Genetic and epigenetic changes during the invasion of a cosmopolitan species (Phragmites australis)

Abstract While many introduced invasive species can increase genetic diversity through multiple introductions and/or hybridization to colonize successfully in new environments, others with low genetic diversity have to persist by alternative mechanisms such as epigenetic variation. Given that Phragmites australis is a cosmopolitan reed growing in a wide range of habitats and its invasion history, especially in North America, has been relatively well studied, it provides an ideal system for studying the role and relationship of genetic and epigenetic variation in biological invasions. We used amplified fragment length polymorphism (AFLP) and methylation‐sensitive (MS) AFLP methods to evaluate genetic and epigenetic diversity and structure in groups of the common reed across its range in the world. Evidence from analysis of molecular variance (AMOVA) based on AFLP and MS‐AFLP data supported the previous conclusion that the invasive introduced populations of P. australis in North America were from European and Mediterranean regions. In the Gulf Coast region, the introduced group harbored a high level of genetic variation relative to originating group from its native location, and it showed epigenetic diversity equal to that of the native group, if not higher, while the introduced group held lower genetic diversity than the native. In the Great Lakes region, the native group displayed very low genetic and epigenetic variation, and the introduced one showed slightly lower genetic and epigenetic diversity than the original one. Unexpectedly, AMOVA and principal component analysis did not demonstrate any epigenetic convergence between native and introduced groups before genetic convergence. Our results suggested that intertwined changes in genetic and epigenetic variation were involved in the invasion success in North America. Although our study did not provide strong evidence proving the importance of epigenetic variation prior to genetic, it implied the similar role of stable epigenetic diversity to genetic diversity in the adaptation of P. australis to local environment.

Among all identified epigenetic mechanisms, including histone modifications, DNA methylation, and small noncoding RNAs, DNA methylation is relatively stable with transgenerational heritability that can be independent of heritable genes (Bird, 2007;Eichten, Schmitz, & Springer, 2014), with the result that DNA methylation has attracted the most attention in epigenetic studies in ecology and evolution (Alvarez, Schrey, & Richards, 2015). The pattern of DNA methylation can affect ecologically important phenotypes and plasticity (Herrera & Bazaga, 2013;Nicotra et al., 2015;Zhang et al., 2013) and may play a significant role in adaptation to various habitat conditions (Foust et al., 2016;Richards et al., 2012;Schulz et al., 2014). For example, a naturally occurring epiallele named "NMR19-4" has been discovered in Arabidopsis accessions, and its DNA methylation status is inheritable and independent of genetic variation (He et al., 2018). This epiallele controls leaf senescence and associates with local climates. Moreover, an easy and efficient technique named methylation-sensitive amplification polymorphism (MS-AFLP) has been widely used to assess DNA methylation status at a great number of random anonymous loci across the entire genome in nonmodel species without sequenced reference genomes (Alonso, Pérez, Bazaga, Medrano, & Herrera, 2016;Schrey et al., 2013).
Phragmites australis (common reed) has a worldwide distribution and has been considered as a model organism for studying plant invasions (Meyerson, Cronin, & Pyšek, 2016) according to the criteria for identifying model organisms in invasion science adapted by Kueffer, Pyšek, and Richardson (2013). Two main introduced lineages in North America have been detected (Guo, Lambertini, Li, Meyerson, & Brix, 2013;Meyerson, Lambert, & Saltonstall, 2010). One, known as Haplotype M (hereafter "INT"), has spread dramatically across much of the North America, especially in the Great Lakes regions (Saltonstall, 2002). The other, represented by Haplotype M1 and I (hereafter "DELTA" and "LAND"), was native to Mediterranean region, sub-Saharan Africa, and the Middle East and has expanded along the Gulf Coast of the United States and in the northwest of South America Lambertini, Mendelssohn et al., 2012). The introduced population of P. australis had a higher level of genetic diversity and heritable phenotypic variation in its invasive range than in parts of its native range, as multiple and uncontrolled immigration events may have occurred from different European regions to North American (Lavergne & Molofsky, 2007). Some heritable traits and ecophysiological differences in the common reed may contribute to invasion success but hitherto cannot be explained by particular genetic lineages (Mozdzer, Brisson, & Hazelton, 2013). However, very few investigations of natural epigenetic variation of P. australis have been reported, and previous studies were conducted just at a small scale such as in midcoast Maine (Spens & Douhovnikoff, 2016) and in the Songnen Prairie of China (Qiu, Jiang, & Yang, 2016). The understanding of potential epigenetic mechanisms in the invasion of P. australis is still very limited.
In this study, we first compared patterns of genetic and epigenetic variation in P. australis grown in a common garden collected from around the globe using AFLP and MS-AFLP methods.
We collected samples from native and introduced groups in North America and samples of P. australis from the native geographic regions from where the introduced groups were believed to have originated, in order to be able to detect the direction and the degree of genetic and epigenetic changes during the invasion of P. australis. We tested the following hypotheses: (1) introduced groups increased epigenetic diversity to compensate for the loss of genetic diversity in response to heterogeneous environments, and (2) introduced groups could acquire epigenetic variation similar to that of the native groups faster than genetic variation. In a word, we would expect rapid epigenetic adaptation of introduced groups to new environments before slow genetic accommodation occurs.

| Plant material
All samples were collected in a common garden at Fanggan Research Station of Shandong University in Shandong Province, China (36°26′N, 117°27′E), where they had been growing for 2 years. This station has a typical warm temperate monsoon climate with a hot rainy summer and a cold dry winter under climate regulation effects of forest (Sun, Li, Guan, Liu, & Zhang, 2017). To avoid developmental epigenetic variation, all leaf samples for molecular analyses were picked at the same position of the plant before flowering phase. A total of 75 specimens of P. australis, representing eight phylogeographic groups, were sampled to investigate genetic and epigenetic variation ( Table 1). The invasive P. australis (INT) in the Great Lakes region was likely introduced from the native European counterpart (EU) based on haplotype evidence, and hybridization between the native and invasive groups in nature has not yet been documented in North America (Saltonstall, 2002(Saltonstall, , 2003Saltonstall, Lambert, & Rice, 2016). The LAND and DELTA groups of P. australis are two of five identified phenotypes growing sympatrically in the Gulf Coast region. The DELTA group is a typical introduction from the Mediterranean region (MED) and has become one predominant lineage in the Gulf Coast region .
The origin of the LAND type is still debated, and we treat this type as a native group in our study due to the following reasons: (1) LAND type has existed for a longer time than DELTA type as the genetic evidence supports an ancient introduction for LAND type and a recent introduction for DELTA type , (2) LAND type is not invasive in this area with only scattered occurrences, and (3) we just used LAND type as a reference for the introduced DELTA group, which coexisted under the homogeneous environment of the Gulf Coast region. Phragmites mauritianus may be a hybrid of LAND type origin . Individuals from Australia (FEAU) were also analyzed as P. mauritianus from Tropical Africa (TA) as an outgroup.

| AFLP and MS-AFLP analysis
Genomic DNA was extracted from fresh leaf tissue according to the cetyltrimethylammonium bromide (CTAB) method (Doyle & Doyle, 1987). The yield and quality of extracted DNA were determined with both 0.8% agarose gels and a microscope spectrophotometer. We

| Statistical analysis
To determine the DNA methylation status of every locus from the fragment presence/absence scores of both EcoRI-MspI and EcoRI-HpaII reactions, the R-based statistical package "msap" was carried out (Pérez-Figueroa, 2013). In the selected scoring strategy, the presence of both EcoRI-MspI and EcoRI-HpaII bands is an unmethylated state, and the presence of only one band, either EcoRI-HpaII or EcoRI-MspI, represents a methylated state.
However, the absence of both EcoRI-HpaII and EcoRI-MspI bands may reveal either the full methylated state or genetic variation.
As these samples were from different ramets with expected high genetic diversity and we cannot ignore the possibility of genetic change in the specific locus, so we adopted the approach considering the absence of both EcoRI-HpaII and EcoRI-MspI bands as an uninformative state (i.e., missing data) with the function msap The global methylated cytosine was established using the sum of the internal methylated and the external hemimethylated cytosines.
The genetic and epigenetic diversity of every population was determined using the population genetic software GenAlEx 6.5 (Peakall & Smouse, 2012) in

| DNA methylation level
Similar DNA methylation levels among all groups were determined through the MS-AFLP markers except for the NAT and TA groups, which exhibited a relatively high level of uninformative state (i.e., full methylation or absence of target) (Figure 1). The DNA methylation levels of each group calculated from the MSL data revealed that the unmethylated sites were the most common among all populations of P. australis, followed by internal cytosine methylated and hemimethylated, and uninformative sites. In addition, we did not detect significant differences across ploidy levels in our samples of P. australis when we observed these uninformative loci as full methylated (ANOVA: p = .607), missing (ANOVA: p = .645), or null data (ANOVA: p = .897) (Figure 2).

| Genetic and epigenetic diversity
We obtained 154 polymorphic AFLP loci and 151 MSL loci, and diversity indices showed high levels of both genetic and epigenetic variation ( Table 2). Among all groups, LAND had the highest level of genetic diversity while EU had the highest level of epigenetic diversity. Almost all groups of P. australis displayed a higher value of all three diversity indices for epigenetic than genetic variation except for LAND where uHe for genetic variation (uHe = 0.40) was slightly higher than that for epigenetic variation (uHe = 0.39). In the Gulf Coast region, the native group (LAND) had a much higher level of genetic diversity than the introduced group (DELTA) but a lower level of epigenetic diversity using the index uHe. The DELTA group had comparable genetic and epigenetic diversity with one possible original group (MED). Compared with another original group, the DELTA group had lower genetic diversity but higher epigenetic diversity.

| Genetic and epigenetic structure
AFLP and MS-AFLP revealed similar variances among and within groups (see Table 3). Most of the genetic and epigenetic variation existed within rather than among groups, but genetic variance among groups (Φ ST = 0.182, p = .001) was more than epigenetic variance (Φ ST = 0.072, p = .001). In

| D ISCUSS I ON
This is not an in situ study as all samples were collected from a common garden, which must affect the status of some unstable epigenetic variation. Nevertheless, our results are practical and valuable for the following reasons: (1) it is such a short time in the common garden compared to a hundred years of introduced history that most relatively stable epigenetic variation could remain; (2) the common environment has few chances to cause new epigenetic differences among all individuals. In other words, common garden can separate plastic and heritable components of epigenetic variation (Kilvitis et al., 2014;Richards et al., 2012), and samples in our study still maintain most transgenerational epigenetic variation in which ecologists and evolutionary biologists are interested (Verhoeven, von-Holdt, & Sork, 2016). Moreover, the common garden allowed us to control well the developmental stage, which is the other important factor influencing epigenetic variation besides the environmental one.

| Massive epigenetic variation is correlated with genetic variation
Consistent with the previous study (Spens & Douhovnikoff, 2016), we found a significant difference in epigenetic variation between introduced and native groups. Furthermore, our result revealed that epigenetic structure showed a positive correlation with genetic structure, and epigenetic (as well as genetic) distance also provided coincident evidence for the identified migration of P. australis Saltonstall, 2002), but epigenetic distance compared with genetic distance was clearly reduced and possibly more related to individual microhabitats. For example, coefficients of genetic differentiation supported the previous haplotype-based conclusion that P. mauritianus contributed to the hybridization of the LAND type of P. australis , but the epigenetic evidence in our study cannot (Table 3).
The cytosine methylation level in P. australis is relatively steady among geographic and genomic groups, and the detected differences were more due to genetic variation. The reason for the relatively abundant loci with uninformative state in the NAT and TA groups was more likely to be absent from the genetic target rather than full methylation, as the two groups were very different from the other groups in genetic variation (Table 2). Polyploidization is an important evolutionary event in plants and may result in substantial changes in genomewide methylation, but we did not detect the expected increased cytosine methylation with ploidy series (Figure 2).
There are very few studies comparing intraspecific DNA methylation levels across wild plants differing in ploidy, and this type of  ) and DELTA group was similar to MED group (c,d).However, INT group was not closer significantly to NAT group in epigenetic variation (b) relative to genetic variation (a), and DELTA group did not significantly tend toward LAND group in epigenetic variation (c) relative to genetic variation (d) study cannot set apart how much the DNA methylation results were from ploidy versus environmental factors (Alonso, Balao, Bazaga, & Pérez, 2016). MS-AFLP actually cannot provide a quantitative global DNA methylation forthright , and the global DNA methylation level in this species should be established by high-performance liquid chromatography (HPLC) to uncover the role of DNA methylation in natural polyploidizations. Some cytotypes of P. australis can naturally coexist but others cannot (Table 1), so ploidy effects are counteracted by the local environment sometimes. However, the ploidy differences within the samples had little effect on genetic and epigenetic changes during natural invasions in this study.

| The epigenetic diversity may compensate for limited genetic diversity
Many indices have been developed to describe diversity, and all indices perform congruously in common situations; but they may conflict in some cases. In our study, we calculated three indices to compare the genetic and epigenetic diversity among all groups.
When divergence occurred, the uHe index was preferred because of the limited number of individuals from each group (Nei, 1978).
Consistent with most previous epigenetic studies of natural population differentiation (Choi, Roy, Park, & Kim, 2016;Foust et al., 2016;Kim, Im, & Nkongolo, 2016;Qiu et al., 2016;Schulz et al., 2014), more epigenetic variance existed within groups than genetic variance. One general explanation for this is that a population with limited genetic diversity, especially after genetic drift, can extend its ecological niche through epigenetic variation, which is potentially sensitive to environmental stimulation (Richards, 2006). As some epigenetic variation is affected by developmental factors instead of the environment, common epigenetic variations within populations could be detected. Therefore, numbers of specific loci with ecological effect can and should be determined in rigorous research through generalized linear models or other statistical methods.
Invasive species can increase their genetic diversity through interbreeding with local or other introduced populations in response to diverse habitat environments (Genton et al., 2005;Lavergne & Molofsky, 2007;Rosenthal et al., 2008). Cross-pollination experiments have demonstrated that that native and introduced populations of Phragmites can hybridize (Meyerson, Viola, & Brown, 2010), but few natural hybridization cases have been detected between the native and introduced lineages (Saltonstall, Castillo, & Blossey, 2014). In the Gulf Coast region, there may be very strong gene flow within and/or among lineages (Meyerson, Lambertini, McCormick, & Whigham, 2012), which caused a relatively high level of genetic diversity. The LAND and DELTA groups were established by independent colonization events , and the LAND group possessed a higher level of genetic diversity than the DELTA group which suggested introduced DELTA group lost some genetic diversity during the invasion process but obtained little genetic variation through interbreeding with other local lineages.
This assumption is also supported by the little genetic and epigenetic differentiation between the DELTA and MED groups. Genetic diversity seems to be a key factor explaining its broad ecological adaptation (Kettenring, McCormick, Baron, & Whigham, 2011;Lambertini, Mendelssohn et al., 2012). Compared with LAND group, the DELTA group can adapt to wetter and slightly more saline habitats. Therefore, there is a conflict between lower genetic diversity and the broader ecological amplitude in DELTA group. As the results showed that the epigenetic diversity in the DELTA was as high as in the LAND group, if not higher, the epigenetic diversity partially explained this contradictory. The finding also indicated that epigenetic variation could compensate for decreased genetic variation after an initialized introduction.

| Stable epigenetic differentiation has not occurred
Epigenetic variation in invasive plant populations can contribute to phenotypic variation and plasticity for adaptation in new environments as a fast mechanism during the introduction (Richards et al., 2012;Schulz et al., 2014). In the Great Lakes region, our result predicted a constant threat to the native P. Australis in which the native group exhibited a very low level of epigenetic variation, as well as genetic variation, possibly because of very rare hybridization between the native and introduced groups (Saltonstall, 2011 Schulz et al., 2014), but it remained unexplored how stable and inheritable the B3 variation is in these studies. Common garden filtered the plastic B3 variation in our study, and epigenetic divergence between populations in the introduced region and in its original region was weakened. Moreover, the heterogeneity of microhabitats within each region can increase epigenetic noise masking the direction of epigenetic changes during the invasion. Finally, it is hard to find the few important adaptive loci using a limited number of anonymous loci provided by AFLP and MS-AFLP (Schrey et al., 2013).
For the above reasons, we failed to detect the significant epigenetic divergence independent of genetic variation.
However, the adaptation of introduced groups cannot be explained entirely by adaptive evolution based on genetic or epigenetic variation in this study. Only a few diverged ecophysiological functional traits of the Mediterranean P. australis M1 lineage were detected between the DELTA and MED groups (Guo et al., 2016), and we determined very little differentiation in genetic and epigenetic variation between them.
Given that the climate in the introduced range was more advantageous than that in the original range for the survival of common reed because of abundant precipitation and warm temperature, the introduced environment did not provide enough selection pressures to shape the stable genetic and epigenetic structure of introduced populations.
Therefore, we cannot ignore the potential effect of preadaptation and ecological fitting based on the inherently high phenotypic plasticity during the invasion process of common reed (Guo, Lambertini, Nguyen, Li, & Brix, 2014;Guo et al., 2016), which may be correlated with plastic epigenetic mechanisms (Gao et al., 2010).

| CON CLUS ION
In conclusion, epigenetic variation in introduced groups of P. australis is often correlated with genetic variation, suggesting the closely correlated effect of genetic and epigenetic variation in species with high genetic diversity. When the introduced population cannot acquire adequate genetic diversity through hybridization with other introduced populations, the importance of epigenetic variation may rapidly emerge. Faster stable epigenetic convergence between introduced and native groups was not observed maybe due to the strong genetic effect. Further study in situ and in a common garden for several generations is necessary to associate the inheritable genetic and epigenetic loci with invasive traits, and it is better to use a reduced representation bisulfite sequencing approach based on high-throughput sequencing technology (Robertson & Richards, 2015) for direct comparison with functional genetic context.

ACK N OWLED G M ENTS
We thank Yunfei Cai for helpful guidance on laboratory technology and thank Ting Yu and Song Zhao for looking after these reeds in the common garden. We also want to sincerely thank Hans Brix in Aarhus University of Denmark for his cooperation in the establishment of our Phragmites common garden. This study was funded by National Natural

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

AUTH O R CO NTR I B UTI O N S
LL, WG, and XG conceived the idea and designed the study. WG and XG transplanted reeds from the field and the cooperation garden to our common garden. SL, ND, and CP contributed to garden management and sample collection. LL and CP performed the molecular laboratory work. LL, ND, and XG analyzed and interpreted the data. LL drafted the manuscript, and all authors participated in manuscript modifications and gave final approval for publication.