0% agarose gels and then used as templates for direct sequencing. Sequencing was conducted from both ends. The cpDNA and ITS sequences data of V. rotundifolia and V. trifolia samples have been submitted to the GenBank database, and the accession numbers for the trnG-trnS, trnH-psbA, and ITS sequences are MG822129 –MG822287, MG822468 –MG822626, and MG822288 –MG822467, respectively.

Phylogenetic analyses

Geneious Pro V4.8.5 (Drummond et al., 2017) was used to check and align the DNA sequences. Bayesian analysis (BI) was performed on MrBayes3.1.2 to construct the phylogenetic tree of haplotypes based on the combined cpDNA data (Ronquist et al., 2012). T92+G was selected as the model for the cpDNA dataset using Modeltest version 3.06 (Posada & Crandall, 1998) and the Akaike Information Criterion (AIC) was used to calculate the parameters and assumptions for the sequence partitions. Two runs of four simultaneous Monte Carlo Markov Chains (MCMC) analyses were performed for 10,000,000 generations, with sampling done every 100 generations. Log-likelihood values were examined for stationarity to determine the burn-in value, which was 5,000 trees in this case. TCS 1.21 (Clement, Posada & Crandall, 2000) was applied to build a genotype network based on the 90% parsimony criteria with the option of treating gaps as a fifth base.

Molecular dating

The program BEAST version 1.5.3 (Drummond et al., 2012) was used to estimate the divergence times of the major lineages to the most recent common ancestor (TMRCA). Considering there are no fossil records and no specific substitution rates to calibrate the molecular clocks, the published nucleotide substitution rates were used to estimate the divergence time between two species, and a rate of about 1.2–1.7 × 10⁻⁹ substitutions per neutral site per year (s/s/y) was used to obtain absolute values of TMRCA (Eyre-Walker, 2000).

Population genetic diversity and genetic differentiation

The program DnaSP v5.10 was used to calculate the haplotype diversity (h_d) and nucleotide diversity (π_d) based on the cpDNA and nrDNA sequences of V. rotundifolia and V. trifolia (Librado & Rozas, 2009). To prevent an insufficient sampling affect, we re-analyzed the nucleotide diversity of V. rotundifolia excluding poorly sampled populations.

The analyses of molecular variance (AMOVA) in ARLEQUIN 3.5 were performed to calculate the genetic variation between V. rotundifolia and V. trifolia groups, among populations within groups, and within populations, using a significance test based on 1,000 permutations (Excoffier & Lischer, 2010). The program PermutCpSSR_1.2.1 (Pons & Petit, 1996) was used to calculate the within-population diversity (h_S), total diversity (h_T), geographical total haplotype diversity (V_T), geographical average haplotype diversity (V_S), the level of population differentiation at the species level (G_ST), and an estimate of population subdivisions for phylogenetically ordered alleles (N_ST). G_ST and N_ST are often used to assess the geographical structure affecting population differentiation.

Demographic history

Tajima’s D test and Fu’s Fs test in the ARLEQUIN 3.5 software, with 1,000 permutations, were applied to detect historical demographic range expansions of V. rotundifolia and V. trifolia (Excoffier & Lischer, 2010). The significance of the D value is associated with bottlenecks, selective effects, population expansion, or heterogeneity of mutation rates (Tanaka et al., 2011). In addition, ‘mismatch distributions’ analysis was performed using the DnaSP v5.10 program to detect any recent demographic expansions of V. rotundifolia or V. trifolia (Librado & Rozas, 2009).

Genetic diversity based on cpDNA and ITS sequences

Based on the 157 samples of V. rotundifolia and V. trifolia individuals, the nucleotide sequence lengths were 848 bp and 271 bp for the trnG-trnS and trnH-psbA regions, respectively. The combined cpDNA sequence length after multiple alignments was 1,119 bp. In total, six polymorphic sites and eight haplotypes were detected, and the aligned sequence variations are summarized in Table 2. The number of haplotypes, haplotype diversity (h_d), and the nucleotide diversity (π_d) within each population are presented in Table 1. Finally, three haplotypes were obtained from 17 populations, plus four sites with a single sample of V. rotundifolia, and five haplotypes were obtained from four sampling sites of V. trifolia (Fig. 2). The value of haplotype diversity (h_d,cp) ranged from 0 to 0.639, and the nucleotide diversity (π_d,cp) from 0 to 0.00065 in the 21 sampling sites for V. rotundifolia. At the species level, h_d,cp and π_d,cp were 0.360 and 0.00010, respectively. When excluding six poorly sampled populations, h_d,cp and π_d,cp slightly changed to 0.363 and 0.00010, respectively. Within the four V. trifolia sampling sites, h_d,cpranged from 0 to 0.639 and π_d,cp from 0 to 0.00065, and at the species level, h_d,cp was 0.812 and π_d,cp was 0.00039. These results indicated that both V. rotundifolia and V. trifolia have relatively low genetic diversity. The highest haplotype diversity occurred in population WN for V. rotundifolia and 3_YN for V. trifolia, indicating that Wanning of Hainan Province and Xishuangbanna of Yunnan Province may be the centers of biological diversity of V. rotundifolia and V. trifolia, respectively.

The length of the ITS sequence was 334 bp for 177 individuals, but only two haplotypes and one polymorphic site were found in V. rotundifolia and V. trifolia (Fig. 3, Table 3). In 23 sampling sites for V. rotundifolia, the total haplotype diversity (h_d,nr) was 0.440, the range of haplotype diversities within population was from 0–0.667; the nucleotide diversity (π_d,nr) was 0.00000; and the range of nucleotide diversities within population was 0–0.00200. When we excluded six poorly sampled populations, the ITS total h_d,nr and π_d,nr slightly changed to 0.443 and 0.00000, respectively. Within the four V. trifolia sampling sites, the total values of h_d,nr and π_d,nr were 0.516 and 0.00000, respectively. These results indicated that based on the ITS nuclear sequence, V. rotundifolia and V. trifolia both have a low genetic diversity.

Furthermore, PermutCpSSR_1.2.1 analysis based on cpDNA sequences deduced that the h_T, V_T, h_S, and V_S values for V. rotundifolia were 0.281, 0.285, 0.225, and 0.285, respectively, and those for the V. trifolia populations were 0.869, 0.921, 0.690, and 0.553, respectively, indicating that V. rotundifolia populations also have low genetic diversity, and that the genetic diversity of the V. trifolia populations was much higher than that of V. rotundifolia populations. The parameters of h_T, V_T, h_S, and V_S based on ITS sequence could not be calculated because only two haplotypes were found in V. rotundifolia and V. trifolia.

AMOVA analysis based on the cpDNA sequences revealed that a total of 79.19% of the variation was between the V. rotundifolia and V. trifolia populations, and 80.36% of the total cpDNA variation existed within V. rotundifolia populations. In addition, 78.25% of the variations based on cpDNA were found between V. rotundifolia and V. trifolia populations (P < 0.01; Table 3), indicating that there was a significant genetic differentiation between these two species. Based on ITS sequences from combined V. rotundifolia and V. trifolia populations, AMOVA analysis showed that 6.29% of the variation existed between the species, 6.70% of the variation was caused by differences between the populations within the species, and 87.02% of the variation existed within the populations (Table 4).

Population structure and phylogeographical analysis

The phylogeographical analysis based on the cpDNA sequences found that the N_ST value was significantly greater than the corresponding G_ST value (0.751 vs. 0.436, P < 0.05), showing a noticeable phylogeographical structure across 25 V. rotundifolia and V. trifolia populations, and the eight haplotypes displayed a clear geographical pattern. However, the N_ST value was higher than its corresponding G_ST value (0.462 vs. 0.201, P > 0.05) in 21 V. rotundifolia populations, indicating that there was an unclear cpDNA phylogeographical structure. As shown in Fig. 4, statistical parsimony analysis found a single haplotype network based on cpDNA samples of V. rotundifolia and V. trifolia populations. Furthermore, the TCS 1.21 network analysis of cpDNA haplotypes revealed the relationship of the interior (ancestral) and the tip (derived) haplotypes. Haplotype H4 was inferred as the ancestral haplotype, as determined by outgroup weight based on haplotype positions in the network. H1, H7, and, H8 were shared only by V. rotundifolia. H2–H6 were shared only by V. trifolia. H1 was the most frequent and widely distributed haplotype in 17 of the 21 populations of V. rotundifolia.

The phylogenetic relationships between the eight cpDNA haplotypes were assessed under Bayesian inferences drawn using V. negundo as an outgroup. The result showed that the phylogenetic tree of the eight cpDNA haplotypes had a comb-like structure (Fig. 5). This result may have been caused by insufficient information sites, or by the rapid expansion of V. rotundifolia and V. trifolia. We could not perform the haplotype network and phylogenetic analyses based on the ITS sequences because there were only two haplotypes.

Divergence times and demographic history

BEAST, a program for Bayesian analysis, was used to estimate the divergence times. The result showed that the effective sample size (ESS) value ranged from 919 to 1,070 for all the nodes, the divergence time between V. rotundifolia and V. trifolia was 0.23 mya (0.096–0.396, 95% highest posterior density (HPD)) (Fig. 6). Hence, the divergence between V. rotundifolia and V. trifolia appeared during the late Pleistocene era.

Based on the cpDNA data, mismatch distribution analysis was performed to test whether the current expansion occurred in V. rotundifolia populations and V. trifolia populations separately. The results found that V. rotundifolia and V. trifolia populations each had a relatively smooth unimodal distribution, which was consistent with distribution under an exponential growth rate (Fig. 7), indicating that the two species had a rapid demographic expansion. In addition, among the 21 V. rotundifolia populations, Tajima’s D and Fu’s test gave apparently positive results, as evidenced by the value of D and Fs (D =  − 0.3997, Fs = 0.2787; both P >0.05), indicating that V. rotundifolia has not passed through a recent demographic expansion. In the four V. trifolia population, Tajima’s D (D = 1.8690, P > 0.05) and Fu’s Fs (Fs = 0.95759, P > 0.05) tests also showed no demographic expansion (Table 5). This result is inconsistent with that derived from the mismatch distribution plots. This discrepancy might be caused the insufficient number of variable sites provided by only two pairs of cpDNA regions (trnG-trnS and psbA-trnH).

A large-scale distribution and low genetic diversity of V. rotundifolia

V. rotundifolia is native to seashores from eastern India to Hawaii and from Japan to Australia, and can also be found in Brazil, Mauritius, Bangladesh, and Sri Lanka (Moldenke, 1977; Munir, 1987). In China, V. rotundifolia grows on the beach, seaside, and lakeside, mainly along the eastern coastal region (Wang & He, 2013). Based on our fieldwork, we investigated 14 sampling sites throughout the coastal areas and three sampling sites at lakesides in inland of China. The whole distribution areas of V. rotundifolia in China were covered and also extended to Korea and Japan, including sites with few samples. The latitude and longitude of these locations range from 18°44′4.68 to 39°51′26.24″  and from 9°28′37.83″ to 140°0′46.99″, with the sampling distance crossing almost 3,000 km. However, the present study found that V. rotundifolia has extremely low levels of genetic diversity (h_d,cp = 0.360, π_d,cp = 0.00010; h_d,nr = 0.440, π_d,nr = 0.00000) compared with most other plants (Yan, 2007; Chen et al., 2008; Chen et al., 2012; Xing et al., 2017). Such low level of nucleotide diversity has also been found in the coastal plant Rhizophora (Rhizophoraceae), with nucleotide diversities for trnG-trnS and trnH-rpl2 being 0.04 ± 0.02 and 0.03 ± 0.02, respectively, and π_nr = 0.02 ± 0.01 (Lo, Duke & Sun, 2014). However, for V. rotundifolia, using allozyme markers (Yeeh et al., 1996) and ISSR markers (Hu et al., 2008), moderate genetic diversity was reported, which was slightly higher than that observed in the present study. This discrepancy might be attributed to differences molecular markers used. In the present study, even when excluding poorly sampled populations, the secondary analysis also indicated that both the chloroplast and nuclear sequences have extremely low nucleotide diversity in V. rotundifolia.

The large-scale distribution of V. rotundifolia with a low level of genetic diversity may be attributed to three reasons: (1) Long-distance dispersal (LDD) by sea-drifted fruit: long-distance dispersal can result in the homogeneous distribution of DNA haplotypes at the spatial scale (Miryeganeh, 2013). Our results revealed that 17 of the 21 V. rotundifolia sampling sites, including the most distant populations covered by the geographical distribution range of sampling, such as XL (Xinglong of Hainan province) and JP (Japan), shared the common haplotype H1. This wide distribution of haplotypes may be caused by long-distance dispersal of this plant’s fruit. The V. rotundifolia fruit is covered with a thick hydrophobic coating that can resist water penetration, rendering it able to float in the ocean, drift onto the beach, and disperse over long distance via sea drift (Munir, 1987; Cousins et al., 2010). Therefore, this rapid long-distance dispersal ability of V. rotundifolia fruit may be responsible for the low genetic diversity of this plant. (2) The similar ecological environment: life history traits and environmental factors affect the genetic diversity and structure of species (Loveless & Hamrick, 1984; Nybom, 2004). V. rotundifolia usually grows on beaches, sand dunes, and rocky shorelines at low elevations and can tolerate the highly salt, drought, full sun, and sandy or well-drained soils (Hawaiian Native Plant Propagation Database, 2018; GeoResources Institute (GRI), 2018). These similar ecological niches could limit genetic differences and lead to a low genetic diversity of V. rotundifolia. (3) The asexual reproductive mode: this species proliferates by sexual reproduction, as well as clonal propagation through the elongation of rhizomes and root systems (Gresham & Neal, 2004). Clonal reproduction usually leads to the loss of genetic diversity within populations (Cook, 1983).

Population structure and demographic history

Based on the cpDNA data, the results showed that all populations of V. rotundifolia and V. trifolia had a significant phylogeographical structure (G_ST = 0.436, N_ST = 0.751; N_ST >G_ST, P <0.05). The current distribution of haplotypes may be caused by climate oscillation during the Quaternary period, which resulted in further large-scale migration of most plants and animals, and the subsequent accumulation of genetic variations and specific phylogeographical structures (Hewitt, 2000; Hewitt, 2004). V. rotundifolia populations showed an insignificant phylogeographical structure. (G_ST = 0.201, N_ST = 0.462; N_ST > G_ST, P >0.05). Furthermore, AMOVA analysis showed low genetic differentiation among V. rotundifolia populations, indicating that rapid spread by sea currents may have played a key role. The unclearly phylogeographical structure and the low genetic differentiation in V. rotundifolia populations may relate to the rapid long-distance dispersal by sea drifts. This is not consistent with the report by Hu et al. (2008), in which ISSR analysis showed a relatively high genetic differentiation (G_ST = 0.587) among populations of V. rotundifolia; the authors stated that such characteristics of V. rotundifolia are likely attributed to its sexual/asexual reproduction and limited gene flow. These contradictory results may have been caused by the different gene markers used (ISSR vs. cpDNA).

Estimated divergence times indicated that the genetic divergence between V. rotundifolia and V. trifolia occurred about 0.23 mya (0.096–0.396, 95% HPD), indicating a relatively short divergence time, corresponding to the late Pleistocene era, before the onset of the last glacial maximum (LGM). Following this divergence, long-distance dispersal may have been a major historical process for V. rotundifolia, because they are present throughout the Pacific, including the coastal areas of the continents and many islands. Additionally, we detected a recent population expansion of V. rotundifolia based on the apparent mismatch distribution, showing that V. rotundifolia had undergone a long-distance dispersal expansion in the short term.

The ancestral haplotypes are located in a central position within the phylogeographical network (Crandall & Templeton, 1993). Haplotype H4 is located in the center of the haplotype network, and is the closest haplotype to H9 (outgroup), and H1 has evolved from H4. Therefore, we suggest that H4 is the ancestral haplotype, which was detected in two populations (3_WC and 3_YN) in V. trifolia. Populations 3_WC and 3_YN are from the southern part of China and have higher haplotype diversities in all populations. Moreover, the results of the TCS network analysis also showed that the haplotypes shared by V. rotundifolia were deduced from those of V. trifolia. Thus, we speculated that Yunnan and Wenchang regions might be two origin areas of V. rotundifolia. During the Quaternary period, V. rotundifolia expanded to eastern China rapidly.

The analysis of cpDNA data in this study showed that the V. rotundifolia and V. trifolia populations have no shared haplotypes. In addition, analysis of molecular variance (AMOVA) showed a significant difference in genetic differentiation between V. rotundifolia and V. trifolia (F_ST,cp = 0.79195). These findings supported the view that V. rotundifolia and V. trifolia are two separate species.

Conservation strategy

It is essential to study the morphological variation, genetic diversity, and population structure to provide basic information on plant conservation (Feng, Wang & Gong, 2014; Fan et al., 2017). Climate change, the rapid growth of the human population, and economic development all contribute to the deterioration of V. rotundifolia’s habitat, reducing the number of individuals in some populations. Therefore, V. rotundifolia resource protection is urgently required. The management priority is to conserve the population of the species that has the greatest diversity. Our results showed that the WN, XL, and QD populations have the highest haplotype diversity and nucleotide diversity in V. rotundifolia; therefore, conservation priority should be considered for these populations. Hu et al. (2008) proposed that the Xinjian population of V. rotundifolia from Jiangxi province of China had a considerably high genotypic diversity index and should be prioritized for protection measures.

Furthermore, population size is an important factor threatening the existence of species, and for many wildlife species, conservation should prioritize the smallest populations. Our field surveys found that most extant V. rotundifolia populations in China face a serious threat of extinction because of the limited number of individuals. Therefore, initial conservation measures for V. rotundifolia should be considered to increase the population size and genetic diversity. Among the extant populations, there are fewer than 10 individuals in the DL, LYG, and XL populations. Hence, we need to increase the population size for in situ conservation with high priority in these populations. In addition, habitat protection is also an important concern in V. rotundifolia conservation because of habitat damage by increased human activities and over-exploitation of the plant as a medical resource. Therefore, it is necessary to establish local nature reserves, especially for those small-sized and seriously disturbed populations. Ex situ conservation strategies, including germplasm collection, culture in botanical gardens (Kunming, Beijing), or reintroductions at appropriate areas with similar habitats, are also effective measures to preserve the genetic resources of V. rotundifolia. When the ex situ conservation is put in practice, samples should be collected from as many individuals as possible, because a large portion of the genetic diversity exists within, rather than among, populations. Sample collection from the WN, XL, and QD populations is recommended. Our experience, together with the results from previous studies, suggested that individuals in a population should be collected at >20 m spatial intervals to avoid collecting individuals with identical or similar genotypes (Hu et al., 2008).