Genome-wide DNA methylation alterations of Alternanthera philoxeroides in natural and manipulated habitats: implications for epigenetic regulation of rapid responses to environmental fluctuation and phenotypic variation


J. Yang. Fax: 86-21-65643494; e-mail:


Alternanthera philoxeroides (alligator weed) is an invasive weed that can colonize both aquatic and terrestrial habitats. Individuals growing in different habitats exhibit extensive phenotypic variation but little genetic differentiation in its introduced range. The mechanisms underpinning the wide range of phenotypic variation and rapid adaptation to novel and changing environments remain uncharacterized. In this study, we examined the epigenetic variation and its correlation with phenotypic variation in plants exposed to natural and manipulated environmental variability. Genome-wide methylation profiling using methylation-sensitive amplified fragment length polymorphism (MSAP) revealed considerable DNA methylation polymorphisms within and between natural populations. Plants of different source populations not only underwent significant morphological changes in common garden environments, but also underwent a genome-wide epigenetic reprogramming in response to different treatments. Methylation alterations associated with response to different water availability were detected in 78.2% (169/216) of common garden induced polymorphic sites, demonstrating the environmental sensitivity and flexibility of the epigenetic regulatory system. These data provide evidence of the correlation between epigenetic reprogramming and the reversible phenotypic response of alligator weed to particular environmental factors.


Alternanthera philoxeroides (alligator weed) is an invasive weed in China, which is native to South America but has now invaded almost all temperate and tropical areas of the world, including most provinces of South China (Ding et al. 2006; Pan et al. 2007). Alligator weed can exploit diverse habitats in the introduced region. Individuals growing in different habitats exhibit extensive ecophysiological and morphological distinctions, but show very little genetic variation within and among populations (Li & Ye 2006; Geng et al. 2007). It has thus been proposed that phenotypic plasticity, rather than the existence of locally adapted ecotypes, allows the invasive alligator weed to colonize a wide range of habitats with different water availability (Geng et al. 2006, 2007; Li & Ye 2006; Pan et al. 2007).

The mechanisms underlying environment-induced plastic responses and phenotypic variation have received considerable attention in recent years (Schlichting & Smith 2002; Pigliucci 2005; Sultan 2005; Feinberg 2007; Bossdorf, Richards & Pigliucci 2008; Marfil, Camadro & Masuelli 2009; Morange 2009). Much evidence indicates that phenotypic changes in response to environmental heterogeneity are associated with the on-off status or the quantitative expression level of genes, regulated by their epigenetic status (van Kleunen & Fischer 2005; Grant-Downton & Dickinson 2006; López-Maury, Marguerat & Bähler 2008; Marden 2008). Epigenetic mechanisms, for example, the alteration of DNA methylation patterns or a variety of covalent histone modifications, allow an organism to respond to the environment through changes in gene expression. Consequently, phenotypic variation can occur without corresponding changes in the genome (Kalisz & Kramer 2008). Epigenetic information is thus of particular interest to the study of plastic responses to environmental fluctuation and phenotypic variation. However, although there is a growing body of evidence showing that epigenetic modifications play an important role in mediating environmentally induced phenotypic variation, several critical questions remain unanswered. These include how much of the epigenetic component is truly independent of genetic changes and to what extent epigenetic changes contribute to phenotypic variation in natural populations (Bossdorf et al. 2008). Furthermore, despite the latest findings indicating that epigenetic modifications underlie reversible alterations in gene expression, and thereby in phenotype, there is as yet little empirical evidence to demonstrate that epigenetic patterns are so sensitive to environmental stimuli that they can undergo a rapid remodeling and alter a phenotype within the lifespan of a single organism. It is also unclear whether the epigenetic control system is flexible enough to provide an effective short-term strategy for organisms to respond to various environmental fluctuations. The answers to such questions must come from studies designed specifically to rule out the confounding effects of genetic factors, and to include manipulation of ecological factors to detect whether changes in epigenetic state occur as rapidly as the environmental change.

Alligator weed provides a suitable model for the study of epigenetic regulation and phenotypic variation in an ecological context. Throughout its introduced range, alligator weed rarely produces viable seeds but reproduces mainly by vegetative propagation, with the fragments of stolon or storage root breaking off and growing into new plants (Julien & Stanley 1999; Geng et al. 2007; Pan et al. 2007). All new plants produced by this means are thus clones with a genetic make-up identical to that of the parent. These genetically identical but phenotypically variant plants provide an ideal research system to explore the correlations between environmental stimuli, epigenetic modification and phenotypic variation.

DNA methylation is now recognized as a primary and perhaps the most extensively characterized epigenetic mark. Genome-wide profiling of DNA methylation using the methylation-sensitive amplified fragment length polymorphism (MSAP) technique has already provided provocative insights into the genomic response to environmental perturbation (Aina et al. 2004; Salmon, Ainouche & Wendel 2005; Peredo et al. 2008; Marfil et al. 2009). In this study, we examined the epigenetic variation of alligator weed in natural populations and under manipulated growing conditions in common gardens, using MSAP. Five morphological traits, that is, leaf length, leaf width, stem diameter, stem pith cavity diameter and internode length, were used as markers of phenotypic variation under different conditions. The purposes of this study were: (1) to investigate whether DNA methylation status varied within and among natural populations of alligator weed; (2) to evaluate the sensitivity and dynamic nature of epigenetic modifications in response to environmental cues by common garden experiments; (3) to explore the potential correlation between epigenetic reprogramming and the reversible phenotypic response of alligator weed to particular environmental factors. The results show that there were considerable methylation polymorphisms within and between natural populations, and that extensive epigenetic reprogramming occurred in the common garden experiment.


Field survey

To gain a broad picture of the variation pattern of alligator weed across southern China, three sites with different climatic and geographical conditions were selected: Kunming of Yunnan Province (E102°43′ N25°02′), Nanning of Guangxi Province (E108°22′ N22°48′), and Zhuji of Zhejiang Province (E120°20′ N29°40′). We sampled alligator weed from different habitat types, that is, aquatic and terrestrial habitats, at each site. Here we refer to ‘aquatic habitats’ as rivers, ponds or reservoirs where plants root in soil at the edge and extend long stems over the water surface or form free-floating mats, whereas ‘terrestrial habitats’ refers to areas that are not flooded even after rain, like roadsides. Three pairs (aquatic versus terrestrial) of natural populations were thus included in this study. Forty individuals were sampled from each population. The distance between individuals was always more than 50 m, to ensure different clones.

Five morphological traits, that is, leaf length, leaf width, stem diameter, stem pith cavity diameter, and internode length, were chosen as markers of phenotypic variation under different conditions. In situ measurements were conducted on individuals collected from different populations using the methods described in Geng et al. (2007). After field survey, the stem cuttings of all 240 sampled individuals were collected and brought back for common garden experiments. The youngest fully expanded leaves were selected and dried with silica gel for genetic and epigenetic profiling analyses in the laboratory.

Common garden experiment

To minimize the confounding effects of different ecological factors and to investigate the specific variation in the feature of epigenetic modification in response to heterogeneous water conditions, two water-manipulated common gardens were set up in the experimental field of Fudan University, Shanghai (E121°29′ N31°14′). One common garden was to simulate the typical aquatic habitat colonized by alligator weed in natural environments, consisting of a pond (7 × 7 × 0.5 m), and the other simulated the typical terrestrial habitat, consisting of an upland (10 × 5 m). The two common gardens were adjacent and experience the same climate conditions.

All 240 stem cuttings brought back from natural populations were grown in the greenhouse for 1 week, allowing the plants to recover from transplant shock, and then cut into segments 5 cm long bearing a vegetative bud each. The fragments were allowed to grow until two new leaves had developed. Two ramets derived from the same stem cutting, of similar size, were then transplanted into pots containing a 1:1:1 mix of peat, vermiculite and sand, one in each pot. The pots were fertilized with 6 g commercial compound fertilizer containing 15:11:13 of N/P/K (Osmocote, the Scotts Company, Marysville, OH). About 5 d later, after all of the 480 ramets were established, they were allocated to the two common gardens. The ramets from different original plants collected from natural aquatic and terrestrial habitats, respectively, were reciprocally transplanted into the two common gardens. The common gardens were monitored every day to ensure that the water surface was kept 10 cm above the pots in the pond, whereas 1 L water was supplied to the plants in the upland garden every day. Plants were harvested and used for morphological and DNA profiling analyses after 4 months of growth under these manipulated conditions.

Amplified fragment length polymorphism (AFLP) and MSAP analyses

The AFLP technique was used to investigate whether there exists genetic variation within and among different natural populations of alligator weed. DNA from 40 individuals was extracted individually for each population, using the TIANGAN (Beijing, China) Plant Genomic DNA Kit. AFLP reactions were performed according to the method of Vos et al. (1995). Selective amplification was carried out using nine primer combinations: Eco-AGG/AGC/AAC/ACA + Mse-CAA/CTT/CTA/CAT. PCR products were resolved on 6% sequencing gels and silver stained according to the protocol described by Bassam, Caetanoanolles & Gresshoff (1991).

The MSAP technique was used to conduct genome-wide screens for epigenetic variation of alligator weed in response to different field conditions. To gain a comprehensive picture of inter-individual and inter-population differences in methylation status, we performed MSAP analyses on 15 randomly selected individuals from each population, and on the bulked DNA of 40 individuals for each population, respectively. Nine selective primer combinations (Eco-AAC/ACG/AGC/AGG + Msp/Hpa-TCAA/TCC/TCG/TCT/TGA/TTG) were used for the MSAP analysis performed on randomly selected individuals, and 50 selective amplification primer pairs were chosen (Eco-AAC/AAG/ACA/ACT/ACC/ACG/AGC/AGG + Msp/Hpa-TCAA/TCC/TCG/TCT/TGA/TGC/TTC/TTG) for MSAP analysis using the bulked DNA. MSAP was carried out according to the protocol described by Portis et al. (2004), which is modified from AFLP with the ‘frequent cutter’MseI substituted by HpaII and MspI. HpaII and MspI are a pair of isoschizomers, which have the same recognition sequence 5′-CCGG but different sensitivities to the methylation state of cytosine. HpaII will not cut if either of the cytosines is fully (both strands) methylated, but will cut if the external cytosine is hemi-methylated (single strand); whereas MspI will not cut if the external cytosine is fully or hemi-methylated, but will cut if the internal cytosine is fully or hemi-methylated. Full-methylation of the external cytosines (on both strands) or full-methylation on both cytosines prevents cutting (both HpaII and MspI), which means these two methylation states cannot be distinguished by the MSAP technique. However, comparison of amplification products from EcoRI + HpaII and EcoRI + MspI, allows the identification of some types of mentylation states. Fragments present in both EcoRI + HpaII and EcoRI + MspI digestions are attributed to the corresponding CCGG sites being non-methylated. Fragments present in EcoRI + HpaII, but absent from the corresponding EcoRI + MspI digestion indicate that the external cytosine is hemi-methylated. Fragments absent from EcoRI + HpaII, but present in the corresponding EcoRI + MspI digestion indicate that the internal cytosine is fully methylated. PCR fragments were separated on 6% sequencing gels and visualized by silver staining according to Bassam et al. (1991). To assess reproducibility, two separate MSAP reactions were performed for each sample using independent DNA isolations. Only intense and reproducible bands were scored for analysis. The epigenetic modification patterns of plants in the common garden environments were also estimated by MSAP. DNA bulks were prepared for each population under different treatments by pooling equal amounts of DNA resulting from individual DNA extraction of 40 individuals.

Statistical analysis

Morphological measurement data were log transformed to meet the assumption of homoscedasticity and analyzed by analysis of variance (anova). The AFLP and MSAP bands were scored as present (1) and absent (0) in binary matrices. GENALEX6.1 (Peakall & Smouse 2006) was used to calculate the proportion of polymorphic alleles, Dice genetic distance (equivalent to Nei & Li 1979), and Shannon diversity index (Lewontin 1972). Nei's unbiased gene diversity (Nei 1987) was calculated in Arlequin 3.1 (Excoffier, Laval & Schneider 2006). The epigenetic relationships among individuals were represented graphically by principal coordinate analyses (PCO) based upon the matrices of pairwise Dice genetic distances. Epigenetic differentiation at different hierarchical levels, that is, among sites, between habitats within sites, and within habitats, was calculated using analysis of molecular variance (amova) in Arlequin 3.1. Significance of variance components was obtained by nonparametric procedures using 1000 random permutations.


DNA methylation alterations of alligator weed in natural environments

Morphological examination revealed significant phenotypic differences among plants growing in different regions and habitats. In particular, plants in aquatic habitats exhibited significantly larger leaves, longer internodes, and larger stem pith cavity diameter than those in terrestrial habitats (Fig. 1). The anova results indicated that both regional and local habitat factors had significant effects on morphological traits (P < 0.001) (see Table S1a in Supporting Information). In contrast to significant variation in morphological characters, few genetic differences were detected among plants from different natural populations. A total of 1089 AFLP bands were scored from the nine primer combinations. Of the 240 individuals included in this study, 225 showed an identical AFLP profile. Nineteen polymorphic bands were observed in 15 individuals scattered in five natural populations. The percentage of polymorphic bands ranged from 0 to 0.73% for different populations, with a mean of 0.34% at the species level. The Nei's index of genetic diversity (He) and the Shannon diversity index (I) were close to zero at the species level.

Figure 1.

Variation of morphological traits in field and common garden plants of A. philoxeroides, with standard errors marked on the bars.

Global DNA methylation patterns of 90 individuals from six field populations were revealed by MSAP. A total of 1000 markers were generated with nine selective primer combinations and 274 bands were polymorphic (27.4%). Different methylation states were observed within and among populations. The relations among individuals were shown by a PCO plot based on pairwise Dice distances (Fig. 2), where the first and second axes explained 33.98 and 20.84% of the variation, respectively. The individuals examined were grouped into three major clusters corresponding to three source field sites: Kunming, Nanning, and Zhuji. Furthermore, individuals from aquatic and terrestrial habitats tended to form distinct subgroups within each local group (Fig. 2). amova showed that a large proportion of epigenetic variation (78.7%, P < 0.001) resided within populations, whereas 13.4% (P = 0.052) and 7.9% (P < 0.001) of the epigenetic variation resided among geographic sites and between habitats within sites, respectively. The results of MSAP analysis based on pooled DNAs from 40 plants of each population, using the same primer combinations, were consistent with those based on individual DNAs, with the positions of each pooled sample being close to the relevant individuals in the PCO plot (Fig. 2). Using 50 selective primer combinations, we generated a total of 5183 clear and reproducible MSAP bands, of which 397 (7.64%) were found to be polymorphic. The numbers of various fragments attributed to non-methylation, hemi-methylation of external cytosine, full-methylation of internal cytosion and full-methylation of external cytosine or both cytosions, respectively, were calculated for each population based on MSAP profiles (see Supporting Information Table S2). We detected 57, 142 and 154 epigenetic sites in Kunming, Nanning, and Zhuji populations, respectively, which show distinct patterns of DNA methylation between aquatic and terrestrial plants. But most DNA methylation polymorphisms scored in wild populations are field site-specific.

Figure 2.

Plot of principal coordinate analysis based on MSAP banding patterns, showing the relationships among individuals from different natural populations. KW, plants from Kunming aquatic population; NW, plants from Nanning aquatic population; ZW, plants from Zhuji aquatic population; KD, plants from Kunming terrestrial population; ND, plants from Nanning terrestrial population; ZD, plants from Zhuji terrestrial population.

Phenotypic alteration and epigenetic reprogramming under common garden conditions

By transplanting plants from different source populations into each of the two experimental gardens and imposing different water treatments, we found that both the morphological traits and the genomic methylation patterns of alligator weed were sensitive to water changes, and were highly flexible. The results showed that plants growing in the pond common garden, no matter where they were from, exhibited significantly thicker stems, longer internodes, and larger stem pith cavity diameter than those growing in the upland common garden, showing a similar variation trend to plants growing in natural habitats (Fig. 1). Genome-wide DNA methylation profiling using MSAP revealed that the genomes of alligator weed were subject to extensive epigenetic reprogramming under different water treatments. A total of 278 polymorphic fragments were found using 50 selective primer combinations. By comparing MSAP profiles of common garden-reared plants and the related natural-reared plants, it was found that, out of the 278 polymorphic sites identified, 62 were conserved and 216 exhibited distinct patterns between common garden- and natural-reared plants. Most intriguingly, of the 216 common garden induced polymorphic sites, 169 showed high regularity of variation corresponding to different water treatments, and the alterations in DNA methylation pattern were shared by all plants, no matter which field site and habitat the plant came from (Fig. 3, Supporting Information Table S3). With respect to the global methylation level of genomes, different frequencies in hemi- and full methylation were found between plants from the pond and upland common garden, although the differences did not reach statistical significance (see Supporting Information Table S2).

Figure 3.

Variation in cytosine methylation patterns revealed by MSAP profiles in field and common garden plants of A. philoxeroides. DNA samples from 40 individuals of each population were pooled and used in profile analysis. (a) An example of the MSAP profile. Arrow-heads indicate DNA methylation variation among field plants. Arrows indicate locus-specific DNA methylation alterations in response to pond/upland common garden treatments. Primer combinations are EcoRI + AGG/HpaII (MspI) + TCC. (b) Polymorphic methylation loci summarized from MSAP profiles of 50 selective primer combinations. The colour code indicates the methylation state of each site in different plants. wKW, Kunming, aquatic, pond common garden; wNW, Nanning, aquatic, pond common garden; wZW, Zhuji, aquatic, pond common garden; wKD, Kunming, terrestrial, pond common garden; wND, Nanning, terrestrial, pond common garden; wZD, Zhuji, terrestrial, pond common garden; dKW, Kunming, aquatic, upland common garden; dNW, Nanning, aquatic, upland common garden; dZW, Zhuji, aquatic, upland common garden; dKD, Kunming, terrestrial, upland common garden; dND, Nanning, terrestrial, upland common garden; dZD, Zhuji, terrestrial, upland common garden; MW, PUC19 DNA Marker. Other abbreviations are the same as Fig. 2.


Every species may experience spatial and/or temporal variation in the environment. Living organisms have developed various strategies to adapt to environmental changes. In contrast to the long-term strategy of generating new traits by selection, phenotypically plastic variation provides an efficient short-term strategy for organisms living in changing environments to sense their environment and to respond to it rapidly and flexibly (Boyko & Kovalchuk 2008; Chinnusamy & Zhu 2009). Given the sensitivity to the environment and potential reversibility of epigenetic modifications, the epigenetic machinery offers an important window to understanding the molecular mechanisms underlying plastic alterations in morphological and physiological traits in response to environmental stimuli.

In this study, we showed that plants of different source populations not only underwent significant morphological changes in common garden environments, but also underwent a genome-wide epigenetic reprogramming in response to different treatments. The epigenetic system shows three important properties. Firstly, the epigenetic pattern can be modified by environmental factors directionally. Current research not only shows the environmental susceptibility of epigenetics, but also shows how environmental factors can influence epigenetic processes, thereby affecting epigenetic marks. Previous studies also demonstrated the directional induction of DNA methylation patterns by experimental treatments (Wheeler et al. 1992; Wade & Archer 2006; Weinhold 2006). Secondly, epigenetic alteration can be induced over shorter timescales than a lifetime. The dynamic nature of rapid reprogramming of the epigenome was highlighted by active de novo methylation and demethylation events induced in common garden experiments. It would be meaningful to test whether the induced alterations in DNA methylation and variations in ecophysiological traits occur independently or are interrelated to elucidate the basis of rapid responses of individuals to environmental fluctuation. Finally, the epigenetic variation is likely to be reversible. By transplanting plants from different source populations reciprocally into each of the two experimental gardens, we showed that the distinct epigenotypes of wild plants are remodeled massively according to the common environments. Plants raised in the same common garden, no matter whether they came from aquatic or upland habitats, display a high level of DNA methylation similarity, this can be taken as indirect evidence for the reversibility of epigenetic variation in response to dynamic environmental conditions (Weinhold 2006).

Genome-wide methylation profiling using MSAP revealed inter-individual and interpopulation differences in global DNA methylation in field plants of alligator weed. Some polymorphic loci exhibited field site-specific variation, including loci that show distinct DNA methylation patterns between aquatic and terrestrial plants. These results suggest the confounding effects of variable environmental factors to epigenetic variation of alligator weed. The environmental factors can be partitioned in a hierarchical fashion based on the spatial scale at which these factors likely operate: sample sites within habitat, habitats within the same geographic region, and different geographic regions. The environmental factors covarying across geographical locations seem to exert a relatively greater impact on DNA methylation patterns in field plants, leading to the formation of three major groups in the PCO plot. When the confounding effects of variable environmental factors were removed in the common garden environment, plants from different geographic regions and habitats exhibited a similar pattern of global DNA methylation in response to different water treatments.

Although fingerprints provided from the MSAP technique are specific to DNA methylation status, and therefore just reflect epigenetic variation partially, they showed a close correlation between the dynamics of epigenotype and the phenotypic variation of alligator weed relevant to specific environmental challenges. Based on complementary field and common-garden studies of alligator weed, it is reasonable to conclude that the wide adaptability of alligator weed to diverse environments is principally determined by an underlying flexible epigenetic regulatory system. The results of this study not only extend our understanding of the relationship between the molecular process of epigenetics and the ecological process of phenotypic plasticity, illustrating how phenotypic expression varies in alternative conditions, but also have even broader implications for the maintenance and benefit of plastic traits under changing environments. It is critical to note that, although the epigenetic changes that can persist over several generations in the population and their potential adaptive significances have received special attention in recent years, relatively little attention has been paid to swift and reversible epigenetic alternations that mediate rapid plastic responses of the organism to environmental perturbation. Such flexible alterations can not only buffer environmental fluctuations by adjusting physiological capacities, but also enable epigenetically induced gene expression to generate specialized morphological adaptations, permitting persistence in varying environments and increasing the potential for evolution. This mechanism is particularly important to allow organisms with no/low genetic diversity to adapt to different environments, and is likely to be a favorable evolutionary response when organisms are exposed to stress periods that last shorter than a single life span (Reiber & Roberts 2005).

Being a malignant invasive species, alligator weed has become the object of many invasion biology studies. However, the molecular mechanisms underpinning the wide range of phenotypic variation and rapid adaptation to novel and changing environments remain uncharacterized (Prentis et al. 2008). It is impossible to make a causal link between the observed methylation changes and particular phenotypic traits based on MSAP fingerprint analysis. We have done de novo transcriptome assembly from paired-end RNA-Seq data for alligator weed. We assembled approximately 55 million reads into 24 241 contigs 500 bp or longer, and identified 2333 differentially expressed consensus-genes in response to different water treatments (unpublished data). We anticipate that the ongoing studies of the functions and regulatory mechanisms of differentially expressed genes will extend our understanding of the role of epigenetics in mediating environmentally induced phenotypic variation in alligator weed, and eventually help to explain why this species is unusually flexibile in its ability to acclimate, and thus highly invasive.


We are grateful to Mr Ming Nie and Ms. Liyan Zeng for their experimental assistance, and to Mr Kun Wang for assistance in the field survey. We thank Dr Beverley Glover (University of Cambridge) for proofreading. This work was supported by the National Key Basic Research Program (973) (2009CB119201), Shanghai Leading Academic Discipline Project (B111), and the Innovative Foundation of graduate students of Fudan University (EYH1322097).