Genetic diversity and genetic structure of Decalobanthus boisianus in Hainan Island, China

Abstract Decalobanthus boisianus is a native plant of Hainan Island, China, which has caused considerable damage to tropical forest ecosystems in recent decades. Understanding the genetic diversity and structure of this species can facilitate uncovering the molecular mechanism of its invasive ability. Here, we collected 77 individuals of D. boisianus spanning 8 distribution areas with a gradient of human disturbance intensity (i.e., low, moderate, and high disturbance intensity groups) to assess patterns of genetic diversity and structure using inter simple sequence repeat (ISSR) markers. We found that a total of 220 loci were scored with 13 primers using ISSR methods, and that 198 loci were polymorphic. The genetic diversity of D. boisianus among these eight forests decreased with increasing human disturbance intensity. Over 70% of the total genetic variation was present within populations, while less than 30% of variation was found among populations. There was a high gene flow (1.27) among them due to a lack of effective geographic barriers. The mean Nei's genetic distance of D. boisianus populations was found to be relatively small (i.e., 0.07), and the average genetic similarity of the eight populations was high (i.e., 0.93). Our findings indicate that the genetic diversity of D. boisianus correlated to human disturbance density, and that D. boisianus populations in Hainan Island have frequent gene exchange. We suggest that reduce deforestation to decrease human disturbance may be a good way to prevent the invasion of D. boisianus.

Environmental pressure is seen as an important factor accounting for plant reproductive strategies, which can further influence levels of genetic diversity in the species (Cook, 1985;Eriksson, 1997;Philbrick & Les, 1996;Sinclair et al., 2016). For example, sexual reproduction typically increases genetic diversity in a population. In contrast, clonal reproduction allows species to rapidly colonize available habitats, but often leads to a decline in genetic diversity (Ellstrand & Roose, 1987;Hamrick, Linhart, & Mitton, 1979;Schmid, Bazzaz, & Weiner, 1995). Several studies have showed that environmental pressures often have negative consequences for genetic diversity. Rusterholz, Kissling, and Baur (2009) demonstrated that soil disturbance induced by human trampling reduced the frequency of sexual reproduction for Anemone nemorosa populations, resulting in lower levels of genetic diversity for these populations. Likewise, Aveliina et al. (2009) Wang et al., 2005;Wang, Peng, Li, & Zhou, 2009;Staples, 2010). This species has recently moved from genus Merremia Hall. f. to Decalobanthus Ooststr since the former now only included polyphyletic species because of the weak support for monophyly and strong support for polyphyly of this genus (Simões & Staples, 2017 (Lian et al., 2007;Liu, 2007;Wang et al., 2009;Wu, Liang, Chen, Li, & Cao, 2007). A number of studies assumed that viable seeds of the species can be transported over streams (Wu et al., 2007;Wang et al., 2009), which helps D. boisianus spread long distance. The species is thus characterized by rapid growth and a strong competitive ability , which allow it to easily invade plantation areas, shrublands, and secondary forests with low canopy density (Wang et al., 2009;, covering more than 2000 ha areas in Hainan Island. After invasion, it often kills a large number of trees and understory plants with low canopy coverage (Huang et al., 2013). Although researchers suggest that D. boisianus can be taken advantage and used for vegetation recovery in barren mountains, as well as leaves feeding livestocks (Sun, Shen, Wan, & Xie, 2006), it has been listed as a major forest harmful pest in China due to the serious damage to secondary forests that the species causes (State Forestry Administration, China).
In this study, we explored genetic diversity and structures of D. boisianus in eight Nature Reserves in Hainan Island. According to field surveys and interviews with local residents, we found that D. boisianus distributed in areas with relatively high human disturbance intensity. Human disturbance intensity has been proved to affect genetic diversity of plant species (Aveliina et al., 2009;Ledo & Schnitzer, 2014;Rusterholz et al., 2009). Increases in human disturbance intensity across the eight distributed areas of D. boisianus likely led to a decrease in its genetic diversity. We thus hypothesized that genetic diversity of D. boisianus was high in areas with the low human disturbance intensity, and vice versa.  (Table 1). In light of the human disturbance intensity, we categorized D. boisianus in these eight forests into three groups, including low disturbance intensity group (WZS), moderate disturbance intensity group (YGL, BWL, JFL, and DLS), and high disturbance intensity group (LMS, BSL, and GSL).

| DNA extraction and PCR amplification
DNA was extracted from our 77 samples using the Plant DNA Isolation Kit (Foregene Co., Ltd., Chengdu, China). We followed the manufacturer's protocol using 25-30 mg of dried leaf material. DNA quality and quantity were determined visually by comparisons with the DNA marker DL2000 on 1.0% (W/V) agarose gel electrophoresis. Samples were then stored at −20°C prior to PCR amplification.
A set of 100 ISSR primers was synthesized by SinoGenoMax Co., Ltd. according to the sequences obtained from the University of British Columbia (Biotechnology Laboratory, University of British Columbia, primer set #9: http://www.biotech.ubc.ca/services/naps/ primers/Primers.pdf). An initial experiment was performed to determine the suitable primer and reaction conditions. One sample randomly selected from each eight D. boisianus populations to test the preliminary number of polymorphic loci with 100 primers and to optimize the reaction and amplification procedure for PCR. Finally, we obtained optimal reaction and amplification procedures as well as 13 primers that produced a high number of variable and readable loci ( Table 2) Table 2), extension at 72°C for 60 s, and then a final extension at 72°C for 6 min, followed by cooling at 4°C until recovery of the samples ( Table 2)

| Data analysis
Only bands that were clear and reproducible were used to construct data matrices. The amplified bands were scored as presence (1) or absence (0) of a particular band for each primer. These data were then used to assemble the data matrix of ISSR phenotypes.

| The genetic diversity of the eight D. boisianus populations
We calculated the proportion of polymorphic loci (PPL), Shannon's diversity index (I), Nei's gene index (H), the observed number of alleles (Na), and the effective number of alleles (Ne) to assess levels of genetic diversity among our populations using the POPGENE program v. 1.32 (Francis, Rong, & Boyle, 1999). The PPL is an important indicator of genetic diversity and of the adaptability of populations to a particular environment. Shannon's index is based on the band phenotypic frequency and is used to estimate the genetic diversity  (King & Schaal, 1989). H is an index to reflect the gene diversity based on Hardy-Weinberg hypothesis (Nei, 1972). Other indices such as Na and Ne indicate the observed and effective numbers of alleles that are maintained in the populations (Johnson, 1974). Here, we compared the differences in genetic diversity of the eight D. boisianus populations across different human disturbance intensity using ANOVA analysis.

| The genetic structure of the eight D. boisianus populations
Genetic differentiation among population (Gst) represents the portion of the total genetic diversity found among populations (Nei, 1972), and the number of migrants (gene flow, Nm) measures the degree of genetic differentiation among populations (Wright, 1950). Here, Gst, Nm, total genetic diversity (Ht), and gene diversity within populations (Hs) were used to determine the genetic differentiation among the eight populations of D. boisianus. A matrix of genetic similarity and Nei's genetic distances were used to depict genetic relationship among the eight D. boisianus populations (Nei, 1972). Populations with high genetic similarity and low Nei's genetic distances value indicate a close genetic relationship. All parameters were calculated using the POPGENE program v. 1.32 (Francis et al., 1999).
We also explored spatial structures of D. boisianus with a Bayesian model-based cluster analysis, using the STRUCTURE program version 2.3.4 (Pritchard, Stephens, & Donnelly, 2000) and calculated the appropriate number of clusters of populations (K) (Evanno, Rengaut, & Gouget, 2005). Groups were chosen that best fit the genetic variability observed in the dataset, irrespective of the number of populations sampled  (Andreakis, Kooistra, & Procaccini, 2009). After initial pilot runs of variable burn-in and run-length, 20 independent runs were performed at K = 2-8 with 10,000 MCMC repetitions and a burn-in period of 10,000 iterations, using no prior information and assuming correlated allele frequencies and admixture model in this study. The STRUCTURE output was further interpreted by STRUCTURE HARVESTE (Earl & VonHoldt, 2012). The CLUMPP version 1.1 (Jakobsson & Rosenberg, 2007) were used with greedy algorithms, with 1,000 random input orders and 1,000 repeats to calculate the average pairwise similarity of runs. The clustered output was visualized using the software Distruct version 1.1 (Rosenberg, 2004). Finally, we conducted cluster analysis for the

| Genetic diversity
We also found the genetic diversity varies significantly with human disturbance intensity, with the high, medium, and low genetic

| Genetic structure
Among the eight populations, the total gene diversity index (H t ) and  populations (Hamrick et al., 1979;Loveless & Hamrick, 1984), to alter genetic diversity of D. boisianus populations (Cook, 1985;Eriksson, 1997;Sinclair et al., 2016). For example, D. boisianus in LMS, BSL and GSL with high-intensity human disturbances may allocate resources to clonal reproduction, which helps it rapidly expand to novel habitats. But the clonal reproduction produces the same offspring gene and leads to low genetic diversity (Ellstrand & Roose, 1987;Philbrick & Les, 1996;Kudoh, 1999). In contrast, low-intensity human disturbances allow WZS population to allocate more resources to sexual reproduction, thereby maintaining higher level of genetic diversity (Prentis et al., 2008;Silander, 1979

| Genetic structure of D. boisianus in Hainan Island
In our study, both Nei's genetic differentiation index (Gst = 0.28) and AMOVA values (24.88%) indicated that there was a large proportion of genetic variation (over 70%) present within the eight D. boisianus populations, while less than 30% of genetic variation was observed among the eight populations. These results are consistent with the previous reports from other invasive species like Carduus acanthoides L. (Bohumil, Petr, Dana, Petr, & Ivana, 2009), Ambrosia artemisiifolia L. (Kočiš et al., 2015), and Spartina densiflora Brongn. (Castillo et al., 2018) but are less than most clonal invasive species, such as Eupatorium catarium Veldkamp (Li, Li, & Liu, 2014), Praxelis clematidea R.M. King & H.Rob. (Wang, Huang, Downie, Chen, & Chen, 2015), Galinsoga quadriradiata Ruiz & Pav. (Li, Qi, Yao, & Liu, 2015), Mikania micrantha H.B.K. (Li, Dong, & Zhong, 2007), and Alternanthera philoxeroides Griseb. (Ye, Li, Cao, & Ge, 2003). These results indicate that a very low genetic differentiation of the eight D. boisianus populations is observed on Hainan Island. One possible reason is that the effective gene flow (1.27) among the eight populations result from the high dispersal ability of D. boisianus to against barriers throughout the expansion of this species (Opedal et al., 2017;Wright, 1950). Indeed, the small mean value of Nei's genetic distance (0.07) and the high mean genetic similarity index (0.93) both indicate the close relationship among D. boisianus populations. Although little is known for the dispersal mode of this species, many studies have assumed that viable seeds of D. boisianus may be transported over streams to achieve long-distance transmission (Wu et al., 2007;Wang et al., 2009

| CON CLUS ION
Generally, we found that genetic diversity of D. boisianus was correlated to the human disturbance intensity, which has been proved to be a major factor governing plant invasion. High dispersal ability and adaptive potential of D. boisianus may be important factors that were attributed to its large distances expansion and cause serious damage to the ecosystems. So far, artificial removal and chemical controls have been implemented to kill and prevent D. boisianus in the study areas, but were seldom effective. We suggest reducing deforestation to decrease human disturbance to forests for one hand.
In addition, effective efforts should be taken to remove D. boisianus, and restore the secondary forests.

ACK N OWLED G EM ENTS
We thank Chao Li, Yikang Chen, and Qing Chen for their assistance in field work. Financial supports came from National Natural Science