Epigenetics and the success of invasive plants

1. Introduction

Phenotypic variation fuels the successful response of organisms to environmental challenges, and a lack thereof would thus seem to be a universal obstacle for introductions of species to new habitats [1,2]. The dilemma results from the fact that species are generally thought to be introduced to new ranges by only limited numbers of individuals and, therefore, to undergo genetic bottlenecks characterized by reduced genetic variation (and presumably phenotypic variation) compared to their native range. This type of bottleneck is expected to result in inbreeding depression and decreased evolutionary potential, presenting a ‘genetic paradox’ for understanding the successful plant invaders and their adaptation to new habitats [2,3].

Our understanding of the importance of genetic bottlenecks is rooted in the modern synthesis of evolutionary theory, which asserts that genes (defined by DNA sequence) are the sole source of heritable phenotypic variation, and that inheritance of environmentally induced non-genetic variation is impossible [4–7]. The recovery of genetic variation has, therefore, long been considered an important driver of successful invasions [8]. However, applications of genomics approaches have revealed the genetic make-up of invasive populations and helped to refine our views of genetic diversity and evolutionary processes during invasions. First, comparative studies in the native and introduced species ranges suggest that the genetic paradox may not be as severe as initially thought because many invasive populations exhibit only modest reductions in genetic variation due to multiple introductions, hybridization or de novo mutations [3]. Moreover, the loss of diversity measured by molecular markers does not necessarily reflect the loss of quantitative trait variation or may reflect successful response to selection to the novel habitat from an initially higher diversity of founding genotypes or new recombinants (e.g. [9]). Genetic bottlenecks can also contribute to performance by purging deleterious alleles, revealing beneficial cryptic variation or creating new beneficial interactions among genomic elements [3,10–14].

In addition to genetic marker studies, accumulating whole-genome and transcriptome sequence data from various species has allowed us to develop a more nuanced understanding of genome dynamics. One important insight is the prominence of genomic redundancy, largely resulting from multiple episodes of whole-genome duplication (polyploidy) followed by genome fractionation and diploidization processes [15–18], which play a major role in plant diversification and adaptation [19]. Expression of duplicated genes (homeologs) can diverge from that of the homologous genes in the parental species, creating functional plasticity even in the absence of genetic variation among individuals [17,20]. Another area of progress is on the effects of the ‘genomic shock’ resulting from interspecific hybridization and polyploidy, which can re-programme gene expression and create phenotypic novelty [18,21–23]. The merging of two different genomes (and corresponding divergent regulatory networks) can lead to an expression repatterning that is non-additive with respect to parental expression. This non-additivity can manifest as parental dominance (reflecting similarities to one of the parental expression patterns) or transgressive (e.g. up- or down- regulation compared to parents) expression [17] at individual loci. These expression changes may differ among tissues, organs or environmental conditions [15,17]. In this context, we describe ‘phenotypic novelty’ according to Pigliucci [24] as new traits or novel combinations of traits (or values of traits) that contribute to ecological functions that are novel for a lineage. However, despite a tremendous increase in information about genomic and transcriptomic processes for many species, biologists generally have only a limited understanding of the actual molecular underpinnings (e.g. the ‘black box’) of organismal responses to complex biotic and abiotic factors [4,25].

Ultimately evolutionary response to challenging environmental conditions relies on heritable phenotypic responses, regardless of the underlying mechanisms of heritability [6,26,27]. We now have evidence that the structural and functional dynamics of genomes along with a variety of epigenetic and other non-genetic effects can affect heritable variation and thus contribute to adaptation [6,28–33]. Recent efforts have specifically addressed the ‘black box’ of the translation from genotype to phenotype by exploring intermediate molecular-level phenomena. For example, genome-level processes are often mediated by epigenetic changes (chromatin modifications, DNA methylation, small RNAs), which can vary among individuals within populations and be selected for, much like genetic variation, and serve as an additional source of heritable variation [31,32,34]. Changes in DNA methylation are known to be associated with allopolyploidization (e.g. [35–38]), exposure to environmental stress [39–41], different habitat types or shifts in species ranges [42–45] and with variation in plant phenotypes [30,46–48]. Epigenetic variation in turn can regulate differential gene expression and transposable elements (TE) activation or repression [49–51]. The importance of TE activity in the context of biological invasion is still largely unexplored (reviewed in [12]; but see [52]), but several studies have explored how differential gene expression underlies variation in ecologically important traits (reviewed in [53]; e.g. [54–56]). Specific to the context of biological invasions, changes in gene expression can translate into variation in inducible defences and secondary metabolism, which combines into complex phenotypes that can be finely tuned and provide a diverse array of appropriate responses to environmental challenges in plants [57–59] and animals [60].

These possible contributions of genome dynamics to phenotypic variation and adaptation are intriguing and relevant for invasion biology because some extremely invasive clonal plant species have almost no detected genetic diversity even after they become well established, e.g. knotweeds in the USA [61] and Europe [62–64], alligator weed (Alternanthera philoxeroides) in China [65–67], Spartina anglica in Europe [68,69] and water hyacinth (Eichhornia crassipes) outside of South America [70]. Researchers attempting to explain the huge success of these species often argue for the importance of phenotypic plasticity or ‘general-purpose genotype’ [8,71–73], which could be mediated by epigenetic effects that translate into gene expression and changes in plant chemistry and other phenotypes [31,58,74–76]. Many of the hypotheses that attempt to explain variation in invasion success involve genetic or plastic response to novel biotic interactions, often associated with changes in secondary metabolites [59,77–81].

In the present paper, we take advantage of work done in plant invaders to outline how epigenetics may impact phenotypes and adaptation of species in their native versus introduced range. We specifically discuss how epigenetic variation can translate into transcriptomic and metabolomic variation, with a special focus on the biotic interactions of invasive plants and how epigenetic mechanisms can facilitate plant invasion by providing sources of variation to clonal plants, or novelty through hybridization and polyploidization.

2. Rapid adaptation during plant invasion

The primary way that epigenetics could play a role in plant invasions is by contributing to phenotypic variation and adaptation of invasive plant populations, which requires linking epigenetic mechanisms to the traits of invaders, their abiotic and biotic interactions and how these are linked to the fitness advantages of invaders, and thus adaptation. As much as epigenetic mechanisms can contribute to variation in gene expression, and act as stress-responsive and developmental regulators [50,74,82,83], they will be important for the plastic response of invasive plants [76,84,85], and some of these responses could be inherited through epigenetic or other forms of non-genetic inheritance (reviewed in [6,32,84,86]; see also figure 1). In addition, invasive plants are generally expected to evolve quickly because of the novel selection pressures they experience, and it seems plausible that epigenetic processes such as DNA methylation—which have a higher mutation rate and respond quickly to environmental changes—may contribute to this. There are many well-documented cases of rapid phenotypic evolution in invasive plant populations [10,13]. However, the possibility of epigenetic contributions has so far hardly been tested [87]. Below we briefly summarize some of the key ecological-evolutionary hypotheses for plant invasion success, and suggest links to underlying genomic (including epigenomic) processes.

• Download figure
• Open in new tab
• Download PowerPoint

Several hypotheses in the invasive species literature have examined how biotic interactions, and their evolutionary consequences, contribute to increased fitness of invasive species in their introduced ranges. The enemy release hypothesis (ERH; [88]) suggests that escape from native herbivores and other specialist enemies explains the success of many exotic species, and the evolution of increased competitive ability hypothesis (EICA; [89]) proposes that after such enemy release, some invasive plants reallocate resources and rapidly evolve towards less defended but more vigorous ecotypes, as reported, for example, in A. philoxeroides [90]. An advancement of EICA is the shifting defence hypothesis (SDH; [7,77]), which postulates that invaders may not reduce all herbivore defences in invasive populations, but there will be a shift from defences against specialist towards those against generalist herbivores (driven by trade-offs between quantitative and qualitative defences). In fact, metabolomic analyses have found some invasive species to harbour more total secondary metabolites related to generalist herbivores, such as flavonoids or glucosinolates, in their introduced range than in their native range (e.g. [59,81,91–94]). To add even more complexity, plant defences can involve both resistance and tolerance mechanisms, where resistance generally reduces damage to plants [95], but tolerance mitigates the fitness impact of the damage, for example, through rapid re-growth [96]. Resistance mechanisms can also be both constitutive, i.e. constantly present, or only induced by herbivore attack [97]. Recent studies of the evolution of invasive plant defences have begun to incorporate these additional levels of complexity (e.g. [98]). For instance, a recent SDH study on invasive Spartina alterniflora in China found increased resistance to specialist herbivores but lower tolerance to generalist herbivores in invasive populations [99].

Ultimately, the evolutionary responses to novel biotic interactions described above must be associated with corresponding changes at the genomic, transcriptomic and metabolomic levels. However, evolutionary studies of invasive plants that integrate molecular data with phenotypes, plant fitness and biotic interactions are still rare ([13]; but see [100,101]). In particular, the potential contribution of epigenetic processes in rapid adaptation of invasive plants has hardly been addressed.

Although so far no study exists that fully encompasses genomic (including epigenomic) data to transcriptome, metabolome, biotic interactions and plant fitness, we already have evidence for several important pieces of this puzzle. For instance, numerous transcriptomic, epigenetic and metabolomic profiling studies have supported the correlations of these mechanisms with herbivory. In Arabidopsis thaliana, herbivory-induced changes in gene expression involved up- and downregulated genes in plant secondary metabolism networks, hormone signalling pathways or plant defence-related genes [102,103]. Changes in expression patterns of genes involved in plant defence against herbivory were observed in Populus [104] and Solanum [105,106] and in the invasive species Ambrosia artemisiifolia [54] and Solidago canadensis [56]. Likewise, some studies in non-invasive species have linked epigenetic variation to herbivory. Herrera & Bazaga [107] found methylation polymorphisms to be correlated with herbivory damage in Viola cazorlensis, and Kellenberger et al. [108] reported herbivory-associated changes in DNA methylation in Brassica rapa. In herbivore-infested plants and when herbivory was simulated by methyl jasmonate exposure, DNA demethylation was correlated with changes in floral signalling to pollinators. Similarly, in A. thaliana response to plant defence hormones like jasmonic acid and salicylic acid was shown to be correlated with differences in DNA methylation among epigenetic recombinant inbred lines (epiRILs; [109]). Other studies in A. thaliana have shown that histone modifications, DNA methylation, or RdDM (RNA-directed methylation by small RNAs) are correlated with the expression of chemical defences such as glucosinolates and flavonoids [110–113], supporting the idea of a role of these epigenetic mechanisms in shifting defences. In summary, we have evidence for links between herbivory and activity of defence genes, between herbivory and epigenetic variation, and between epigenetic variation and plant defences. Together this makes a role of epigenetic mechanisms also in the evolution of defences during plant invasions very plausible.

An important additional tool in this context is metabolomic approaches that quantify plant secondary chemistry and provide a better understanding of the translation of genomic variation into plant defences and success in novel environments [59]. With appropriate experimental designs, metabolomic analyses can reveal how plant secondary chemistry is induced by abiotic and biotic stresses and thus provide a functional readout of genome-level processes complementary to transcriptomic studies. Metabolomics approaches have already offered mechanistic support for the SDH hypothesis through surveys that showed herbivory impacts on the production of defence metabolites [81,114,115]. Still, how genomic (including epigenomic) variation translates into variation in metabolites and, as a consequence, interactions with herbivores and invasion success, has been largely unexplored in invasive plants. There is, again, some evidence from model species and other non-invasive study systems.

In Arabidopsis, studies of epiRILs have identified epigenetic quantitative trait loci underlying response to herbivore attack, providing evidence that methylation is involved in the control of secondary metabolism [58]. Differences in glucosinolates, flavonoids and additional metabolites among epiRILs support the importance of variation in DNA methylation in plant responses to herbivory. In Brassica hybrids, the integration of transcriptome and metabolome data allowed for the investigation of enhanced flavonoid and glucosinolates production, but no details about epigenetic variation were surveyed [116]. Such analyses are much more challenging in non-model species without extended genomics resources, but more restricted approaches such as methylation-sensitive amplified fragment length polymorphism (AFLP) or epigenotyping-by-sequencing (epiGBS) coupled with untargeted metabolomics can nevertheless provide important insights [32,117]. Combining (epi)genomic and metabolomic analyses in an integrative approach will reveal new understanding of the molecular mechanisms underlying exotic species' responses to biotic interactions.

3. Clonal plant invasions

In addition to a general contribution to plant adaptation, epigenetic changes may be especially important in clonal plants [31,118,119], and clonal growth is common among invasive plant species. If epigenetic variation can provide a source of phenotypic variation within clonal individuals, which translates into fitness differences, it could contribute to invasion success [86] (figure 1). Some of the world's most successful invasive plants are thought to be entirely genetically uniform in their introduced ranges. Among the 468 invasive plants from the IUCN Global Invasive Species Database (http://www.issg.org/database), 70% reproduce clonally, which is similar to the estimate for flowering plants overall [120]. But out of the plants included in the list of 100 worst invasive species [121], a greater majority (30 of 37, or 81%) reproduce clonally, and for nine, it is the main mode of reproduction in their introduced range. In a survey of invasive plants in China, Liu et al. [122] found that almost half (44%) of the 126 invasive plants studied were clonal, and among the 32 most invasive ones, the fraction of clonal plants was even 66%. The high frequency of clonality among invasive plants suggests that this mode of reproduction is beneficial for plant invasion.

One of the best-known cases of an invasive clonal plant is Japanese knotweed (Reynoutria japonica aka Fallopia japonica), where a single octoploid clone has spread aggressively through a broad range of habitats in temperate Europe and North America [61,123–127]. In the USA, Richards et al. [61] found only individuals of the same single AFLP haplotype of R. japonica that is the same one present throughout Europe, while 12 other populations were made up of only a few haplotypes of the hybrid hexaploid species R. × bohemica. They reported that some individuals of each haplotype were found in beaches and marshes— novel habitats from the perspective of Japanese knotweed— with epigenetic variation correlated to the different habitats [61,128]. Another study across central Europe confirmed that all individuals of R. japonica belonged to one haplotype, but different populations harboured significant epigenetic and phenotypic variation which was associated with climate of origin and thus possibly related to adaptation [63].

Another group of clonal species that provides intriguing cases of successful invasion is represented by several members of the polyploid genus Spartina. The molecular evolutionary history of the Spartina genus has been investigated well by Ainouche and colleagues [129–132]. Cases of intercontinental invasions are illustrated in S. alterniflora, S. anglica and S. densiflora [133–135]. Unlike in the case of Japanese knotweed, the hexaploid S. alterniflora has also been studied extensively in its native range (North American Atlantic and Gulf Coasts), and recent studies have compared native and invasive populations, providing some insight into the potential mechanisms underlying its invasive abilities [136–139]. In its native range, S. alterniflora is a critical foundation species that supports fisheries, provides storm protection and has a broad environmental tolerance [137,140]. The species is adapted to anoxic and salt conditions, and is highly tolerant to reduced, sulfidic sediments, and oil/polycyclic aromatic hydrocarbons pollution [137,141–146]. Like other marsh species, S. alterniflora exhibits extensive variation in traits across habitats, and a range of ecological studies have shown that traits of native S. alterniflora are correlated to the environmental heterogeneity of salt marshes [137,140,147]. More recently, several studies have also found correlations between DNA methylation and habitat or oil pollution [43,143,145]. Comparisons of native North American and invasive populations in China support the idea that S. alterniflora benefitted from multiple introductions that created a hybrid swarm which contributed to the rapid evolution of increased fecundity [138,139]. Chinese plants grew taller, denser and set more seeds than those from the native range [139].

From the Atlantic North American coast, S. alterniflora colonized the South American Atlantic coast (Argentina), where it is now well established [148]. The species was also introduced to California [134], Western Europe [133], South Africa [149] and China, where its spread has been spectacular [150]. Clonal propagation seems to have contributed to the spread of this otherwise wind-pollinated perennial species in many cases ([151,152]; but see [138,153]). Although Chinese populations appear to be genetically diverse [138,139,153], the introduction from China to Japan so far has resulted in a lack of diversity in the Japanese populations, suggesting both founder effect and predominantly clonal propagation [152]. Other invasive Spartina species spread extensively by clonal means, such as S. patens (syn. S. versicolor; [154]) introduced from North America to Europe [155] and S. anglica, which has expanded in Europe and is now introduced on several continents [68,133]. Studies in this allododecaploid species suggest that high intra-genomic diversity can contribute to the evolution of gene expression repatterning even within a single clone [156–158].

Other prominent examples of nearly genetically uniform clonal plant invaders are the South American alligator weed (A. philoxeroides; [66,159]), ‘Bermuda buttercup’ (Oxalis pes-caprae; [160]), hawkweed (Hieracium aurantiacum; [72]), crimson fountain grass (Pennisetum setaceum; [161]), common reed (Phragmites australis; [162,163]) and water hyacinth (E. crassipes; [70]). In addition to the epigenetics studies in knotweeds [61,63], DNA methylation marker studies in invasive A. philoxeroides populations have also found patterns of greater epigenetic than genetic variation, and that epigenetic differences were partly explained by habitat [67]. Even after 10 asexual generations, some of the epigenetic differences presumed to be induced by environmental differences were maintained when the plants were grown in a common garden [159]. Similarly, in P. australis, the introduced subspecies harboured significantly more epigenetic variation compared to the native from the same watersheds in Maine, USA, and epigenetic variation was partly explained by site of origin [163]. Another study found that in the Gulf region of the USA, epigenetic diversity in invasive populations was comparable to that of native populations while genetic diversity was much lower in invasive populations [164]. There are many more successful clonal plant invaders that have not yet been studied at the molecular level, even though clonality has been found to be overrepresented among invasive plants (e.g. [122]).

Several authors have argued that epigenetic mechanisms could be particularly important for invasive species that are clonal or have low genetic diversity since they could provide a non-genetic source of heritable variation [31,40,41,118,119,165,166]. Although asexual reproduction is predicted to result in slower rates of evolution, and accumulation of deleterious mutations, clonality is thought to be adaptive in challenging environments including geographically marginal populations where reliable pollination vectors may be absent [167–171]. In addition, Schoen & Schultz [170] argued recently that the modular nature of plants in general allows for detrimental mutations to be terminated in specific cell lines while beneficial mutations can be maintained (see also [172]). This should be even more true in clonal plants, which would also allow for beneficial combinations of mutations to become nearly immortal [172,173]. Phenotypic plasticity has long been implicated as an important component of adaptive response in clonal plants [174,175], which has recently also been associated with epigenetic mechanisms that underly plastic responses [48,118,119]. In addition to the importance of epigenetic mechanisms that underlie plasticity in clonal plants, epigenetic mechanisms could be particularly important in clonal plants because asexual reproduction ‘bypasses' the resetting of epigenetic effects that is thought to occur through meiosis (even though meiotic resetting is not universal; see [176–178]). For example, the recent domestication of a wild plant by clonal propagation identified novel patterns of epigenetic diversity that did not conform to patterns of genetic diversity in domesticated compared to wild populations of the same species [179].

Although there are compelling arguments for the importance of epigenetic effects in clonal plants [31,118,119,173], so far all studies of clonal plant invasions have been limited by a lack of genomic information, and we have limited understanding of how important somatic mutation might be in this context. This is important since several studies have reported that high rates of somatic mutation may allow asexual species to maintain abundant genetic variation and adapt to changing environmental conditions [170,180,181]. We also know from whole-genome studies in the model plant Arabidopsis thaliana and in human cancers that novel epigenetic variation can be dramatically shaped by de novo sequence mutation. For example, studies have found that single nucleotide polymorphisms can have a remarkable impact on the methylome by modifying a methyl transferase or a nucleotide context where methyltransferases act [182–186]. Therefore, the relevance of somatic mutations cannot be dismissed, and could contribute to the generation of genetic or epigenetic variation in natural clonal lineages.

Deciphering the contributions of genetic and epigenetic variation to invasion success is consequently complicated even in clonal plants ([31]; e.g. [128]). Structural and regulatory mutations can have large effects on phenotype, and mutation may be common enough to fuel adaptation even in the short time frame of an invasion [10–12]. This is thought to be specifically true for the multi-locus traits which have increased opportunities for mutations to occur [187]. Dlugosch et al. [11] argued that in the introduced range, a greater number of mutations may be advantageous, and fast population growth provides more opportunities for new mutations to become fixed. They also argue that copy number variation (CNV) mutations occur almost as often as point mutations but more frequently contribute to the emergence of beneficial phenotypes [3,11]. Further, TEs activated by environmental stress can create novel genetic variation, and/or epigenetic regulation, and further contribute to adaptation (reviewed in [3,12]). Understanding how important these genomic (and epigenomic) processes are in clonal plant invasions is currently constrained by low genomic resolution and will require developing a relevant reference genome for species of interest like the knotweed, alligator weed and Spartina species [31,188]. Further work should combine improved genomic resources with a known history of invasion on the landscape to explore how somatic mutation and exposure to novel environments contribute to genomic and epigenomic dynamics that can be preserved uniquely in clonal plants and contribute to invasions. Even with these genomics resources in place, it will be challenging to capture the rare events that might shape evolutionary trajectories [32,33,186].

4. Contribution of hybridization and polyploidy to invasive potential

Like the pervasiveness of clonal reproduction, a growing number of studies have emphasized the potential importance of hybridization and polyploidy in invasions [10,189–194]. These genome-level processes are known to be significant drivers of speciation and genetic diversity in plants, occur frequently throughout plant evolution, and involve genetic and epigenetic changes [18–20,195–197]. An assessment of ploidy levels for 128 globally invasive taxa by te Beest et al. [198] determined that 64.8% of screened species contained polyploid cytotypes. In a systematic review and meta-analysis, Hovick & Whitney [199] found that in 14 established hybrid plant taxa, the hybrids were significantly more fecund and larger than their parental taxa. In a global analysis of plant species, Pandit et al. [191] compared ploidy level and chromosome number in 890 native and invasive congeners from 62 genera in 27 families, and found that invasiveness was positively related to chromosome number and ploidy (see also [190,193,200]). As plant history is punctuated by recurrent cycles of genome duplication and diploidization [18,201], quantitative estimations of ‘polyploids’ and ‘diploids’ (that are actually more or less diploidized paleopolyploids), and comparative studies (e.g. to address whether ‘polyploids’ are more invasive than ‘diploids’) need to be considered in an evolutionary context. Young hybrid and polyploid complexes appear to be especially relevant to address such questions in invasions. Several studies have emphasized the rapid range expansion of recent hybrid or polyploid lineages (e.g. in knotweeds, [127,202,203]; in Phragmites australis, [162,204–206]; in Spartina, [133–135]), which suggests a major immediate ecological consequence of hybridization and genome doubling.

Hybridization is generally thought to facilitate successful invasions through the transgressive segregation of traits, whereby extreme novel phenotypes are produced through the recombination of parental alleles [198,207–209]. Even intraspecific hybridization can facilitate successful invasions through increasing genetic diversity and heterosis (e.g. S. alterniflora; [134,138,153]). Kagawa & Takimoto [208] modelled transgressive segregation in theoretical hybrid species and found that adaptive radiation into suitable novel niches was most likely to occur in hybrid offspring from parental species with moderate genetic differentiation [208,209]. However, several recent studies support that hybridization between closely related lineages within species, between more divergent lineages within species or between species can contribute to standing levels of diversity, transgressive segregation, increased heterozygosity or purging of deleterious alleles [10,192]. In addition, artificial selection for vigorous hybrids during intentional introductions (e.g. in Spartina; [138,139]) and introgression following recurrent backcrosses between hybrids and the parental species can contribute to the invasion process ([210]; e.g. in Quercus petraea and Q. robur; [211]). The vigorous introgressant hybrids between S. alterniflora and the native S. foliosa in California exhibited greater male fitness (viable pollen production of the hybrid was 400 times that of the native plants) and they rapidly invaded the San Francisco Bay area [212].

Broad ecological tolerance is also generally reported in polyploid species [198,200,213–216], and may be attributed to a large range of mechanisms, involving intrinsic (e.g. evolutionary history and genome dynamics) or external (e.g. selective) evolutionary forces. As with hybridization, polyploidy events are often maladaptive. However, genome doubling can mask deleterious mutations through gene redundancy, release constraints on gene function or result in offspring that exhibit heterosis, thereby increasing the potential for adaptive radiation and invasion success [198,208,209,217,218]. Polyploids may arise from diverse pathways in natural populations [219], ranging from autopolyploidy (genome duplication within species) to allopolyploidy (hybrid genome duplication), resulting in duplication of more or less divergent genomes [15]. The relative abundance of auto- versus allopolyploids has been much debated (reviewed in [220]), but ecological success and invasiveness has been widely correlated with allopolyploidy. This increased ecological success could be related to the combined advantages of the heterosis effects that are a consequence of hybridization, with the fertility, genetic redundancy and plasticity provided by genome duplication [221].

In practice, the importance of hybridization and polyploidization in invasions can be difficult to tease apart [192], especially since seed plants have undergone several polyploidization events in their evolutionary history [196,222,223], and deciphering the autopolypoid or hybrid (allopolyploid) nature of ancient polyploids is an additional challenge [201]. We know, however, that the formation of hybrids and polyploids results in wide-scale genomic changes, sometimes referred to as ‘genomic shock’ due to the genetic and epigenetic processes involved in the merging of genomes or whole-genome duplications, as well as subsequent deletions of genomic regions, and chromosomal rearrangements [21,217,224]. Most of this knowledge has been gained from experimentally resynthesized hybrids or allopolyploids [17,20], and these phenomena remain underexplored in natural populations.

At the genomic level, the two components of the allopolyploid speciation process (namely differentiated genome merger and genome redundancy) have important, though different effects, especially with regard to epigenetic consequences like changes in DNA methylation, small RNAs and subsequent gene expression patterns [36,51,132], and these mechanisms are poorly explored in invasive species. The recent hybridization and genome duplication events that occurred in the genus Spartina provide an excellent opportunity to explore the impacts of these different genome-level phenomena. Comparisons between the F1 hybrid S. × towsendii (resulting from hybridization between the hexaploids S. maritima and S. alterniflora) and its allododecaploid derivative S. anglica, which formed in the nineteenth century, allow for distinguishing the effects of hybridization and genome duplication per se [69,225]. During the allopolyploidization process, hybridization (rather than genome doubling) appears to have induced major DNA methylation alterations in S. × townsendii, which were transmitted to the invasive allododecaploid S. anglica [36]. Most of these changes affected regions flanking TEs [226]. Post-transcriptional gene regulation (via microRNAs), as well as repeat-associated small-interfering RNAs (targeting repetitive sequences) changes were also detected following genome merger and genome duplication [51].

The parental hexaploid Spartina genomes have similar repetitive DNA contents (ca. 45%), but higher amounts of TEs are found in S. maritima (ca. 751 Mb/2C) than in S. alterniflora (724 Mb/2C; [227]). Transcriptomic analyses indicated TE repression following hybridization and genome duplication [228], which is congruent with previously reported changes in DNA methylation. Epigenetic control of TEs has been shown to affect the expression of neighbouring genes (e.g. [229]), and in Spartina, both hybridization and genome doubling entailed gene expression evolution, with non-additive parental expression patterns affecting genes involved in stress tolerance and epigenetic regulation [157,228]. Non-additive patterns of parental expression contributed to enhance plasticity in gene expression and putatively adaptive responses to fluctuating environments observed in the allopolyploid populations [158]. The increased tolerance to xenobiotic stress recently reported in S. anglica compared to its parental species could also be partly due to these novel expression patterns [146]. Additional studies in greenhouse or natural field experiments correlating specific changes in DNA methylation, small RNAs, TE activity and gene expression to phenotypic responses, performance and expansion of invasive Spartina populations will be important for understanding how these processes contribute to invasion in these populations. This approach would be even more insightful when compared with the same processes in native populations of the parental species (sensu studies of S. alterniflora in North America and China, [138,139]).

Other studies suggest that Spartina species have high stress tolerance and increased phenotypic variation and tend to occur more frequently in newly created, harsh and recently disturbed environments [230]. Further, phenotypic plasticity in newly formed hybrids promotes niche expansion [198,230–232]. The vigorous intercontinental invader S. densiflora is a heptaploid created from hybridization between a hexaploid and a tetraploid species, and is native to the South American Atlantic coast [233]. Grewell et al. [230] reported high phenotypic plasticity in response to salinity in invasive populations of S. densiflora on the Pacific coast of North America, but an examination of the F1 offspring of S. maritima and S. densiflora showed that Spartina hybrids had even greater tolerance to salinity than their parental species [207]. In California, where S. densiflora hybridized with the native S. foliosa, the hybrids exhibit higher salt tolerance than their parents [234]. Spartina densiflora has also invaded the southern coasts of the Iberian Peninsula, hybridized with the native S. maritima and produced hybrids that grow better than the parent species in most cases [235]. However, the molecular-level phenomena that accompanied these hybridization events have not yet been explored.

Precisely how polyploidy and hybridization might influence invasion success is highly dependent on whether these microevolutionary processes exist within a founding population or arise later as a result of introduction [198,236]. Hence, the apparent association of hybridization or increased ploidy with traits that appear to impart invasiveness may not reflect that hybridization or increased ploidy caused invasion success, but may instead be coincidental with the invasion process. Detecting causal linkages may be challenging since traits with high adaptive potential at one life stage could be deleterious either at another life stage or under diverse environmental conditions [236–238]. Ploidy level and hybridization are both known to affect cell size and biomass [239–242], as well as seed weight and size, suggesting that each could play a role in vegetative and reproductive trait variation [243]. Polyploidization is also known to result in changes in flowering phenology between diploid and polyploid cytotypes [198]. For example, this partitioning influenced the persistence of a novel cytotype of Anacamptis pyramidalis that might otherwise have been outcompeted by its diploid progenitor [244]. Niche differentiation of allopolyploid species relative to their progenitors could help explain the abundance of invasive allopolyploid hybrids [198].

A growing body of evidence supports that the evolutionary phenomena that we have reviewed are important for successful plant invasions [198,215,245–247]. Anthropogenic forces such as the global plant trade, increased land use change, habitat fragmentation and climate change have been proposed to increase opportunities for hybridization and polyploidization events that lead to invasiveness [245]. These forces create novel environments and shape new ecological niches that existing and novel hybrids and polyploids may exploit [19,232]. New tools and approaches are now available for a better understanding of the molecular mechanisms underlying hybridization and polyploidy [20,248] as well as phenotypic plasticity [85,249]. Such studies so far remain limited to a few model systems or rely mostly on phenotypic associations that accompany changes in ploidy without identifying the underlying molecular mechanisms [19]. Further work should explore how genomic and epigenomic dynamics associated with natural hybridization and polyploidization contribute to invasions across the landscape in systems like the Spartina species. Documenting changes in natural populations combined with experimental manipulations in the greenhouse or field experiments within this type of system could allow for a deeper understanding of the link between genome-level processes and performance.

5. Conclusion and future perspectives

In studies of natural populations and through ecological experiments, we have some level of information about how genomic and epigenetic mechanisms (mainly DNA methylation) may contribute to organismal response to environmental challenges [31–33], but these ideas have rarely been applied to the understanding of invasions [87]. Further, most of our knowledge about ecological epigenetics is fairly coarse-grained and lacking critical fine-scale genomic context and understanding of the dynamics involved in clonal plant propagation and the processes of hybridization and polyploidization—processes that are often involved in invasion. Better understanding of other molecular epigenetic mechanisms, like the action of small RNAs, chromatin modifications and cellular location, is also required to truly flesh out how epigenetic mechanisms interact with each other to contribute to heredity along with genetic and other non-genetic mechanisms that ultimately translate into organismal performance [4,6,32,33].

In spite of the progress that has been made in investigating the molecular machinery involved in epigenetic processes, a priority for understanding the importance of epigenetics in invasions (and ecology and evolution more generally) in the coming years must be to develop tools that enable higher-resolution probing of genomes in more realistic scenarios, and in a wider diversity of systems [31–33,188]. This is a priority because epigenetic variation can be dramatically shaped by chance genetic variants [182–186]. Further, the idea that epigenetic signatures may correlate with invasion relies partly on the assumption that changes in DNA methylation are involved in the regulation of gene expression, particularly since methylation of gene promoters has been associated with gene silencing (reviewed in [188]). However, studies over the last 10 years do not always support this idea since many transcription factors show increased DNA binding affinity for methylated DNA [250]. Although DNA methylation of the 5-prime end of genes has been correlated with gene silencing in plants, the functional relevance of gene body methylation varies by context and across taxa, and it is not always correlated to gene expression [50,251,252]. Changes in gene expression have also been shown to cause variation in patterns of DNA methylation [253,254], e.g. in nearby TEs [255]. Understanding the causal patterns of DNA methylation will require further study and combination of methylation data with multiple levels of molecular variation, plant traits and performance.

Similarly, epigenetic effects are thought to be important in clonal plant expansion and in phenotypic plasticity more generally, but the molecular basis and genetic architecture of these processes are not fully understood [85,172,173,249]. Deciphering the role of epigenetics in clonal plant evolution and in plasticity will require targeted experiments to investigate genes underlying plasticity, with classic designs that measure plant responses across different environments [249], and how this might uniquely vary, given the biology of clonal plants [118,172,175]. This is a complicated question considering that most traits are complex, epigenetic effects will impact multiple regulatory developmental switches, and associated genetic regulatory networks (GRNs) have not been identified for many traits [85]. Organisms integrate complex environmental conditions into the expression of traits, and reaction norms of individual traits may change differently in the context of novel conditions. Evaluating multidimensional response could, therefore, be critical to understanding the importance of molecular-level mechanisms that underlie phenotypic plasticity (i.e. multidimensional phenotypic plasticity or MDPP; [238]). The gene networks underlying flowering time are among the best described, and in that case, hundreds of genes are involved in multiple pathways [256,257]. Mutant and transgenic studies have identified individual genes that respond directly to specific environmental factors, but also that natural polymorphisms (i.e. FRIGIDA and Flowering Locus C) can result in dramatic phenotypic differences [101,258,259]. But how multiple layers of molecular mechanisms coordinate expression of these genes and incorporate environmental cues to appropriately regulate flowering time at the level of the whole plant remains unresolved [260].

Perhaps as critical as the development of genomics resources, we have argued extensively about the importance of appropriate experimental design, and that phenotypic plasticity and epigenetic response are not interchangeable terms: there are plastic responses that are not epigenetic, including provisioning and biochemical function [31,32,75,76,261]. In the context of understanding the importance of hybridization and/or genome duplications in invasions, future studies could follow the lead of previous experiments which have taken advantage of synthetic hybrids ([20]; e.g. [100,262–265]) and synthetic or recent polyploids ([17,20]; e.g. [266,267]) to examine the effects of these processes on molecular, phenotypic and ecological levels. This type of approach combined with simultaneous measures of epigenomic, transcriptomic, metabolomic and plant trait response under natural environmental challenges will be important to understand the complexity of organismal response in invasions.

Investigations into the epigenetic effects involved in clonal plant expansion, in natural hybridization and in polyploids are still in their infancy, with most research predominantly investigating methylation [248]. This is especially true for invasive species [87]. In the postgenomic era, as whole-genome sequences from natural, non-model organisms accumulate, we should be able to get a more accurate view of the genomic, epigenomic and transcriptomic landscapes of invasive clonal plants, hybrids and polyploids in various ecological and evolutionary contexts. Although the genetic and epigenetic bases underlying defence mechanisms of invasive plants are virtually unexplored, increased access to omics approaches and tools to analyse (epi)genetic markers are providing novel insights for this perspective. Integrating epigenetics, expression and defence chemical production (i.e. metabolome) to predict molecular-level mechanisms involved in invasion could provide novel insights for understanding the success of invasive species.

Data accessibility

This article has no additional data.

Authors' contributions

All authors made substantial contributions to the conception, drafting or revising of the manuscript. (All authors will give final approval of the version to be published.)

Competing interests

We declare we have no competing interests.

Funding

This work was supported by a Chateaubriand STEM Fellowship through the Office for Science & Technology (OST) of the United States Embassy of France (to J.M.), the German Research Foundation (DFG; grant no. 431595342 to O.B., C.R. and B.L.), the National Science Foundation of China (grant no. 31961133028 to B.L., O.B. and C.R.) and the Deutscher Akademische Austauschdienst (DAAD; MOPGA Project ID 306055 to C.R.).