Plastic responses to environmental stress can be transmitted across plant generations via multiple mechanisms, independently or even in the absence of DNA sequence variation (Cortijo et al., 2014; Zhang et al., 2013). In particular, adaptive transgenerational plasticity can be mediated by starch reserves, mRNAs, proteins, hormones, and other primary and secondary metabolites packaged in the seed (Roach and Wulff, 1987; Leishman et al., 2000; Fenner and Thompson, 2005; Moles and Leishman, 2008), and by epigenetic mechanisms (Rossiter, 1996; Boyko et al., 2010; Richards et al., 2017), via the alteration of gene expression through heritable changes in cytosine methylation or histone modification (Richards, 2006; Bird, 2007; Richards et al., 2017). Since provisioning effects are mediated directly by maternal individuals, environmental effects that persist for multiple generations must be mediated by mechanisms capable of longer-term stability (e.g. epigenetic mechanisms). However, these mechanisms are neither completely separate nor mutually exclusive. On the contrary, multiple mechanisms can cooperatively influence heritable phenotypes (see paragraph: "Combined effects of transgenerational plasticity mechanisms").

a. Seed provisioning and maternally derived proteins and mRNAs

Seed provisioning refers to the carbohydrate, lipid, protein, and mineral nutrient reserves stored by the mother plant in the developing seed (Koller, 1972; Srivastava, 2002). It is often reduced in the deprivation of resources in maternal plants, resulting in an offspring with diminished early growth rates, seedling size, and competitive ability (Haig and Westoby, 1988; Fenner and Thompson, 2005), causing maladaptive transgenerational effects. Alternatively, resource-deprived maternal plants can maintain or even increase seed provisioning (Roach and Wulff, 1987; Schmitt et al., 1992; Sultan, 1996, 2001; Donohue and Schmitt, 1998), maximizing in this way seedling survival. For example, well-provisioned offspring can produce more extensive root systems in dry soil or larger shoot systems under canopy shade (Silvertown, 1984; Wulff, 1986; Leishman et al., 2000). In natural populations, however, the adaptive benefit of such enhanced provisioning may be limited. Resource-deprived maternal plants, in such cases, tend to produce fewer seeds with more chances of surviving. Furthermore, an increased seed provisioning can correlate with decreased persistence in the soil seed bank (Sultan, 1996; Donohue and Schmitt, 1998; Fenner and Thompson, 2005). Hence, in stressful maternal conditions, transgenerational effects mediated via seed provisioning can promote offspring success in specific ecological settings.

Besides seed provisioning, stressed maternal plants can transmit to the offspring also proteins, mRNAs, small RNAs, secondary metabolites, and hormones. Maternally-derived proteins can act both as regulatory molecules and as nutritive elements (via seed provisioning). Together with maternally-derived mRNAs, they are key regulators of seed dormancy and germination (Donohue, 2009). Since stress can significantly alter maternal gene expression, maternally-derived mRNAs and proteins may facilitate adaptive growth responses in seeds germinating under stressful conditions (Rajjou et al., 2004).

b. Epigenetic inheritance: DNA methylation, histone modifications and small RNAs

Epigenetic mechanisms can also mediate transgenerational effects. DNA methylation seems to be both environmentally sensitive and heritable over multiple (i.e., ≥8) generations (Johannes et al., 2009; Reinders et al., 2009; Hauser et al., 2011; reviewed by Jablonka and Raz, 2009). Therefore, DNA methylation (and possibly other epigenetic mechanisms) may also play a role in regulating transgenerational effects of environmental stress (Kalisz and Purugganan, 2004; Grant-Downton and Dickinson, 2006; Boyko and Kovalchuk, 2011). After exposing A. thaliana plants to salt stress, Boyko et al. (2010) found an increased tolerance to the same stress in the progeny that correlated with the inheritance of stress-induced DNA methylation marks. In tobacco plants, infection with tobacco mosaic virus (TMV) also caused heritable changes in DNA methylation. It increased resistance to viral, bacterial, and fungal pathogens in the progeny (Kathiria et al., 2010).

To some degree, the transgenerational effects mediated by DNA methylation seem to be genotype-specific (Herman and Sultan, 2016; Rendina González et al., 2018). Herman and Sultan (2016) investigated the transmission mechanisms of adaptive transgenerational responses to drought stress in different genetic lines of the annual plant Polygonum persicaria. The offspring of the drought-stressed parental plants were treated with the demethylating agent zebularine and grown in dry soil. These plants did not present the adaptive phenotypes shown by the naturally methylated offspring (more extended root systems and greater biomass), suggesting that demethylation removed the adaptive effect of parental drought stress (without significantly altering phenotypic expression in offspring of well-watered parents). Since the seed provisioning between offspring of drought-stressed and well-watered parents was equivalent, differential seed provisioning did not contribute to the effect of parental drought on offspring phenotypes. Furthermore, the effect of demethylation on the expression of the parental drought effect was found to differ among the genetic lines. These results suggest that DNA methylation can mediate adaptive, genotype-specific effects of parental stress on offspring phenotypes. However, demethylation of the whole genome is not targeted and may result in random demethylation variation among replicate lines (see also “Chapter 3: Plant defense response”) or may activate previously inactive transposable elements (see also “Chapter 4: Epigenetics in evolution”).

Despite being to some extent genotype-specific, plastic transgenerational responses to environmental stress can occur even in the absence of genetic variation. To investigate this aspect, a potent approach is the use of A. thaliana epigenetic recombinant inbred lines (epiRILs), characterized by high DNA methylation variation but no DNA sequence variation (Johannes et al., 2009; Reinders et al., 2009; Teixeira et al., 2009). In A. thaliana, studies of epiRILs showed that DNA methylation variants can cause substantial heritable variation in key traits such as primary root length and flowering time (Cortijo et al., 2014; Zhang et al., 2013). In particular, Zhang et al. (2013) tested the response of a large number of epiRILs of A. thaliana to drought and increased nutrient conditions, and they found significant variance components and heritabilities in several phenotypic traits, including flowering time, plant height and total biomass, fruit number, and root:shoot ratio. Thus, this study provides evidence that variation in DNA methylation can cause substantial heritable variation of ecologically important plant traits even in the absence of genetic variation.

In sexually reproducing plants, another layer of complexity is represented by the parental sex, since the inheritance of epigenetic modifications mediating transgenerational effects seems to be sex-specific. In the yellow monkeyflower plants (Mimulus guttatus), Akkerman et al. (2016) tested for differences between maternal and paternal transmission of the transgenerational induction of increased glandular trichome density in response to simulated insect damage. Both maternal and paternal damage resulted in similar and additive increases in progeny trichome production. Notably, the treatment of germinating seeds with 5-azaC erased the effect of maternal but not paternal damage. These results indicate that transgenerational effects can occur through maternal and paternal germlines, but they differ in the proximate mechanism of epigenetic inheritance.

Not only DNA methylation, but even histone modifications can affect gene expression by altering chromatin structure. Histone modifications can also be transferred from one generation to the other, as shown in A. thaliana (Lang-Mladek et al., 2010). Lang-Mladek et al. (2010) found that both heat stress and UV-B exposure could induce heritable changes in gene expression in A. thaliana, correlating with histone H3 deacetylation and with no DNA methylation changes. This effect was found only in small groups of cells within the plant and persisted for two offspring generations.

In addition to DNA methylation and histone modifications, small RNAs (sRNAs) can also have a role in plant transgenerational effects. Changes in sRNA composition have been associated with heat (Ito et al., 2011; Bilichak et al., 2015; Song et al., 2016), drought (Matsui et al., 2008; Tricker et al., 2012), salinity (Borsani et al., 2005; Matsui et al., 2008; Ding et al., 2009; Song et al., 2016), cold, and osmotic stress (Song et al., 2016), and in some cases they even persisted in the offspring of stressed plants (Bilichak et al., 2015; Morgado et al., 2017). In A. thaliana, mutants in the biogenesis of sRNA showed compromised transgenerational caterpillar herbivore resistance (Rasmann et al., 2012), suggesting that sRNAs were required to sustain induced defense responses across generations.

c. Combined effects of transgenerational plasticity mechanisms

As mentioned previously, several modes of transgenerational responses can act together to influence offspring phenotypes. In the wild radish Raphanus raphanistrum, both seed mass-dependent and seed mass-independent transgenerational responses to maternal caterpillar herbivory were found (Agrawal et al., 1999; Agrawal, 2001, 2002). Seed mass is often used as a proxy for seed provisioning (Moles and Leishman, 2008, Lacey et al., 1997). The molecular mechanism behind non-provisioning effects was not investigated in this study (Agrawal, 2001, 2002). However, these results show how even a relatively simple environmental change can induce multiple physiological changes that together enhance offspring performance.

d. Transgenerational plasticity in clonally reproducing plants

Transgenerational effects have been studied mostly across sexual generations, but they have high potential relevance, especially for asexual (or clonal) species (e.g. Latzel and Klimešová, 2010). When plants reproduce clonally, they produce genetically identical offspring not arising from seeds (except in the case of apomixis, in which seeds are asexually produced). Therefore, seed provisioning cannot play a role in transgenerational effects occurring in non-apomictic clonal plants. On the contrary, epigenetic inheritance can occur both across clonal and sexual reproduction (reviewed by Jablonka and Raz, 2009). Epigenetic inheritance even seems to be more prominent across clones, which bypass the epigenetic erasure associated with meiosis (Feng et al., 2010). Thus, it has been suggested that epigenetic inheritance could play a key role in transgenerational plasticity in clonally reproducing plants (Latzel and Klimešová, 2010; Verhoeven and Preite, 2014; González et al., 2016; Rendina González et al., 2018).

González et al. (2016) exposed plants of white clover (Trifolium repens) to different drought treatments and analyzed the transgenerational effects (i.e. offspring biomass) on the untreated clonal offspring. To assess whether DNA methylation was essential to mediate these effects, half of the plants were treated with the demethylating agent 5-azacitidine. In the naturally methylated clonal offspring, they found stress-driven transgenerational effects. However, these effects were not present in the demethylated plants, suggesting that DNA methylation was involved in the observed transgenerational effects.

In the same system, Rendina González et al., 2018 explored whether the effects of transgenerational plasticity were genotype-specific and under epigenetic control. They analyzed the effects of transgenerational plasticity induced by multiple parental stresses on the clonal offspring using five different genotypes. They found that transgenerational plasticity induced by different stresses was genotype-specific and that at least one stress (drought) induced DNA methylation variation that was maintained across several clonal offspring generations. These results suggest that transgenerational effects in Trifolium repens are genotype-specific, potentially under epigenetic control and inherited across several clonal generations.

Despite the abundance of studies investigating the phenotypes produced in response to changing environments, the molecular mechanisms underlying phenotypic variation remain still largely unknown in both plants and animals. Organisms need to process environmental signals and respond appropriately accordingly to their significance. They achieve this through a system common to all organisms based on signal reception, transduction and translation, and the resulting product(s). In particular, environmental stimuli are perceived and transmitted by different signal transduction pathways. These pathways switch on specific transcription factors, which activate genes, ultimately affecting the phenotype.

Notably, in some cases, this process driving phenotypic plasticity was affected by epigenetic mechanisms (Schlichting, 1986; Mirouze and Paszkowski, 2011). Epigenetic mechanisms (e.g. changes in chromatin structure and small RNAs) can regulate gene expression at both the transcriptional and post-transcriptional level. Thus, epigenetic modifications could be a key mechanism in the modulation of gene expression after the environmental signal perception. This hypothesis received support by accumulating evidence that epigenetic modifications may contribute to the regulation of phenotypically plastic traits (e.g. Tatra et al., 2000; Bossdorf et al., 2010; Kooke et al., 2015). One of the best-studied examples is the timing of flowering, in which epigenetic factors evidently play an essential role (He, 2009; Jeong et al., 2015). Proper flowering time is crucial to ensure reproductive success, especially in annual species. Therefore, it is finely regulated by integrating various endogenous and environmental signals to ensure that flowering initiates under the most favorable conditions. In A. thaliana, among other environmental signals, flowering time is regulated by cold exposure through the vernalization pathway. In this pathway, prolonged exposure to low temperatures triggers flowering by silencing the flowering repressor FLOWERING LOCUS C (FLC) through epigenetic modifications (Bastow et al., 2004; Sung and Amasino, 2004; Sung et al., 2006; Finnegan and Dennis, 2007; Greb et al., 2007; Schmitz et al., 2008). This repression is mitotically stable and allows rapid flowering in spring. It functions as a memory of prior cold exposure. However, FLC expression is re-activated during embryogenesis so that the next generation requires vernalization again to accelerate flowering (Sheldon et al., 2008). Therefore, vernalization-mediated FLC repression is mitotically stable but meiotically unstable, and this way reliably times one of the most critical transitions for an annual plant. Clearly, the vernalization response (accelerated flowering) is not immediately triggered by the stimulus (low temperatures); on the contrary, the plant starts flowering when the original stimulus (low temperatures) is removed. This is possible thanks to the epigenetic basis of vernalization, which allows preserving the low-temperature stimulus occurring in winter until the following spring, promoting in this way the floral transition (Lang, 1965).

In conclusion, despite the genetic and epigenetic basis of some phenotypic responses are relatively well characterized (e.g. flowering time), the molecular mechanisms regulating the majority of phenotypic responses remain poorly understood. More studies are thus needed to unravel the role of epigenetic mechanisms in phenotypic plasticity, improving our understanding of the role of epigenetic mechanisms in evolution and adaptation.

The study and understanding of phenotypic plasticity require a multidisciplinary approach, including different areas of biology, such as molecular biology, ecology and evolutionary biology. From one side, the multifaceted nature of phenotypic plasticity justifies its attraction; from the other side, it explains the difficulty to unravel a plastic phenotype in toto. In fact, the ecological and evolutionary role of phenotypic plasticity, as well as the molecular mechanisms regulating its responses, are still relatively unknown.

This chapter provides insight into the current state of knowledge of phenotypic plasticity ranging from its historical definition, passing through the known molecular mechanisms and transgenerational responses, and concluding with the possible role of phenotypic plasticity in evolution and adaptation. These concepts are explained using real examples, emphasizing the importance of epigenetic mechanisms as a key regulator of these responses.

a. Historical insights

Phenotypic plasticity started to gain the interest of the scientific community at the beginning of the 20th century. In 1909, Richard Woltereck performed the first experiments on plastic characters and coined the term "reaction norm" to describe the relationship between the expressions of phenotypic traits across a range of different environments (Schlichting and Pigliucci, 1998). Johannsen (1911) was then the first to distinguish between genotype and phenotype, introducing the concept of genotype-environment interaction, which was developed further by the British developmental biologist Waddington. In particular, Waddington introduced the concept of genetic assimilation, which is how a phenotype initially produced in response to an environmental alteration becomes later genetically encoded (Waddington, 1961). Such a process was identified by studying the phenotypes displayed by Drosophila pupae exposed to environmental stresses (i.e. heat shock or ether vapor) (Waddington, 1953; Waddington, 1956). These phenotypes were indeed initially induced by environmental stress and became then genetically fixed, i.e. they were expressed even without environmental induction. Since these phenotypes were not innate but induced by an environmental event, Waddington argued that genetic assimilation was a change in the pathways of the developmental program due to an environmental change. He claimed that phenotypes result from the interaction between genes and often environment-sensitive developmental factors, which he called for the first time epigenetic factors (Waddington, 1957).

Another conceptual advancement for plasticity research came in 1965 when Anthony Bradshaw suggested that plastic traits may be influenced by natural selection (Bradshaw, 1965). According to his model, plastic traits are environmentally induced and genetically controlled, so that selection can directly act on them. The link between plasticity and evolution was then further developed by Mary-Jane West-Eberhard. She proposed that the genetic accommodation of environmentally induced phenotypes can lead to morphological or behavioral diversification in animals and plants, arguing that phenotypic plasticity plays a crucial role in evolution and speciation (West-Eberhard, 2005).

b. Current understanding in plants

Phenotypic plasticity is today described as the ability of one genotype to produce more phenotypes when exposed to different environments (Pigliucci, 1997). Phenotypic plasticity includes all types of environmentally induced changes, such as physiological, morphological and life-historical traits. Even though it is common in all organisms, plasticity is more broadly expressed in plants, which cannot move away from unfavorable environments because of their sessile nature.

The relationship between phenotype and environment is often represented by a reaction norm (Stearns, 1992; Roff, 1997). In other words, the reaction norm shows the range of phenotypes that a single genotype can express across a range of environments. The reaction norm for any specific trait of a genotype can be visualized as a line or a curve on a two-dimensional plot of the environmental factor (x-axis) versus the phenotype/trait (y-axis) (Fig. 1). The reaction norm is linear when representing two distinct environmental states (e.g. two different temperatures) (Fig. 1A, 2B), while it is linear or non-linear when representing more than two environmental states (e.g. a whole range of temperatures) (Fig. 1D).

The slope of the reaction norm gives hints on the estimation of phenotypic plasticity: a positive (Fig. 1A) or a negative slope (Fig. 1B) implies that the genotype is phenotypically plastic, whereas a flat reaction norm with slope zero represents a non-phenotypically plastic genotype (Fig. 1C).

Figure 1: reaction norms of a single genotype. (A, B, C) Linear reaction norms with positive (A), negative (B) or flat (C) slopes. The positive or negative slopes indicate a phenotypically plastic genotype, while the flat slope a non-phenotypically plastic genotype. (D) Non-linear reaction norm representing more than two environmental states.

Reaction norms representing different genotypes can also be plotted on the same graph, which allows comparing the responses of different genotypes to the same environmental states (Fig. 2). If the two reaction norms are congruent, the two genotypes respond phenotypically in the same way when exposed to the same range of environments (Fig. 2A). When the two reaction norms have identical slope and shape and are shifted along the y-axis, the genotypes show phenotypes that differ on average, but the phenotypic response to the environment is the same (Fig. 2B). When the reaction norms are non-parallel, the two genotypes differ in their phenotypic response to the environment (Fig. 2C-E). Different genotypes of the same species showing non-parallel reaction norms indicate the presence of genotype-by-environment interactions.

Figure 2: reaction norms of two genotypes. (A) Congruent reaction norms indicate that two genotypes display the same phenotypic responses when exposed to the same environments. (B) Reaction norms with identical slope and shape but shifted along the y-axis show genotypes with different phenotypes on average. (C-E) Non-parallel reaction norms indicate that the two genotypes respond phenotypically differently to the same environments.

The sources of phenotypic variation within a population grown in different environments can be summarised by the following equation:

VP = VG + VE + VGxE + Vε.

Here VP is the total phenotypic variance in a trait, VG is the phenotypic variance attributed to differences between the genotypes, VE is the phenotypic variance attributed to differences between the environments, VGxE is the phenotypic variance attributed to the genotype-by-environment interaction, and Vε is the residual or error variance not explained by any of the other sources of variation (Scheiner and Goodnight, 1984).

Notably, each component of this equation can be influenced by epigenetic processes (Banta and Richards, 2018). Epigenetic processes can alter gene expression, ultimately shaping phenotypic variation in response to the environment (Duncan et al., 2014). However, phenotypic responses can also be mediated by indirect physiological and developmental events, such as biochemical mechanisms (reviewed in Kelly et al., 2012), not all of which involve epigenetic processes. For example, extreme temperatures or poor nutrition can directly alter cellular and developmental processes that may also influence complex phenotypic traits.

In summary, phenotypic plasticity arises from diverse mechanisms that can act together, resulting in a wide range of different phenotypes. Although phenotypic plasticity can potentially result also from epigenetic-independent biochemical and nutrient constraints, epigenetic mechanisms can still have a crucial role in mediating many phenotypic responses.

Phenotypic plasticity seems to be adaptive and inherited across generations. Therefore, it can play a significant role in population dynamics and evolution. However, our understanding of the importance of adaptive transgenerational plasticity across different environmental factors and taxa is still limited. For example, plasticity might be even more important in perennial species with long life cycles, respect to annual plants (Walter et al., 2016; Herman and Sultan, 2011). In perennial species, genetic adaptation through natural selection could be indeed too slow to keep pace with rapid environmental changes (but see Yin et al., 2019 for more frequent adaptive transgenerational effects in annual plants).

Phenotypic plasticity may be an adaptive strategy also for clonal plants, as it can allow them to colonize new environments even with low standing genetic diversity. It may be particularly relevant also across clonal generations since it can be mediated by epigenetic mechanisms, which seem to be better maintained across clonal than sexual generations. The transgenerational effects mediated by heritable environmentally-induced epigenetic changes can, therefore, enable a rapid adaptation to changing environments, which infers implications in the short-term microevolution of clonal plants (Latzel and Klimešová, 2010; Verhoeven and Preite, 2014; Dodd and Douhovnikoff, 2016).

Notably, transgenerational plasticity gives rise to adaptive heritable variation precisely when required, as opposed to randomly occurring genetic variation (Verhoeven et al., 2010). Furthermore, particular environmental stress can induce the same adaptive phenotype in various offspring individuals in a population at the same time, as opposed to a phenotype based on genetic variation, which is shared only by part of the population. Consequently, populations can undergo rapid and extensive phenotypic adaptation based on transgenerational plasticity even in the absence of changes in the DNA sequence (Jablonka and Raz, 2009).

Transgenerational plasticity might be relevant also for understanding the ecological impacts of climate change, including global temperature and moisture changes predicted to arise very rapidly due to human activities (IPCC, 2014). Since adaptive transgenerational plasticity can be established even within one generation, it can shortly buffer populations against the immediate effects of climate change and provide time for genetic adaptation or genetic assimilation to act in the long run (Chevin, Lande, & Mace, 2010; Kopp & Matuszewski, 2014). For more details, refer to "Chapter 4: Epigenetics in evolution". In conclusion, our understanding of plant phenotypic plasticity has drastically improved in the last decades. Many phenotypic responses seem to be affected by epigenetic mechanisms, and they seem to facilitate plant adaptation to environmental changes. However, a vast majority of these phenotypic responses seem to be genotype-, species- and taxa-specific, which makes it hard to draw general conclusions on the role of epigenetically-driven phenotypic plasticity in plant adaptation. But then again, it is the nature of local adaptation that the effects are specific to the specific local environment and the genetic properties of the local population. There is thus the need to explore epigenetically-driven phenotypic responses across several genotypes, non-model plant species and taxa.

Overview

‌As sessile organisms of divergent life-history, plants are constantly exposed to a wide range of environmental fluctuations of abiotic conditions (e.g., temperature, drought, precipitation, nutrients) and biotic interactions (e.g., herbivory, pest, mycorrhiza) that can affect their growth, reproduction, and survival. When fluctuations of abiotic or biotic factors become extreme, plants experience stress, and their responses vary across life history, traits, and environmental context. Both for ecology and plant breeding/cultivation, it has become important to understand the means of plant stress responses because changes in plant productivity affect both biodiversity and agriculture. Natural populations can show differences in performance when they are exposed to changes in environmental conditions, partly because of their genetic variation but also because of their epigenetic variation. A general overview of the different types of plant stress factors, their physiological effects, and individual plant defense response mechanisms to overcome their impacts have been previously reported mainly for model and crop species (Mosa et al. 2017; Liza M. Holeski et al. 2012). In earlier chapters, we have discussed phenotypic plasticity and life-history traits in a more general way. In this chapter, we focus on the plant defense response to stresses from the viewpoint of epigenetics in order to understand the full depiction of the plant defense response.

‌

Understanding the role of epigenetic regulation in plant stress responses in the ecological context of natural populations has been a topic that received wide attention in ecological research (Thiebaut et al. 2019; Richards, Alonso, et al. 2017). For untangling the contribution of genotypes and epigenotypes to the stress response in natural populations, our current knowledge needs to be improved by conferring the epigenetic contribution to different types of stress that involves plant phenotypic variation in key traits of populations (Balao et al. 2018). For example, the diversity of life history (reported in Chapter 2) plays a role in the epigenetic regulation of plant stress responses for species with different longevities (annual, perennial, long-lived, etc.) and type of reproduction (sexual, asexual) (Alonso, Medrano et al. 2019). However, to date, the adaptive and evolutionary importance of epigenetic variation in terms of the plant stress response has only rarely been addressed. The incorporation of epigenetics for understanding the plant defense response is only starting (Balao et al. 2018). Different key traits, analysis of varied plant tissue types (root, leaf, bud, etc.), stress relevant features, and 'epiphenotypes' along with DNA sequence information should be analyzed to have more empirical information, which is needed to comprehend plant defense response in natural populations.

In natural plant populations, the most relevant features of environmental stress are the stress intensity, the occurrence frequency, and its predictability, as they will determine the most successful defense strategy to overcome its negative impact on plant performance. These features are properties of the natural environment of a population and are usually not taken into consideration in studies with model species. The defense strategies that plants can develop in nature after multidimensional biotic and abiotic stress exposure are highly related to the environments in which they live. In general, the intensity and frequency of stress factors tend to be inversely correlated, and the defense strategy is most likely selected by the dominant stress features the plants experience (derived from Grativol et al. 2012 and Walter et al. 2013). Further, the predictability of stress or the reliability of the environmental cue is important that plants can prepare a suitable plastic response, either within- or across generations (Reed et al. 2010). And, at which stage of lifespan a cue is received may also dictate the response strategy for different stresses. Studies suggested that when the environmental cue is persistent, plants also show stress-induced or environmentally induced defense mediated by epigenetic variation that could be transmitted over generations (Mauch-Mani et al. 2017). Both within and across generations, the stability of phenotypic responses depends on the degree and the predictability of environmental variation and on the (epi)genetic architecture (Herman et al. 2014). Therefore, the defense strategies of plants are selected according to the particular stress features in their natural environment, and they are specific for individuals or populations within a species and can also differ from species to species.

‌Plants use stress features as cues to fine-tune their defense responses so that they can minimize detrimental effects on plant fitness (Brown and Rant 2013). We can describe three general ways a plant can respond in accordance to the frequency and intensity of the stress: tolerance, avoidance, and induced defense (Walter et al. 2013; Grativol et al. 2012) (Fig 1).

‌

i) Induced defense: Plants show enhanced activation of induced defense when the intensity of stress is low but is encountered more frequently. Under such circumstances, plant performance will be higher if they have a plastic and reversible defense mechanism and, through prior frequent stress cues, an improved responsiveness or priming/acclimation (Fig 1). This means plants acquire this kind of defense by receiving frequent stimuli from pathogens, arthropods, chemicals, and abiotic cues that can trigger the establishment of priming (Mauch-Mani et al. 2017).

‌ii) Tolerance: Plants can show a neutral response of tolerance or a constitutive strategy where plants perform uniformly (same fitness and reproduction) regardless of the intensity and the frequency of stress occurrence. e.g., exposure to heavy metals or a high concentration of salts select plants with higher tolerance against that specific factor (Yaish 2013).

‌iii) Avoidance: Plants show avoidance or an escape strategy only during less frequent stress events of strong intensity because this strategy is usually associated with high costs. For example, fires, flooding, or complete defoliation events often favor avoidance by re-sprouting from below-ground shoots (Boyko & Kovalchuk 2008, "Epigenetic Control of Plant Stress Response"; Gong and Zhang, 2014).

Otherwise, plants also can have negative responses or damage after stress and eventually can suffer greater damage or even complete collapse when stress is recurrent.

Fig 1. Schematic representation of the relationship between plant performance and stress frequency and intensity according to the three plant defense strategies explained within the text. The two axes are continuous and relative. The X-axis denotes the usually inverse relationship between frequency and intensity for any given stress factor, and the Y-axis denotes plant performance, with zero being the average performance for the standard conditions at a certain environment. The dashed-dotted, straight, and dashed lines indicate the induced, the tolerance, and the avoidance plant defense strategy, respectively.

This ability of the immobile plants to survive under fluctuating conditions is sometimes aided by epigenetic mechanisms that can store information at a potentially low cost (Boyko & Kovalchuk 2008, "Epigenetic Control of Plant Stress Response"; Kranner et al. 2010; Grativol et al. 2012). Epigenetic regulation involves histone variants, histone post-translational modifications, small RNA, and DNA methylation that together alter the chromatin structure and determine changes in individual phenotypes without changing the DNA sequence. In plants, methylation of the 5th carbon of the DNA nuclein base cytosine (DNA methylation hereafter) is found within three sequence contexts along the genome: CG, CHG (H = A, T, C), and CHH. DNA methylation regulates the activation and movement of transposable elements and the expression of genes (see Chapter 6 for details). Furthermore, different histones (H2A, H2B, H3, and H4) can be covalently modified at different positions (mostly lysine and arginine residues) by different chemical marks (see Chapter 7 for details). Finally, small regulatory RNAs (sRNAs; approximately 21–24 nt in size, see Chapter 8 for details) also have emerged as important regulators of gene expression. The main aim of this chapter is to review how these epigenetic factors can contribute to the plants' stress responses and discuss how to fill the gaps in our current understanding, as well as how to untangle the genetic and epigenetic contributions.

This book was written and the lectures were recorded in the framework of the European Training Network “EpiDiverse” funded from the EU Horizon 2020 program under Marie Skłodowska-Curie grant agreement No 764965.

Authors: Adam Nunn, Adrián Contreras Garrido, Anupoma Niloya Troyee, Bárbara Díez Rodríguez, Bhumika Dubay, Cristian Peña, Daniela Ramos-Cruz, Dario Galanti, Iris Sammarco, María Estefanía López, Morgane van Antro, Nilay Can, Paloma Perez-Bello Gil, Panpan Zhang, Samar Fatma

Lecturers: Detlef Weigel, Katrin Heer, Vitek Latzel

Editors: Christian Lampei, Katrin Heer, Lars Opgenoorth

Getting started with the EpiDiverse online learning platform.

EpiDiverse is a Marie Skłodowska-Curie Innovative Training Network aimed at the study of epigenetic variation in wild plant species. The network joins research groups from ecology, molecular (epi)genetics, and bioinformatics to explore the genomic basis, molecular mechanisms, and ecological significance of epigenetic variation in natural plant populations.

The 15 (former) Ph.D. students of the network decided that they wanted to provide a textbook and online learning material for beginners in Ecological Plant Epigenetics. Or in other words, they wanted to help others with the kind of material, they were missing when they started this project in 2017. The platform consists of an introductory textbook, the Epidiverse toolkit, and lectures. We hope that you like the result and that it helps you to get started in this exciting research field.

Agrawal A. 2002. Herbivory and maternal effects: mechanisms and consequences of transgenerational induced plant resistance. Ecology 83: 3408–3415.

Agrawal, A. A. 2001. Transgenerational consequences of plant responses to herbivory: An adaptive maternal effect? American Naturalist 157:555–569

Agrawal, A., Laforsch, C., and Tollrian, R. (1999). Transgenerational induction of defences in animals and plants. Nature 401, 60–63.

Akkerman, K. C., Sattarin, A., Kelly, J. K., & Scoville, A. G. (2016). Transgenerational plasticity is sex-dependent and persistent in yellow monkeyflower (Mimulus guttatus). Environmental epigenetics, 2(2), dvw003. doi:10.1093/eep/dvw003

Alexander H. M. and Wulff R. D. 1985. Experimental ecological genetics in Plantago: X. The effects of maternal temperature on seed and seedling characters in Plantago lanceolata. J. Ecol. 73: 271–282.

Andalo C. et al. 1998. Direct and maternal effects of elevated CO2 on early root growth of germinating Arabidopsis thaliana seedlings. Ann. Bot. 81: 405–411.

Badyaev, A. V. & Uller, T. Parental effects in ecology and evolution: Mechanisms, processes and implications. Philos. Trans. R. Soc. B Biol. Sci. 364, 1169–1177 (2009).

Banta and Richards, 2018. Quantitative epigenetics and evolution. Heredity 121, pages 210–224

Bastow, R., Mylne, J.S., Lister, C., Lippman, Z., Martienssen, R.A., and Dean, C. (2004). Vernalisation requires epigenetic silencing of FLC by histone methylation. Nature. 427, 164–167.

Bilichak A, Ilnytskyy Y, Woycicki R, Kepeshchuk N, Fogen D, Kovalchuk I. 2015. The elucidation of stress memory inheritance in Brassica rapa plants. Front Plant Sci. 6:5

Bird A., 2007. Perceptions of epigenetics. Nature volume 447, pages 396–398

Borsani O, Zhu JH, Verslues PE, Sunkar R, Zhu JK. 2005. Endogenous siRNAs derived from a pair of natural cis-antisense transcripts regulate salt tolerance in Arabidopsis. Cell 1237:1279–1291

Bossdorf O. et al. 2009. Genotype and maternal environment affect belowground interactions between Arabidopsis thaliana and its competitors. Oikos 118: 1541–1551.

Bossdorf, O., Arcuri, D., Richards, C.L. et al. Evol Ecol (2010) 24: 541. https://doi.org/10.1007/s10682-010-9372-7

Boyko, A., and Kovalchuk, I. (2011). Genome instability and epigenetic modification – heritable responses to environmental stress? Curr. Opin. Plant Biol. 14, 260–266.

Boyko, A., Blevins, T., Yao, Y. L., Golubov, A., Bilichak, A., Ilnytskyy, Y., Hollander, J., Meins, F., and Kovalchuk, I. (2010). Transgenerational adaptation of Arabidopsis to stress requires DNA methylation and the function of dicer-like proteins. PLoS ONE 5, e9514. doi: 10.1371/journal.pone.0009514

Bradshaw AD. 1965 Evolutionary significance of phenotypic plasticity in plants. Adv. Genet. 13, 115–155.

C.H. Waddington. The Strategy of the Genes - 1st Edition - C.H. Waddington - Routledge. https://www.routledge.com/The-Strategy-of-the-Genes/Waddington/p/book/9781138998131.

Callahan HS, Pigliucci M & Schlichting CD (1997) Developmental phenotypic plasticity: where ecology and evolution meet molecular biology. BioEssays 19: 519–525.

Case, A., Lacey, E., and Hopkins, R. (1996). Parental effects in Plantago lanceolata L. II. Manipulation of grandparental temperature and parental flowering time. Heredity 76, 287–295.

Chevin, L.-M., Lande, R., & Mace, G. M. (2010). Adaptation, plasticity, and extinction in a changing environment: Towards a predictive theory. PLoS Biology, 8, e1000357. https://doi.org/10.1371/journal.pbio.1000357

Colicchio, J. M. & Herman, J. Empirical patterns of environmental variation favor adaptive transgenerational plasticity. Ecol. Evol. 10, 1648–1665 (2020).

Colicchio, J. Transgenerational effects alter plant defence and resistance in nature. J. Evol. Biol. 30, 664–680 (2017).

Cortijo Set al.2014. Mapping the epigenetic basis of complex traits. Science 343, 1145–1148. (doi:10.1126/science.1248127)

Diggle, P. K., Abrahamson, N. J., Baker, R. L., Barnes, M. G., Koontz, T. L., Lay, C. R., Medeiros, J. S., Murgel, J. L., Shaner, M. G. M., Simpson, H. L., Wu, C. C., and Marshall, D. L. (2010). Dynamics of maternal and paternal effects on embryo and seed development in wild radish (Raphanus sativus). Ann. Bot. 106, 309–319.

Ding D, Zhang LF, Wang H, Liu ZJ, Zhang ZX, Zheng YL. 2009. Differential expression of miRNAs in response to salt stress in maize roots. Ann Bot. 1031:29–38.

Dodd and Douhovnikoff, 2016. Adjusting to Global Change through Clonal Growth and Epigenetic Variation. Frontiers in Ecology and Evolution. https://www.frontiersin.org/article/10.3389/fevo.2016.00086

Donohue, K. (2009). Completing the cycle: maternal effects as the missing link in plant life histories. Philos. Trans. R. Soc. Lond. B Biol. Sci. 364, 1059–1074.

Donohue, K., and J. Schmitt. 1998. Maternal environmental effects in plants. Pages 137–158 in T. A. Mousseau and C. W. Fox, editors. Maternal effects as adaptations. Oxford University Press, New York, New York, USA

Duncan EJ, Gluckman PD, Dearden PK (2014) Epigenetics, plasticity, and evolution: How do we link epigenetic change to phenotype? J Exp Zool B Mol Dev Evol 322(4):208–220

Feng, S. H., S. E. Jacobsen, and W. Reik. 2010. Epigenetic reprogramming in plant and animal development. Science 330:622–627.

Fenner, M., and Thompson, K. (2005). The Ecology of Seeds. Cambridge: Cambridge University Press

Finnegan, E.J., and Dennis, E.S. (2007). Vernalization-induced trimethylation of histone H3 lysine 27 at FLC is not maintained in mitotically quiescent cells. Curr. Biol. 17, 1978–1983.

Galloway, L. F. & Etterson, J. R. Transgenerational plasticity is adaptive in the wild. Science (80-. ). 318, 1134–1136 (2007).

Galloway, L. F. 1995. Response to natural environmental heterogeneity: maternal effects and selection on life-history characters and plasticities in Mimulus guttatus. Evolution 49: 1095–1107

González, A. P., Chrtek, J., Dobrev, P. I., Dumalasová, V., Fehrer, J., Mráz, P. and Latzel, V. (2016), Stress‐induced memory alters growth of clonal offspring of white clover (Trifolium repens). American Journal of Botany, 103: 1567-1574. doi:10.3732/ajb.1500526

Grant-Downton, R., and Dickinson, H. (2006). Epigenetics and its implications for plant biology. 2. The "epigenetic epiphany": epigenetics, evolution and beyond. Ann. Bot. 97, 11–27.

Greb, T., Mylne, J.S., Crevillen, P., Geraldo, N., An, H., Gendall, A.R., and Dean, C. (2007). The PHD finger protein VRN5 functions in the epigenetic silencing of Arabidopsis FLC. Curr. Biol. 17, 73–78.

Groot, M. P. et al. Effects of multi-generational stress exposure and offspring environment on the expression and persistence of transgenerational effects in Arabidopsis thaliana. PLoS One 11, 151566 (2016).

Haig, D., and Westoby, M. (1988). "Inclusive fitness, seed resources, and maternal care" in Plant Reproductive Ecology, eds J. Lovett-Doust, and L. Lovett-Doust (Oxford: Oxford University Press), 60–79.

Hauser, M. T., Aufsatz, W., Jonak, C., and Luschnig, C. (2011). Transgenerational epigenetic inheritance in plants. Biochim. Biophys. Acta 1809, 459–468.

He, 2009. Control of the transition to flowering by chromatin modifications. Molecular plant, ISSN: 1674-2052, Vol: 2, Issue: 4, Page: 554-564.

Ito H, Gaubert H, Bucher E, Mirouze M, Vaillant I, Paszkowski J. 2011. An siRNA pathway prevents transgenerational retrotransposition in plants subjected to stress. Nature 4727341:115.

Jablonka, E., and Raz, G. (2009). Transgenerational epigenetic inheritance: prevalence, mechanisms, and implications for the study of heredity and evolution. Q. Rev. Biol. 84, 131–176.

Jacob J. Herman, Sonia E. Sultan, 2016. DNA methylation mediates genetic variation for adaptive transgenerational plasticity283Proceedings of the Royal Society B: Biological Sciences http://doi.org/10.1098/rspb.2016.0988

Jeong H. J., Yang J., Yi J. & An G. Controlling flowering time by histone methylation and acetylation in arabidopsis and rice. J. Plant Biol. 58, 203–210 (2015).

Johannes, F., Porcher, E., Teixeira, F., Saliba-Colombani, V., Simon, M., Agier, N., Bulski, A., Albuisson, J., Fabiana, H., Audigier, P., Bouchez, D., Dillmann, C., Guerche, P., Hospital, F., and Colot, V. (2009). Assessing the impact of transgenerational epigenetic variation on complex traits. PLoS Genet. 5, e1000530. doi: 10.1371/journal.pgen.1000530

Johannsen Schlichting CD, Pigliucci M, 1998. Phenotypic evolution. Sunderland, MA: Sinauer Associates.

Johannsen, W. The genotype conception of heredity1. Int. J. Epidemiol. 43, 989–1000 (2014).

Kalisz, S., and Purugganan, M. D. (2004). Epialleles via DNA methylation: consequences for plant evolution. Trends Ecol. Evol. (Amst.) 19, 309–314.

Kathiria, P., Sidler, C., Golubov, A., Kalischuk, M., Kawchuk, L. M., and Kovalchuk, I. (2010). Tobacco mosaic virus infection results in an increase in recombination frequency and resistance to viral, bacterial, and fungal pathogens in the progeny of infected tobacco plants. Plant Physiol. 153, 1859–1870.

Kelly SA, Panhuis TM, Stoehr AM (2012) Phenotypic plasticity: molecular mechanisms and adaptive significance. Compr Physiol 2(2):1417–1439

Koller, D. (1972). "Environmental control of seed germination," in Seed Biology, ed. T. T. Kozlowski (New York: Academic Press), 2–102.

Kooke R, Johannes F, Wardenaar R, et al. Epigenetic basis of morphological variation and phenotypic plasticity in Arabidopsis thaliana. Plant Cell. 2015;27(2):337-48.

Kopp, M., & Matuszewski, S. (2014). Rapid evolution of quantitative traits: Theoretical perspectives. Evolutionary Applications, 7, 169–191. https://doi.org/10.1111/Eva.12127

Lacey E. P. 1996. Parental effects in Plantago lanceolata L. I. A growth chamber experiment to examine pre‐ and post‐ zygotic temperature effects. Evolution 50: 865–878.

Lacey, E., Smith, S., and Case, A. (1997). Parental effects on seed mass: seed coat but not embryo/endosperm effects. Am. J. Bot. 84, 1617–1620.

Lampei, C., Metz, J. & Tielbörger, K. Clinal population divergence in an adaptive parental environmental effect that adjusts seed banking. New Phytol. 214, 1230–1244 (2017).

Lang A. Physiology of flower initiation. Berlin: Springer-Verlag; 1965

Lang-Mladek, C., Popova, O., Kiok, K., Berlinger, M., Rakic, B., Aufsatz, W., Jonak, C., Hauser, M.-T., and Luschnig, C. (2010). Transgenerational inheritance and resetting of stress-induced loss of epigenetic gene silencing in Arabidopsis. Mol. Plant 3, 594–602.

Latzel, V., and Klimešová, J. (2010). Transgenerational plasticity in clonal plants. Evol. Ecol. 24, 1537–1543.

Latzel, V., Janeček, Š., Doležal, J., Klimešová, J. and Bossdorf, O. (2014), Adaptive transgenerational plasticity in the perennial Plantago lanceolata. Oikos, 123: 41-46. doi:10.1111/j.1600-0706.2013.00537.x

Leimar O, McNamara JM. 2015The evolution of transgenerational integration of information in heterogeneous environments. Am. Nat. 185, E55–E69. (doi:10.1086/679575)

Leishman, M. R., Wright, I. J., Moles, A. T., and Westoby, M. (2000). "The evolutionary ecology of seed size," in Seeds: The Ecology of Regeneration in Plant Communities, ed. M. Fenner (Wallingford: CAB International), 31–57.

Matsui A, Ishida J, Morosawa T, Mochizuki Y, Kaminuma E, Endo TA, Okamoto M, Nambara E, Nakajima M, Kawashima M., et al. 2008. Arabidopsis transcriptome analysis under drought, cold, high-salinity and ABA treatment conditions using a tiling array. Plant Cell Physiol. 498:1135–1149.

Mazer, S., and Gorchov, D. (1996). Parental effects on progeny phenotype in plants: distinguishing genetic and environmental causes. Evolution 50, 44–53.

McIntyre, P. J. & Strauss, S. Y. Phenotypic and transgenerational plasticity promote local adaptation to sun and shade environments. Evol. Ecol. 28, 229–246 (2014).

Mirouze M, Paszkowski J. Epigenetic contribution to stress adaptation in plants (2011). Current Opinion in Plant Biology. Elsevier

Moles, A. T., and Leishman, M. R. (2008). "The seedling as part of a plant's life history strategy," in Seedling Ecology and Evolution, eds M. A. Leck, V. T. Parker, and R. L. Simpson (New York: Cambridge University Press), 217–238.

Morgado, L., Preite, V., Oplaat, C., Anava, S., Ferreira de Carvalho, J., Rechavi, O., Johannes, F., … Verhoeven, K. (2017). Small RNAs Reflect Grandparental Environments in Apomictic Dandelion. Molecular biology and evolution, 34(8), 2035-2040.

Rajjou, L., Gallardo, K., Debeaujon, I., Vandekerckhove, J., Job, C., and Job, D. (2004). The effect of alpha-amanitin on the Arabidopsis seed proteome highlights the distinct roles of stored and neosynthesized mRNAs during germination. Plant Physiol. 134, 1598–1613.

Rasmann S, De Vos M, Casteel CL, Tian DL, Halitschke R, Sun JY, Agrawal AA, Felton GW, Jander G. 2012. Herbivory in the previous generation primes plants for enhanced insect resistance. Plant Physiol. 1582:854–863.

Reinders, J., Wulff, B. B., Mirouze, M., Mari-Ordonez, A., Dapp, M., Rozhon, W., Bucher, E., Theiler, G., and Paszkowski, J. (2009). Compromised stability of DNA methylation and transposon immobilization in mosaic Arabidopsis epigenomes. Genes Dev. 23, 939–950.

Rendina González, A. P., Preite, V., Verhoeven, K., & Latzel, V. (2018). Transgenerational Effects and Epigenetic Memory in the Clonal Plant Trifolium repens. Frontiers in plant science, 9, 1677. doi:10.3389/fpls.2018.01677

Richards, C. L., Alonso, C. , Becker, C. , Bossdorf, O. , Bucher, E. , Colomé‐Tatché, M. , Durka, W. , Engelhardt, J. , Gaspar, B. , Gogol‐Döring, A. , Grosse, I. , van Gurp, T. P., Heer, K. , Kronholm, I. , Lampei, C. , Latzel, V. , Mirouze, M. , Opgenoorth, L. , Paun, O. , Prohaska, S. J., Rensing, S. A., Stadler, P. F., Trucchi, E. , Ullrich, K. and Verhoeven, K. J. (2017), Ecological plant epigenetics: Evidence from model and non‐model species, and the way forward. Ecol Lett, 20: 1576-1590. doi:10.1111/ele.12858

Richards, E.J. (2006). Inherited epigenetic variation – revisiting soft inheritance. Nat. Rev. Genet., 7, 395–401.

Riginos C. et al. 2007. Maternal effects of drought stress and inbreeding in Impatiens capensis (Balsaminaceae). Am. J. Bot. 94: 1984–1991.

Roach, D. A., and R. D. Wulff. 1987. Maternal effects in plants. Annual Review of Ecology and Systematics 18:209–235.

Roff, DA (1997) Evolutionary Quantitative Genetics. New York: Chapman and Hall.

Rossiter, M.C. (1996). Incidence and consequences of inherited environmental effects. Annu. Rev. Ecol. Syst., 27, 451–476.

Salinas, S., Brown, S. C., Mangel, M. & Munch, S. B. Non-genetic inheritance and changing environments. Non-Genetic Inherit. 1, (2013).

Scheiner SM, Goodnight CJ. The comparison of phenotypic plasticity and genetic variation in populations of the grass Danthonia spicata. Evolution. 1984;38(4):845–855. doi: 10.1111/j.1558-5646.1984.tb00356.x

Schlichting C D. The Evolution of Phenotypic Plasticity in Plants (1986).. Annual Review of Ecology and Systematics 1986 17:1, 667-693

Schmid B. and Dolt C. 1994. Effects of maternal and paternal environment and genotype on offspring phenotype in Solidago altissima L. Evolution 48: 1525–1549.

Schmitt, J., J. Niles, and R. D. Wulff. 1992. Norms of reaction of seed traits to maternal environments in Plantago lanceolata. American Naturalist 139:451–466

Schmitz, R.J., Sung, S., and Amasino, R.M. (2008). Histone arginine methylation is required for vernalization-induced epigenetic silencing of FLC in winter-annual Arabidopsis thaliana. Proc. Natl Acad. Sci. U S A. 105, 411–416

Sheldon, C.C., Hills, M.J., Lister, C., Dean, C., Dennis, E.S., and Peacock, W.J. (2008). Resetting of FLOWERING LOCUS C expression after epigenetic repression by vernalization. Proc. Natl Acad. Sci. U S A. 105, 2214–2219

Silvertown, J. W. (1984). Phenotypic variety in seed germination behavior: the ontogeny and evolution of somatic polymorphism in seeds. Am. Nat. 124, 1–16.

Song YP, Ci D, Tian M, Zhang DQ. 2016. Stable methylation of a non-coding RNA gene regulates gene expression in response to abiotic stress in Populus simonii. J Exp Bot. 675:1477–1492.

Srivastava, L. M. (2002). Plant Growth and Development. San Diego: Academic Press.

Stearns, SC (1992) The Evolution of Life Histories. Oxford: Oxford University Press.

Sultan, S. E. 1996. Phenotypic plasticity for offspring traits in Polygonum persicaria. Ecology 77:1791–1807.

Sultan, S. E. 2001. Phenotypic plasticity for fitness components in Polygonum species of contrasting ecological breadth. Ecology 82:328–343

Sung, S., and Amasino, R.M. (2004). Vernalisation in Arabidopsis thaliana is mediated by the PHD finger protein VIN3. Nature. 427, 159–164

Sung, S., Schmitz, R.J., and Amasino, R.M. (2006). A PHD finger protein involved in both the vernalization and photoperiod pathways in Arabidopsis. Genes Dev. 20, 3244–3248.

Tatra GS, Miranda J, Chinnappa CC, Reid DM (2000). Effect of light quality and 5-azacytidine on genomic methylation and stem elongation in two ecotypes of Stellaria longipes. Physiol Plant 109:313–321

Teixeira, F. K., Heredia, F., Sarazin, A., Roudier, F., Boccara, M., Ciaudo, C., Cruaud, C., Poulain, J., Berdasco, M., Fraga, M. F., Voinnet, O., Wincker, P., Esteller, M., and Colot, V. (2009). A role for RNAi in the selective correction of DNA methylation defects. Science 323, 1600–1604.

Tricker PJ, Gibbings JG, Lopez CMR, Hadley P, Wilkinson MJ. 2012. Low relative humidity triggers RNA-directed de novo DNA methylation and suppression of genes controlling stomatal development. J Exp Bot. 6310:3799–3813.

Verhoeven, K. J. F., Jansen, J. J., Van Dijk, P. J., and Biere, A. (2010). Stress-induced DNA methylation changes and their heritability in asexual dandelions. New Phytol. 185, 1108–1118.

Waddington, C. H. (1957). The Strategy of the Genes (Geo Allen & Unwin, London).

Waddington, C. H. GENETIC ASSIMILATION OF AN ACQUIRED CHARACTER. Evolution (N. Y). 7, 118–126 (1953).

Waddington, C. H. Genetic Assimilation. Adv. Genet. 10, 257–293 (1961).

Walter, J., Harter, D. E. V., Beierkuhnlein, C., & Jentsch, A. (2016). Transgenerational effects of extreme weather: Perennial plant offspring show modified germination, growth and stoichiometry. Journal of Ecology, 104, 1032–1040.

West-Eberhard, M. J. Developmental plasticity and the origin of species differences. Proc. Natl. Acad. Sci. U. S. A. 102, 6543–6549 (2005).

Whittle, C., Otto, S., Johnston, M., and Krochko, J. (2009). Adaptive epigenetic memory of ancestral temperature regime in Arabidopsis thaliana. Botany 87, 650–657.

Wulff, R. (1986). Seed size variation in Desmodium paniculatum. II. Effects on seedling growth and physiological performance. J. Ecol. 74, 99–114.

Yin, J., Zhou, M., Lin, Z., Li, Q. Q. and Zhang, Y. (2019), Transgenerational effects benefit offspring across diverse environments: a meta‐analysis in plants and animals. Ecol Lett. doi:10.1111/ele.13373

Zhang, Y., Fischer, M., Colot, V. and Bossdorf, O. (2013), Epigenetic variation creates potential for evolution of plant phenotypic plasticity. New Phytol, 197: 314-322. doi:10.1111/nph.12010

The term priming is mostly used for biotic interactions (e.g., with herbivores or parasites), while the term acclimation is commonly used for abiotic stress events (e.g., heat or frost). They both denote the sensitivity and responsiveness to stress that results from a prior experience and often causes enhanced defense readiness. For simplicity, we will use the term 'priming' in this section for both biotic and abiotic stressors (Mauch-Mani et al. 2017). Priming is a robust defense because it has no or minimal fitness costs in terms of growth or reproduction, and often it is transient and only activated by a stimulus (Mauch-Mani et al. 2017).

‌Defense priming has been observed in a wide range of plant taxa, including wild species and model organisms. Notably, experimental evidence suggests that epigenetic changes were observed following priming. Several studies have hypothesized that epigenetic changes could influence the way how plants respond to biotic and abiotic stresses (Bruce et al. 2007; Mauch-Mani et al. 2017). The general idea is that an induced defense can associate molecular, biochemical, and physiological cues with stronger and/or activated phenotypic defense mechanisms in primed versus unprimed plants (Fig 2)(Bruce et al. 2007; Mauch-Mani et al. 2017). A possible response could be in the form of DNA hypo-hyper methylation (Boyko et al. 2007; Verhoeven et al. 2010), inducing methylation of small, non-coding RNAs as the stress response.

Preceding frequent abiotic stress can acclimatize plants by inducing a change in the epigenetic state, and the persistence of the induced state can be plastic. The epigenetic regulation of the abiotic stress response is complex in nature and could be interlinked with genetic networks or can be an independent event. From literature, we could observe a gap in in-depth studies conducted in natural populations. Most studies are done in artificial environments with model plants or cultivars. Reviews by Bej and Basak (2017) and Li et. al (2017) combined the information on abiotic factors and the contribution of different epigenetic mechanisms for different species (Bej & Basak 2017; Li et al. 2017). ‌In the natural environment, plants are exposed to many abiotic stresses such as salinity, drought, temperature, heavy metals. Epigenetic control of stress-responsive mechanisms was observed in several plant species under various abiotic stress conditions, such as extreme temperatures (Ding et al. 2019), drought (Huang et al. 2019), salinity (Yang and Guo 2018), herbivory, and pathogen (Holeski 2007). For example, increased salinity was associated with DNA methylation changes, histone acetylation, methylation, and phosphorylation in species like rice, Arabidopsis thaliana, tobacco, and mangrove plants (Kim et al. 2015).

In Arabidopsis thaliana, histone modifications are involved in the drought stress response (Kim et al. 2015). For heat stress, DNA methylation differed between heat-sensitive and heat-tolerant genotypes in rapeseed (Gao et al.). And in forest trees (Cork oak), an interplay between DNA methylation and H3 acetylation was observed at elevated temperatures (Correia et al. 2013). Also, cold stress response in A. thaliana and maize affect DNA methylation and histone acetylation (Steward et al. 2002). An interesting example was recently reported by Song et al. (2015), who found that the alpine subnival plant Chorispora bungeana revealed DNA methylation changes that correlated with the exposure to chilling and freezing. Notably, several of the candidate genes were related to physiological chilling and freezing resistance pathways (Song et al. 2015). In summary, induced plant response following abiotic stress seems to be closely related to epigenetic mechanisms that even take an active role in the acclimatization to changing environmental conditions.

Table 1: List of epigenetic modifications reported for abiotic stress:

Prior exposure to biotic stresses has been reported to have improved the defense response of plants. Studies on this topic have a similar bias towards Arabidopsis and cultivated plants as studies on abiotic stress. Reviews by Alonso et al. (2018) and Mauch-Mani et al. (2017) combined the information on biotic interactions and the epigenetic contribution in the response for several plant species (Alonso, Ramos‐Cruz et al. 2019; Mauch-Mani et al. 2017).‌

In the natural environment, plants encounter biotic stress like the occurrence of herbivores, parasites, and pathogens, or the absence of symbiotic partners, and competition with other plants, and hence plant defense response has coevolved with the evolution of interacting species and consequently developed diverse strategies of plant defense mechanisms (Karban et al. 1999; Holeski et al. 2012). These strategies have provided plant fitness benefits against stressors, e.g., herbivore damage (Baldwin et al. 1998). Priming and somatic memory to stress, like the exposure to a pathogen, has been repeatedly reported in correlation with epigenetic changes (Lämke and Bäurle 2017). The task for the future, as the authors note, is to overcome "correlation" and test causation. In other words, it is yet unresolved if these epigenetic changes are actively involved in the priming response. However, a notable study by Yu et al. (2013) demonstrated that treating Arabidopsis thaliana with a peptide of bacterial origin induced the active demethylation of transposable elements (TE), which mobilized short-interfering RNAs/siRNAs and led to the transcriptional activation of genes involved in the defense against bacteria (Yu et al. 2013). The study further found that DNA demethylation negatively affected the growth of a bacterial pathogen, suggesting a close link between the two.

Table 2: List of epigenetic modification reported for biotic stress:

Understanding the within- and inter-generation plant defense response through epigenetics requires the combined analyses of diverse species, their functional phenotypes, and the associated epigenetic variation. This chapter simultaneously shows the significance of epigenetics and the dire necessity of new studies that also should be conducted long-term with diverse species under conditions better matching the plant's natural environment to observe an ecologically more meaningful plant defense response. I also show that low genomic resolution has hindered the investigation of the correlation between ecological factors and epigenetic mechanisms in non-model organisms.

Furthermore, to have complete knowledge of underlying mechanisms, it is compulsory to have a collaboration between ecologists and molecular biologists to develop a proper toolkit. Following these recommendations, it should be possible to unravel the combined contribution of genetic and epigenetic variation to the expression of phenotypes and contribute to filling current knowledge gaps.

Transgenerational defense induction denotes a change in offspring phenotype guided by the environmental signal in the parental generation. This is a form of priming spanning across two generations. To assess the transmission of a priming signal, it is important to study parental environmental effects for different offspring traits (e.g., time of germination, resource allocation, plant architecture, and chemistry, etc.) (Verhoeven and van Gurp 2012; Liza M. Holeski et al. 2012). Plants in each generation can face combinations of different environmental challenges or stress and can often increase the resistance, growth, and reproduction success of their offspring under similar conditions (Karban et al. 1999; Herman and Sultan 2011).

A recent meta-analysis suggested that transgenerational transmission is also influenced by the developmental stage of parental and offspring during stress exposure, such that the transgenerational effect is stronger if the stress was experienced during the early development of the parental plants (Yin et al. 2019). Stress-relevant features like stress occurrence and its predictability in the parental environment can enhance the defense in offspring for the same stress and thus transmit as transgenerational defense (Yin et al. 2019). For example, Arabidopsis thaliana plants showed enhanced performance when exposed to single parental stress treatment (temperature shock or clipping), but not when two different treatments were combined, suggesting that environmental complexity is an important driver of efficient transgenerational defense (Lampei 2019).

Current studies reported that the transgenerational transmission of stress defense includes specific growth variations that are functionally adaptive to the parental conditions, but there are strong differences between the typical laboratory conditions to the complex environment of natural plant populations (Herman and Sultan 2011). However, first evidence exists that the required predictable environmental conditions for the selection of transgenerational effects indeed exist in nature. Lampei et al. (2017) showed in a recent study with a Mediterranean annual plant that winter precipitation predicted the following winter's seedling densities. In order to reduce competition in the offspring generation (avoidance strategy), plants from high water treatments reduced offspring germination (stronger seed dormancy) proportional to the long-term correlation between precipitation and the following winter's seedling density (Lampei et al. 2017). This example demonstrates the high complexity of the natural environments in which the plants live, but at the same time, it also demonstrates how seemingly random and very noisy variables, such as weather, can produce strong environmental correlations between generations that select for transgenerational plasticity.

But how does the transgenerational defense transmission work on the molecular scale? There are several possible mechanisms, including components of seed provisioning. But epigenetic mechanisms are often indicated as one of the most significant underlying mechanisms of transgenerational stress response (Boyko and Kovalchuk 2010; Luna et al. 2012). There are three major reasons supporting this suggestion.

1. The state of epigenetic modifications can be heritably transmitted, it is principally reversible, and the transition can take place rapidly (see Chapters 6, 7, and 8).

2. Epigenetic modifications are known to shape ecologically meaningful traits that also could be stably inherited. For example, histone modifications have been reported to be responsible for priming and transgenerational plant defense (Bej & Basak 2017; Jablonka and Raz 2009).

3. Epigenetic modifications can keep the memory potentially for a longer time than other types of transgenerational phenotype transmitters, such as components of seed provisioning.

Therefore, it is not unlikely that epigenetic modifications play a role also in the phenotype transition to the next generation. For a more detailed discussion of the molecular transmission of transgenerational plant defenses, refer to Chapter 1.

Phenotypic plasticity can occur both within and across generations. It is defined as within-generation plasticity if it occurs within a single generation: an individual encounters distinct environments and reacts accordingly. Conversely, transgenerational plasticity includes heritable phenotypic responses that persist for multiple generations, in which the conditions experienced in one generation can interact with conditions experienced by subsequent generations (Agrawal et al., 1999; Salinas et al., 2013). It is essential to point out that some authors distinguish even further between "intergenerational" and "transgenerational" plastic responses. Intergenerational responses (or parental effects) refer to phenomena spanning short timescales (i.e. one generation after the initial trigger) (reviewed in: Badyaev & Uller, 2009). Transgenerational effects refer to phenomena persisting for multiple generations in the absence of the initial trigger and thus cannot be ascribed to direct effects of the trigger. For simplicity reasons, in this chapter, we will use the broad definition of "transgenerational" to encompass both of these sub-categories. Furthermore, when not specified either “within-generation” or “transgenerational, we will refer to the broad definition of “transgenerational” plasticity.

Intuitively, one could think that similar conditions cause the expression of analogous within-generation and transgenerational phenotypic responses. However, simulation models suggest a different scenario. Natural selection may act on them independently. In particular, within-generation plasticity seems to be favored in an environment characterized by high temporal variability. In contrast, transgenerational responses are favored when the offspring environment is predictable across generations, showing low temporal variability and a slow rate of environmental change (Leimar and McNamara, 2015). That such environmental conditions exist in nature was recently demonstrated using 120 years of climate records for the United States of America (Colicchio and Herman, 2020). However, selection can also favor transgenerational plasticity in highly variable environments (Lampei et al., 2017). It all depends on the environmental correlations between the parental and the offspring generation. This correlation can have the form of a temporal autocorrelation in, e.g. temperatures (Leimar and McNamara, 2015; Colicchio and Herman, 2020) or a correlation between different environmental variables so that, e.g. the amount of rain in the parental season affects the density of seedlings in the offspring season (Lampei et al., 2017).

In particular, transgenerational plasticity can be adaptive, maladaptive or neutral. Transgenerational plasticity is neutral when the parental environment has no effects on the fitness of the offspring, and it is maladaptive when the parental environment limits the fitness of the offspring (Sultan, 1996; Roach and Wulff, 1987; Galloway, 1995; Donohue and Schmitt, 1998). Transgenerational plasticity is adaptive when the parental environment increases the fitness of the offspring, or in other words, when the parental environment allows the plants to pre-adapt their offspring to the conditions they will experience.

For example, Whittle et al. (2009) reported evidence for adaptive transgenerational responses to heat stress in A. thaliana that persisted over at least two generations. Using a set of homozygous inbred lines, they exposed the F0 and F1 generations to heat stress (30°C), and they found that the F3 progeny increased the reproductive output fivefold under heat stress. This effect persisted into the F3 generation, even when F2 plants were grown at a moderate temperature (23°C) (Whittle et al., 2009). The specific mechanisms leading to this fitness increase in the offspring were, however, not documented. The effects of temperature stress also appear to be heritable and potentially adaptive in the cosmopolitan weed Plantago lanceolata. Case et al. (1996) showed that the effects of cold temperature treatment persisted across two generations, significantly enhancing seed weight and fitness-related leaf and life-history traits in adult grandchild plants. Contrary to what was often assumed, this result clearly showed that transgenerational effects might not be confined to the seedling stage. Furthermore, in this species, the paternal temperature also significantly affected offspring traits via interactions with the maternal temperature environment (Lacey, 1996). Furthermore, the paternal environment may also be relevant in outcrossing species, though less than the maternal environment (Roach and Wulff, 1987; Mazer and Gorchov, 1996; Diggle et al., 2010).

Beyond temperature, also other environmental variables can trigger adaptive transgenerational plasticity. In P. lanceolata, offspring showed improved relative fitness when exposed to maternal nutrient availability (Latzel et al., 2014). In Mimulus guttatus, parental wounding induced resistance against natural herbivory (Colicchio, 2017). Also known for strong adaptive parental effects is the maternal shading status that prepares offspring of several herbs for growing in the open (sun) or under canopy (shade) (Galloway and Etterson, 2007; McIntyre and Strauss, 2014). So, despite the rather specific nature of environments that favor transgenerational plasticity, it appears to be frequently found among plants. However, this seems to depend on life history. In a recent meta-study, transgenerational plasticity was frequent and adaptive among annual plants but less frequent and neutral or negative among perennial plants (Yin et al., 2019).

Transgenerational effects are also genotype‐specific as seed families differed in transgenerational plasticity (Alexander and Wulff, 1985; Schmitt et al., 1992; Schmid and Dolt, 1994; Andalo et al., 1998; Agrawal, 2001, 2002; Riginos et al., 2007; Bossdorf et al., 2009). More recently, these differences were used to demonstrate that the reaction norm of parental effects can correlate with climate variables (Groot et al., 2016). In other words, the differences between genotypes in transgenerational plasticity seem to occur not at random but were likely favored by past selection.

To conclude, we do have evidence from different sources that transgenerational plasticity contributes to plant adaptation. However, for a more comprehensive overview, we need more examples, especially studies that include many genotypes or many species, to increase the generality of conclusions.

History of evolutionary theory

The evolutionary theory as we now know it, originated in the second half of 19th century, when Charles Darwin and Alfred Russel Wallace independently came up with the idea of evolution by natural selection (Darwin 1859; Wallace 1858). This revolutionary theory was opposing the Creationist theory, which considered different species as distinct entities created independently. Darwin and Wallace instead argued that species can produce diverging phenotypes which, given a large enough number of generations, can result in different species. This theory was providing a single explanation for the observation of micro- and macroevolution, justifying the entire diversity of life-forms present on Earth. Through his observations Darwin also understood that the force driving evolution is natural selection, acting on variation arising by chance. In other words, during the lifespan of a generation, slight differences arise stochastically between individuals and are transmitted to the progeny. These slight differences or “variation” make the individuals in the new generation diverge. The new generation is then selected by the environment as only individuals that are more fit for that specific environment will more likely survive and reproduce, propagating further only the “positive variation” arisen by chance. However, how this variation originated and was then passed on to the progeny was an important missing piece. It was not until the beginning of the 20th century that an answer emerged.

When Gregor Mendel published his studies on the heredity of traits in plants (Mendel 1865) he described the heredity of discrete traits and for years his findings seemed to be incompatible with the gradual evolution by natural selection hypothesized by Darwin and Wallace. It was only at the beginning of the 1900s that Darwinian evolution by natural selection and Mendel’s laws of heredity were shown to be compatible and were joined in a common mathematical framework by the work of Ronald Fisher, Sewall Wright and J. B. S. Haldane, giving birth to the field of “Population genetics” (Provine 1978). This unification was initially made possible by showing that continuous traits result from the independent inheritance of several genetic loci with small additive effects (Fischer 1919), solving the apparent contrast between Darwin’s gradual evolution and Mendel’s discrete traits.

This unification, started in the 1920s with the work of population geneticists, continued until the 1940s, further developing the evolutionary theory in what was then called the “Modern synthesis” by Julian Huxley (Huxley 1943). The discovery of DNA in 1953 by James Watson and Francis Crick as the molecule responsible for the inheritance of traits answered the last major question left open. The MS was initially implemented with some numerical differences by its founders (Mayer, Stebbins and Dobzhansky 1950; Smocovitis 1997), but the basic principle was common: natural selection is acting on heritable genetic variation generated by random mutations.

Alonso, C., Medrano, M., Pérez, R., Bazaga, P., & Herrera, C. M. (2017). Tissue-specific response to experimental demethylation at seed germination in the non-model herb Erodium cicutarium. Epigenomes, 1(3), 16.

Alonso, C., Medrano, M., Pérez, R., Canto, A., Parra‐Tabla, V., & Herrera, C. M. (2019). Interspecific variation across angiosperms in global DNA methylation: phylogeny, ecology and plant features in tropical and Mediterranean communities. New Phytologist, 224(2), 949-960.

Alonso, C., Ramos‐Cruz, D., & Becker, C. (2019). The role of plant epigenetics in biotic interactions. New Phytologist, 221(2), 731-737.

Balao, F., Paun, O., & Alonso, C. (2018). Uncovering the contribution of epigenetics to plant phenotypic variation in Mediterranean ecosystems. Plant Biology, 20, 38-49.

Baldwin, I. T. (1998). Jasmonate-induced responses are costly but benefit plants under attack in native populations. Proceedings of the National Academy of Sciences, 95(14), 8113-8118. ‌

Bej, S., & Basak, J. (2017). Abiotic stress induced epigenetic modifications in plants: How much do we know? In Plant epigenetics (pp. 493-512). Springer, Cham.

Berr, A., McCallum, E. J., Alioua, A., Heintz, D., Heitz, T., & Shen, W. H. (2010). Arabidopsis histone methyltransferase SET DOMAIN GROUP8 mediates induction of the jasmonate/ethylene pathway genes in plant defense response to necrotrophic fungi. Plant physiology, 154(3), 1403-1414.

Bian, X. Y., Rasheed, M. S., Seemanpillai, M. J., & Rezaian, M. A. (2006). Analysis of silencing escape of tomato leaf curl virus: an evaluation of the role of DNA methylation. Molecular plant-microbe interactions, 19(6), 614-624.

Bossdorf, O., Richards, C. L., & Pigliucci, M. (2008). Epigenetics for ecologists. Ecology letters, 11(2), 106-115.

Boyko, A., & Kovalchuk, I. (2008). Epigenetic control of plant stress response. Environmental and molecular mutagenesis, 49(1), 61-72.

Boyko, A., & Kovalchuk, I. (2010). Transgenerational response to stress in Arabidopsis thaliana. Plant signaling & behavior, 5(8), 995-998.

‌Boyko, A., Kathiria, P., Zemp, F. J., Yao, Y., Pogribny, I., & Kovalchuk, I. (2007). Transgenerational changes in the genome stability and methylation in pathogen-infected plants: (Virus-induced plant genome instability). Nucleic acids research, 35(5), 1714-1725.

‌Brown, J. K. M., & Rant, J. C. (2013). Fitness costs and trade‐offs of disease resistance and their consequences for breeding arable crops. Plant Pathology, 62, 83-95.

Bruce, T. J., Matthes, M. C., Napier, J. A., & Pickett, J. A. (2007). Stressful “memories” of plants: evidence and possible mechanisms. Plant science, 173(6), 603-608.

‌Chen, Z., Zhang, H., Jablonowski, D., Zhou, X., Ren, X., Hong, X., ... & Gong, Z. (2006). Mutations in ABO1/ELO2, a subunit of holo-Elongator, increase abscisic acid sensitivity and drought tolerance in Arabidopsis thaliana. Molecular and cellular biology, 26(18), 6902-6912.

Choi, C. S., & Sano, H. (2007). Abiotic-stress induces demethylation and transcriptional activation of a gene encoding a glycerophosphodiesterase-like protein in tobacco plants. Molecular Genetics and Genomics, 277(5), 589-600.

Choi, S. M., Song, H. R., Han, S. K., Han, M., Kim, C. Y., Park, J., ... & Noh, B. (2012). HDA19 is required for the repression of salicylic acid biosynthesis and salicylic acid‐mediated defense responses in Arabidopsis. The Plant Journal, 71(1), 135-146.

Correia, B., Valledor, L., Meijon, M., Rodriguez, J. L., Dias, M. C., Santos, C., ... & Pinto, G. (2013). Is the interplay between epigenetic markers related to the acclimation of cork oak plants to high temperatures?. PLoS One, 8(1), e53543.

Ding, Y., Shi, Y., & Yang, S. (2019). Advances and challenges in uncovering cold tolerance regulatory mechanisms in plants. New Phytologist, 222(4), 1690-1704.

Dowen, R. H., Pelizzola, M., Schmitz, R. J., Lister, R., Dowen, J. M., Nery, J. R., ... & Ecker, J. R. (2012). Widespread dynamic DNA methylation in response to biotic stress. Proceedings of the National Academy of Sciences, 109(32), E2183-E2191.

Durrant, W. E., Wang, S., & Dong, X. (2007). Arabidopsis SNI1 and RAD51D regulate both gene transcription and DNA recombination during the defense response. Proceedings of the National Academy of Sciences, 104(10), 4223-4227.

Gao, G., Li, J., Li, H., Li, F., Xu, K., Yan, G., ... & Wu, X. (2014). Comparison of the heat stress induced variations in DNA methylation between heat-tolerant and heat-sensitive rapeseed seedlings. Breeding science, 64(2), 125-133.

Gong, B., & Zhang, G. (2014). Interactions between plants and herbivores: a review of plant defense. Acta Ecologica Sinica, 34(6), 325-336.

Grativol, C., Hemerly, A. S., & Ferreira, P. C. G. (2012). Genetic and epigenetic regulation of stress responses in natural plant populations. Biochimica et Biophysica Acta (BBA)-Gene Regulatory Mechanisms, 1819(2), 176-185.

Hashida, S. N., Uchiyama, T., Martin, C., Kishima, Y., Sano, Y., & Mikami, T. (2006). The temperature-dependent change in methylation of the Antirrhinum transposon Tam3 is controlled by the activity of its transposase. The Plant Cell, 18(1), 104-118.

Herman, J. J., & Sultan, S. E. (2011). Adaptive transgenerational plasticity in plants: case studies, mechanisms, and implications for natural populations. Frontiers in plant science, 2, 102.

Herman, J. J., Spencer, H. G., Donohue, K., & Sultan, S. E. (2014). How stable ‘should’epigenetic modifications be? Insights from adaptive plasticity and bet hedging. Evolution, 68(3), 632-643.

Herrera, C. M., & Bazaga, P. (2011). Untangling individual variation in natural populations: ecological, genetic and epigenetic correlates of long‐term inequality in herbivory. Molecular ecology, 20(8), 1675-1688.

Herrera, C. M., Medrano, M., & Bazaga, P. (2016). Comparative spatial genetics and epigenetics of plant populations: heuristic value and a proof of concept. Molecular Ecology, 25(8), 1653-1664.

‌Holeski, L. M. (2007). Within and between generation phenotypic plasticity in trichome density of Mimulus guttatus. Journal of evolutionary biology, 20(6), 2092-2100.

Holeski, L. M., Jander, G., & Agrawal, A. A. (2012). Transgenerational defense induction and epigenetic inheritance in plants. Trends in ecology & evolution, 27(11), 618-626.

Huang, S., Zhang, A., Jin, J. B., Zhao, B., Wang, T. J., Wu, Y., ... & Xu, Z. Y. (2019). Arabidopsis histone H3K4 demethylase JMJ 17 functions in dehydration stress response. New Phytologist, 223(3), 1372-1387.

Huang, S., Zhang, A., Jin, J. B., Zhao, B., Wang, T. J., Wu, Y., ... & Xu, Z. Y. (2019). Arabidopsis histone H3K4 demethylase JMJ 17 functions in dehydration stress response. New Phytologist, 223(3), 1372-1387.

Jablonka, E., & Raz, G. (2009). Transgenerational epigenetic inheritance: prevalence, mechanisms, and implications for the study of heredity and evolution. The Quarterly review of biology, 84(2), 131-176.

Kanazawa, A., Inaba, J. I., Shimura, H., Otagaki, S., Tsukahara, S., Matsuzawa, A., ... & Masuta, C. (2011). Virus‐mediated efficient induction of epigenetic modifications of endogenous genes with phenotypic changes in plants. The Plant Journal, 65(1), 156-168.

Karban, R., Agrawal, A. A., Thaler, J. S., & Adler, L. S. (1999). Induced plant responses and information content about risk of herbivory. Trends in Ecology & Evolution, 14(11), 443-447.

‌Kim, J. M., Sasaki, T., Ueda, M., Sako, K., & Seki, M. (2015). Chromatin changes in response to drought, salinity, heat, and cold stresses in plants. Frontiers in plant science, 6, 114.

Kim, S. Y., Zhu, T., & Sung, Z. R. (2010). Epigenetic regulation of gene programs by EMF1 and EMF2 in Arabidopsis. Plant Physiology, 152(2), 516-528.

Kovar˘ik, A., Koukalová, B., Bezde˘k, M. et al. (1997). Hypermethylation of tobacco heterochromatic loci in response to osmotic stress. Theor Appl Genet 95, 301–306.

Kranner, I., Minibayeva, F. V., Beckett, R. P., & Seal, C. E. (2010). What is stress? Concepts, definitions and applications in seed science. New Phytologist, 188(3), 655-673.

Kreutz, C., Can, N. S., Bruening, R. S., Meyberg, R., Mérai, Z., Fernandez-Pozo, N., & Rensing, S. A. (2020). A blind and independent benchmark study for detecting differentially methylated regions in plants. Bioinformatics, 36(11), 3314-3321

‌‌Lämke, J., & Bäurle, I. (2017). Epigenetic and chromatin-based mechanisms in environmental stress adaptation and stress memory in plants. Genome biology, 18(1), 1-11.

‌‌Lampei, C. (2019). Multiple simultaneous treatments change plant response from adaptive parental effects to within‐generation plasticity, in Arabidopsis thaliana. Oikos, 128(3), 368-379.

Lampei, C., Metz, J., & Tielbörger, K. (2017). Clinal population divergence in an adaptive parental environmental effect that adjusts seed banking. New Phytologist, 214(3), 1230-1244

Li, S., Castillo‐González, C., Yu, B., & Zhang, X. (2017). The functions of plant small RNA s in development and in stress responses. The Plant Journal, 90(4), 654-670.

Luna, E., Bruce, T. J., Roberts, M. R., Flors, V., & Ton, J. (2012). Next-generation systemic acquired resistance. Plant physiology, 158(2), 844-853.

Mauch-Mani, B., Baccelli, I., Luna, E., & Flors, V. (2017). Defense priming: an adaptive part of induced resistance. Annual review of plant biology, 68, 485-512.

Mosa, K. A., Ismail, A., & Helmy, M. (2017). Introduction to plant stresses. In Plant stress tolerance (pp. 1-19). Springer, Cham.

‌Niederhuth, C. E., & Schmitz, R. J. (2017). Putting DNA methylation in context: from genomes to gene expression in plants. Biochimica et Biophysica Acta (BBA)-Gene Regulatory Mechanisms, 1860(1), 149-156.

Rasmann, S., De Vos, M., Casteel, C. L., Tian, D., Halitschke, R., Sun, J. Y., ... & Jander, G. (2012). Herbivory in the previous generation primes plants for enhanced insect resistance. Plant physiology, 158(2), 854-863.

‌Reed, T. E. et al. (2010). Phenotypic plasticity and population viability: the importance of environmental predictability. Proc. R. Soc. B 277: 3391–400.

‌Richards, C. L., Alonso, C., Becker, C., Bossdorf, O., Bucher, E., Colomé‐Tatché, M., ... & Verhoeven, K. J. (2017). Ecological plant epigenetics: Evidence from model and non‐model species, and the way forward. Ecology letters, 20(12), 1576-1590.

Richards, C. L., Bossdorf, O., & Verhoeven, K. J. (2010). Understanding natural epigenetic variation. New Phytologist, 187(3), 562-564.

‌Schulz, B., Eckstein, R. L., & Durka, W. (2014). Epigenetic variation reflects dynamic habitat conditions in a rare floodplain herb. Molecular ecology, 23(14), 3523-3537.

Shen, Y., Conde e Silva, N., Audonnet, L., Servet, C., Wei, W., & Zhou, D. X. (2014). Over-expression of histone H3K4 demethylase gene JMJ15 enhances salt tolerance in Arabidopsis. Frontiers in plant science, 5, 290.

Slaughter, A., Daniel, X., Flors, V., Luna, E., Hohn, B., & Mauch-Mani, B. (2012). Descendants of primed Arabidopsis plants exhibit resistance to biotic stress. Plant physiology, 158(2), 835-843.

‌Song, Y., Liu, L., Feng, Y., Wei, Y., Yue, X., He, W., ... & An, L. (2015). Chilling-and freezing-induced alterations in cytosine methylation and its association with the cold tolerance of an alpine subnival plant, Chorispora bungeana. PLoS One, 10(8), e0135485.

Steward, N., Ito, M., Yamaguchi, Y., Koizumi, N., & Sano, H. (2002). Periodic DNA methylation in maize nucleosomes and demethylation by environmental stress. Journal of Biological Chemistry, 277(40), 37741-37746.

Steward, N., Ito, M., Yamaguchi, Y., Koizumi, N., & Sano, H. (2002). Periodic DNA methylation in maize nucleosomes and demethylation by environmental stress. Journal of Biological Chemistry, 277(40), 37741-37746.

Steward, N., Kusano, T., & Sano, H. (2000). Expression of ZmMET1, a gene encoding a DNA methyltransferase from maize, is associated not only with DNA replication in actively proliferating cells, but also with altered DNA methylation status in cold-stressed quiescent cells. Nucleic acids research, 28(17), 3250-3259.

Stockinger, E. J., Mao, Y., Regier, M. K., Triezenberg, S. J., & Thomashow, M. F. (2001). Transcriptional adaptor and histone acetyltransferase proteins in Arabidopsis and their interactions with CBF1, a transcriptional activator involved in cold-regulated gene expression. Nucleic acids research, 29(7), 1524-1533.

Sunkar, R., & Zhu, J. K. (2004). Novel and stress-regulated microRNAs and other small RNAs from Arabidopsis. The Plant Cell, 16(8), 2001-2019.

Sunkar, R., Chinnusamy, V., Zhu, J., & Zhu, J. K. (2007). Small RNAs as big players in plant abiotic stress responses and nutrient deprivation. Trends in plant science, 12(7), 301-309.

‌Thiebaut, F., Hemerly, A. S., & Ferreira, P. C. G. (2019). A role for epigenetic regulation in the adaptation and stress responses of non-model plants. Frontiers in Plant Science, 10, 246.

Verhoeven, K. J., & van Gurp, T. P. (2012). Transgenerational effects of stress exposure on offspring phenotypes in apomictic dandelion. PloS one, 7(6), e38605.

Verhoeven, K. J., Jansen, J. J., Van Dijk, P. J., & Biere, A. (2010). Stress‐induced DNA methylation changes and their heritability in asexual dandelions. New Phytologist, 185(4), 1108-1118.

Vlachonasios, K. E., Thomashow, M. F., & Triezenberg, S. J. (2003). Disruption mutations of ADA2b and GCN5 transcriptional adaptor genes dramatically affect Arabidopsis growth, development, and gene expression. The Plant Cell, 15(3), 626-638.

Wada, Y., Miyamoto, K., Kusano, T., & Sano, H. (2004). Association between up-regulation of stress-responsive genes and hypomethylation of genomic DNA in tobacco plants. Molecular Genetics and Genomics, 271(6), 658-666.

Walter, J., Jentsch, A., Beierkuhnlein, C., & Kreyling, J. (2013). Ecological stress memory and cross stress tolerance in plants in the face of climate extremes. Environmental and Experimental Botany, 94, 3-8.

‌‌Yaish, M. W. (2013). DNA methylation-associated epigenetic changes in stress tolerance of plants. In Molecular stress physiology of plants (pp. 427-440). Springer, India.

‌Yang, Y., & Guo, Y. (2018). Elucidating the molecular mechanisms mediating plant salt‐stress responses. New Phytologist, 217(2), 523-539.

‌Yin, J., Zhou, M., Lin, Z., Li, Q. Q., & Zhang, Y. Y. (2019). Transgenerational effects benefit offspring across diverse environments: A meta‐analysis in plants and animals. Ecology Letters, 22(11), 1976-1986.

Yu, A., Lepère, G., Jay, F., Wang, J., Bapaume, L., Wang, Y., ... & Navarro, L. (2013). Dynamics and biological relevance of DNA demethylation in Arabidopsis antibacterial defense. Proceedings of the National Academy of Sciences, 110(6), 2389-2394.

Zhang, Z., Zhang, S., Zhang, Y., Wang, X., Li, D., Li, Q., ... & Bao, S. (2011). Arabidopsis floral initiator SKB1 confers high salt tolerance by regulating transcription and pre-mRNA splicing through altering histone H4R3 and small nuclear ribonucleoprotein LSM4 methylation. The Plant Cell, 23(1), 396-411.

Summary

Plants are continuously exposed to fluctuating environmental conditions. As sessile organisms, they have a restricted capacity to select the features of their environment; hence, it is crucial for them to respond to environmental changes successfully. A way how plants can cope with environmental changes is phenotypic plasticity (i.e. the ability of one genotype to produce more phenotypes when exposed to different environments). Despite the importance of phenotypic plasticity in plant adaptation to environmental changes, the molecular mechanisms underlying phenotypic variation remain still largely obscure.

Phenotypic plasticity seems to be regulated by epigenetic modifications (i.e. histone modifications, DNA methylation and RNA molecules), which affect phenotypes by regulating gene activity. Interestingly, epigenetic modifications are partly heritable across mitotic and (to some extent) meiotic cell divisions, suggesting that they may mediate phenotypic responses even across clonal or sexual generations (i.e. transgenerational plasticity). In particular, epigenetic modifications seem to be better maintained across clonal than sexual generations. Thus, transgenerational plasticity may be more relevant for clonal species, as it may allow them to bypass their potentially low-standing genetic diversity. However, if and when plastic responses are inherited across generations is further complicated by several factors, such as the predictability of the offspring environment and specific plant characteristics (e.g. sex, genotype and species). Furthermore, plastic responses can also occur independently of epigenetic modifications when they are mediated, for example, by physiological and developmental events or maternal seed provisioning. Thus, the combined effects of all these variables make the study and understanding of phenotypic plasticity difficult.

This chapter aims to overview the variables affecting phenotypic plasticity and the potential role of phenotypic plasticity in rapid plant adaptation to environmental changes.

When willing to study the role of epigenetics in plant adaptation and evolution, a fundamental step is to study epigenetic patterns in natural populations. Different environments offer different selective pressures and may induce the presence of different DNA methylation patterns. Moreover, isolation by distance may induce the accumulation of different epimutations in geographically distant populations. The only way to study the occurrence of these phenomena (or whether they occur at all) is to study them in real natural populations. Furthermore, it is important to bear in mind that what happens in one species may differ in another, especially for species with very different life-history traits (see the Chapter “Life history traits”). Several studies have tried to answer this kind of questions but many struggle to overcome the large-scale vs high-accuracy problem (see section “Genetics-Epigenetics”) focusing on either one of the two aspects (Richards et al. 2017). Nevertheless, each approach was able to address some fundamental questions. Studies based on large-scale surveys were able to show that DNA methylation correlates with habitat of origin and environmental variables (Gugger et al. 2016; Lira-Medeirose et al. 2010; Gáspár, Bossdorf, and Durka 2018), with some studies having found correlations with phenotypic traits (see section “Phenotypic effects”). Some experiments analysing subsequent generations also showed that DNA methylation is at least partially heritable (Gáspár, Bossdorf, and Durka 2018), but most of these studies lack the genomic resolution to fully test whether the observed epigenetic variation is independent of genetic polymorphisms. At present, the only plant species where the large-scale vs high-accuracy problem could be effectively overcome is Arabidopsis thaliana. Yet in this species most of the DNA methylation variation seems to be under genetic control (Dubin et al. 2015; Kawakatsu et al. 2016). Nevertheless, Arabidopsis was shown to be an outlier in terms of DNA methylation studies, harbouring a very low global genome methylation and TE content compared to other plant species (Alonso et al. 2015).

In addition to all these approaches, as previously discussed in the section “Phenotypic effects”, few naturally occurring epialleles were discovered in plant populations. These discoveries were made “by chance”, meaning that the studies that found them did not intentionally look for such epigenetic variation. Nevertheless, the presence of naturally occurring pure epialleles constitutes the most evident proof that epigenetic variation could potentially be adaptive.

In a natural setting, plants are exposed to a complex environment, often with multiple stress conditions, and thus, to simulate ecological realism, we should have more than one stress factor in experimental design (see Lampei 2019). To study plant stress response, it is important to establish the causal links between epigenetic variation and phenotypes, as well the interference of genetic and epigenetic variation. Thus, one challenge lies in identifying the relevance of epigenetic factors for modulating the phenotypic response to specific stress factors. To this aim, it may be beneficial to include plant species with different life histories and genomic features. A few approaches have been reported by which epigenetic variation can be potentially measured for larger populations in an ecological setting.

Study system

Where possible, experiments should include several genetically divergent lines or diverse natural populations from one species. Single line, or genotype, experiments may provide very detailed information but lack generality of conclusion. Often, repeating the experiment with another genotype of the same species yields different results. This is not only true for epigenetic experiments but even more important in these because of frequent interactions with the genomic background. However, the enhanced generality comes with an increased complexity of the study. Therefore, epigenetic studies should preferably be conducted in ecological settings that control genotypic variability. For example, non-model species or natural populations that contain a low level of genetic diversity and reproduce asexually like, for instance, clonal plants, provide these properties (Richards, Bossdorf, et al. 2010).

‌Another approach is to study outcrossed species by evaluating both the DNA sequence and DNA methylation profiles of individuals using statistical approaches to understand the relation of genetic and epigenetic variation (Herrera and Bazaga 2011; Schulz et al. 2019). As the field is still developing for non-model species, we must be aware of the trade-off between the depth/resolution of functional information and cost-effectiveness. Following experimental methods to study epigenetic inheritance could help to fill the gaps in the field (Bossdorf et al. 2008).

Experimental method

An interesting option is the use of inhibitors for epigenetic factors such as the DNA methyltransferases (e.g., 5-azacytidine, zebularine) or histone deacetylases (e.g., Trichostatin). Together with knockout mutants, experimental demethylation can be used to establish the link between epigenetic factors and phenotypic response. However, this method has disadvantages in the first place because it is not targeted and reduces DNA methylation scattered across the genome so that connections between phenotype and methylation can be made, but it is often difficult to identify the specific methylation changes that were involved in the response. ‌

The study should choose the molecular mechanisms that are confirmatory and cost-effective to study epigenetics in natural populations. For example, for evaluating the methylation status, Reduced Representation Bisulfite Sequencing methods (RRBS) can also work without the availability of a high-quality genome (Niederhuth and Schmitz 2017), and it is cost-effective. This is a very nice option for studying methylation in non-model species. However, we need more methods with such criteria for studying other epigenetic factors. Quantitative genetics mapping approaches such as Epigenome-wide association studies (EWAS) can be another method to study the approximate genetic and epigenetic associations with the phenotype (Kreutz et al. 2020).

Gaps in current method and empirical knowledge: the transition from model to non-model plants

Representation of plant species:

First and foremost, the representation of diverse plant species in epigenetic studies of plant stress responses is important as most of the existing knowledge is based on studies conducted using model systems or cultivated plants that tend to be plants with a fast life-cycle. To gain generality in ecological settings, we need to study more non-model species with different life histories that can explain the phenotypic consequences and the epigenetic contribution to different stress responses.

Porper toolkit & statistical relevance:

The most studied epigenetic mechanism in plants is DNA methylation, though others like SmallRNA and histones are increasing in frequency. However, while for DNA methylation, reduced representation methods like provide the means to conduct large-scale ecological studies, such methods are mostly non-existent in other epigenetic contexts. This makes it often difficult to study more than DNA methylation in such projects or in plants with large and polyploid genomes.

Distinguish genetic and epigenetic contribution

A prevailing gap includes the lack of measures to untangle genetic and epigenetic contribution as ecological studies are often focused on collections from natural populations in situ. Ecological relationships should be evaluated across environmental gradients to gain an overview of the stress response, paired with an analysis of spatial genetic and epigenetic structure in the wild populations to understand the respective contribution (Herrera, Medrano, et al. 2012).

Understanding the cross-talk of epigentic regulators:

The link between epigenetic regulators (e.g., DNA methylation, histone modifications, and small RNAs) is often missing in current projects. However, it is often indispensable to understand the mechanisms of both the biotic and the abiotic stress response. This is because the cross-talks between epigenetic regulators themselves may have a role in gene expression (Grativol et al. 2012). More synthesis studies are needed to connect the dots in plant stress response by understanding the cross-talk of significant epigenetic regulators.

The nomenclature of “Modern synthesis” and “Population genetics” can be seen as two sides of the same coin, with the first one representing the theoretical framework and the latter the mathematical implementation of our current evolutionary theory. In the most recent version of one of the classic modern textbooks, the basic principles of the MS, i.e., the basic mechanisms driving evolution, were listed by and are summarized in Box1.

Box 1: Evolutionary mechanisms

Natural selection: This is the key process at the base of the modern synthesis. It refers to the differential survival and reproduction rate of individuals as a consequence of their individual phenotype. In a nutshell, individuals with higher fitness for a specific environment, have higher chances of survival and reproduction. Their genes, providing the higher fitness features, will be passed on to the next generation at a higher frequency.

Genetic drift: Genetic drift is a change in allele frequencies due to the random sampling of alleles for the next generation (Masel 2011). It results in random allele frequency changes and depends on population size. The smaller the population, the more likely alleles will be lost compared to the originally larger genepool due to stochasticity. A conceptual example is given in figure 1.

Figure 1: Schematic example of genetic drift. Simply by chance, in generation 1 only four individuals produce progeny and only one of them carries the blue allele. Consequently, in generation 2 only a minority of individuals carry the blue allele.

Genetic hitchhiking: Genetic hitchhiking occurs when an allele changes in frequency not for being under selection itself but for being in linkage with another allele which is under selection. An extreme case of genetic hitchhiking was observed in the model plant Arabidopsis thaliana: a glyphosate resistance mutation occurred in the chloroplast genome causing an increase in the occurrence of the entire nuclear genome associated with the mutation (Flood et al. 2016). The stronger the selection and the quicker it acts, the less recombination events will occur between the positive mutation and the rest of the genome, extending the hitchhiking to normally unlinked parts of the genome.

Epistasis: Epistasis occurs when a locus A is interfering with the phenotypic effect of another locus B. Selection on locus B will therefore be dependent on the allele present at locus A. A straightforward example is eye colour in many animals. For example, in the parasitic wasp Nasonia vitripennis, the compound determining eye colour is synthesized through a biosynthetic pathway. The last steps of the pathway are catalysed by two enzymes that define the eye colour depending on the alleles present on the loci coding the enzymes, which we will call A and B for simplicity (Figure 2).

Figure 2: Example of epistasis. Nasonia vitripennis is a wasp with diploid females and haploid males. This species has Wild Type (WT) individuals with dark eyes, but loss of function mutations (a and b) exist for two genes (A and B) acting at the end of the biosynthetic pathway responsible for a compound of the eye color. As gene B variation produces phenotypic variation only when a functional copy of gene A is present, gene A has an epistatic effect on gene B.

Recessive loss of function mutations exist for both genes A and B. While in the presence of a functional copy of Enzyme A, enzyme B variation produces phenotypic variation (dark eyes if a functional copy of B is present, red eyes otherwise), this is not the case when there are no functional copies of enzyme A (grey eyes regardless of locus B variation).

Pleiotropic effects: Pleiotropic effects are present when a single gene has multiple phenotypic effects. Natural selection on that gene will therefore pass through selection of multiple phenotypes. When one of the phenotypes is selected for, the other affected phenotypes will also change their occurrence in the next generations, even though they are not directly under selection.

Population bottleneck: A population bottleneck refers to a drastic reduction of population size due to environmental events (such as drought, flood, fires, pest attacks…). Such events drastically reduce the genetic variation of the population. Population bottleneck can result in genetic drift when the surviving individuals are selected randomly. Alternatively, a bottleneck can result in an extreme form of natural selection when the surviving individuals are the most fit to overcome the stress event.

Founder effect: Founder effect refers to the low genetic variation observed in new colonizing populations originated from few individuals that naturally poorly represent the gene-pool of the larger population from which they originate. Founder effects are closely related to bottlenecks but involve the colonization of new space for the species. Like in bottlenecks, mostly genetic drift and but also natural selection are the main evolutionary mechanisms at work.

Gene flow: Gene flow or Genetic migration refers to the transfer of genetic material from one population to another. When two populations have a really high gene flow and their genetic material is exchanged without limit, they may be considered a single population.

The discipline of population genetics describes, in mathematical terms, how the mechanisms in Box 1 cause changes in allele frequencies. To do so, the starting point is the null hypothesis in which no changes is allele frequencies are acting on a population and there is random mating and consequently random combination of alleles (individuals carrying allele A have the same chance to mate with individuals carrying allele A and B). In this theoretical null hypothesis, the population reaches the Hardy-Weinberg Equilibrium and there is no variation in allele frequencies between generations. So, in a locus with two alleles A and a, where we indicate the allele frequencies with f = p and f = q respectively, the corresponding genotype frequencies can be represented in a Punnett square (Figure 3) and they will be:

Freq(AA) = p2 Freq(aa) = q2 Freq(Aa) = 2pq

Figure 3: Punnett Square representing allele and genotype (AA, aa, Aa) frequencies in a population harbouring two alleles (A and a) at locus A. Allele frequencies (p and q) are represented by the length of the arrows, genotype frequencies (f(…)) by the area of the squares.

When the population is in HWE, the allele frequencies in the next generation will remain equal and can be calculated as:

p’ = (p2 + 2pq * 1⁄2) = p2 + pq = p(p+q) = p

q’ = (q2 + 2pq * 1⁄2) = q2 + pq = q(p+q) = q

In natural populations, it is possible to test for deviation from the HWE, i.e. to test whether a specific locus is under selection. To do so, it is necessary to measure genetic variation in two subsequent generations and compare genotype frequencies between these. Deviation from the HWE can be tested through a χ2 (chi-square) test.

Population genetics and genomics allow to draw a variety of information from the genetic variability measured in populations and species. In contrast, quantitative genetics focuses on the study of complex quantitative traits controlled by several genes. The name of this discipline refers to the sometimes-miniscule phenotypic differences among individuals that are quantified and utilized in order to unravel the nature of the trait’s genetic basis. More information and examples on the field of population genetics can be found on quantitative genetics in Falconer & Mackay (1996). A technical report of the most recent analysis available in quantitative genetics can be found on chapter “Differential Methylation”.

When we look at the epigenetic machinery as a whole, with its complexity and its tight links to the genetic background, epigenetic mechanisms could contribute to or impact local adaptation and evolution in a few different ways. Excluding plasticity induced genetic accommodation (Table 1), which is not a direct mechanism and is not exclusively driven by epigenetics, we can envision at least four main direct mechanisms in which epigenetic marks could contribute to local adaptation and evolution:

1. Epimutations can arise stochastically and over evolutionary time they could undergo natural selection and, when functionally relevant, produce stable phenotypic variation (see phenotypic plasticity chapter). This would work in a similar manner as genetic mutations, with the difference that epimutations arise at a higher rate than genetic mutations but are also much more likely to revert (Becker et al. 2011; Johannes and Schmitz 2018).

2. Environmentally induced epigenetic variation could arise in response to changing environmental conditions and, if stably inherited over generations, could provide a means of rapid evolution (Richards 2006; Whitelaw and Whitelaw 2006).

3. DNA methylation variation may generate further genetic variation when transposable elements TEs are released (Dubin et al. 2015; Secco et al. 2015). This could result in different environmental cues releasing different TE families or in epigenetically distant accessions being more likely to generate different genetic variation.

4. In addition to these mechanisms, it is relevant to mention that methylated DNA has a higher mutation rate due to 5-methylcytosine being more prone to C to T transitions by deamination (Cambareri et al. 1989; Pfeifer 2006). This could increase the likelihood of epigenetically distant individuals to accumulate genetic mutations in different regions of the genome.

Nevertheless, in order to be taken into account, the main candidate mechanisms need to meet some basic requirements that are discussed in the next chapters. Most importantly, evolutionary relevant epigenetic variation has to be at least partially “stable”, meaning heritable through successive generations (see section “Stability of epigenetic marks”), have an effect on the phenotype (see section “Phenotypic effects”) and should be “pure” i.e. not rely on DNA sequence variation for its heritability (see section “Genetics – epigenetics”).

Despite the efficacy of the Modern synthesis in describing the genetic implications of evolution, further advances were made in the last 20 years and exceptions to the basic assumptions of the MS were identified. These exceptions raised the question of whether these basic assumptions should be revisited or whether additional drivers should be simply added (Wilkins 2008; Laland et al. 2015). Following these findings and this line of reasoning Massimo Pigliucci and Gerd B. Müller postulated a new “Extended Evolutionary Synthesis” (EES; Müller 2007; Pigliucci 2009). This consists of revisiting the basic assumptions of the ES by giving more importance to the role of phenotypes and including additional evolutionary processes, as described in Table 1. The basic assumptions of the EES embrace and extend the ones of the ES. The three main EES assumptions are listed below (for mor details see Laland et al. 2015):

1. Inclusive inheritance: Genetic mutations but also other forms of inheritance generate variation.

2. Reciprocal causation: Evolutionary processes cause phenotypic variation in the new generations but vice versa the phenotypes influence the occurrence of evolutionary processes.

3. Non-random variation: There is evidence that variation is not generated fully randomly. One controversial example is “developmental bias” , referring to the observation that some phenotypes are more likely than others to arise due to developmental processes. Moreover, it was shown that, responding to specific stimuli, extra chromosomal DNA molecules can be replicated to create redundancy of beneficial genes. This mechanism was shown to induce glyphosate resistance in Amaranthus palmeri and constitutes another striking example of non-random genetic variation . Again, some families of transposable elements seem to be preferentially inserted in the proximity of genes and not fully randomly in the genome . Environmentally induced epigenetic variation, if stable through at least few generations, would constitute another case of non-random variation.

Table 1: Evolutionary processes considered by the Extended Evolutionary Synthesis as described in Laland et al. 2015.

Despite the evident effect of some of the EES additional drivers (especially when thinking about human evolution), some authors argue that these additional drivers are only affecting evolution indirectly and they introduce variation that can be already captured by the original drivers (Futuyma 2017). To give real life examples, it is possible to explain humans loss of body hairs as a result of niche construction, removing the selective pressure (cold protection) or to explain it directly with the loss of the selective pressure for that trait. In the same manner, genetic divergence between populations can be explained as the result genetic accommodation following colonization events only provided by plasticity, or directly as the result of different populations being exposed to different selective pressures.

Without getting into the details of such discussion, it is important to point out that transgenerational epigenetics, referring to epigenetic marks stably inherited through generations, when having a phenotypic effect, would be directly influencing evolution and not in an indirect way like other EES mechanisms. For this reason, in this chapter we will primarily focus on stable epigenetic variation (see “Stability of Epigentic marks”). The concept of “stable” strictly depends on the mode of reproduction of the species, hence in asexually reproducing species mitosis-heritable epigenetic marks should be considered stable, while stability through meiosis would be necessary for sexually reproducing species.

This considered, transgenerational epigenetics is one of the most prominent candidates to be included in new evolutionary models. Nevertheless, according to Occam’s razor, a good model should be simple, and the question of whether or not transgenerational epigenetics would significantly increase the accuracy and power of the evolutionary theory is still under debate (Dickins and Barton 2013; Haig 2007; Pigliucci and Finkelman 2014; Mesoudi et al. 2013). It is therefore crucial to determine the relative contribution of epigenetics to evolution, in order to implement it correctly in future evolutionary models.

Single locus epigenetic effects on phenotype

The ultimate requirement for DNA methylation variation to be adaptive, in addition to being stable and not under genetic control, is to have phenotypic relevance. There are several examples of phenotypic effects of epigenetic marks, from heterophylly (Herrera and Bazaga 2013) to stress responses (Kinoshita and Seki 2014) and others, but only few meet the requirements of being stable and “pure”, i.e. not under genetic control. Although rare, there are few examples of pure naturally occurring epimutations at a single locus which cause mutant phenotypes. The most famous example is the peloric mutant of Lynaria vulgaris, in which spontaneous DNA methylation of the Lcyc gene promoter drives a change in flower symmetry from bilateral to radial (Cubas, Vincent, and Coen 1999). Such a phenotype can be transmitted to the next generation and can sometimes revert back to WT during somatic development. This can sometimes result in branches of peloric and WT flowers originating from the same plant, confirming that the Lcyc promoter methylation is pure and not genetically controlled. A second striking example is the naturally occurring hypermethylation of the Colorless non-ripening (Cnr) gene promoter, inhibiting fruit ripening in Tomato (Manning et al. 2006). This epimutation is even more stable than the Lynaria case, as the fruit phenotype only very rarely reverts to WT. Other examples of epialleles were described in few more plant species, some naturally occurring and some resulting from induced mutagenesis. Some examples were listed by (Kalisz and Purugganan 2004) and few cases confidently exclude any chance of genetic control of the described epimutations.

Figure 4: Examples of single locus phenotypic effects of DNA methylation epialleles. WT phenotype is at the top while mutant phenotype at the bottom for A) tomato (Manning et al. 2006), B) Linaria vulgaris (Cubas et al. 1999), and C) oil palm (Ong-Abdulla et al. 2015).

Epigenetic effects on complex traits

Although single locus epialleles are undoubtedly rare, most plant traits are polygenic. Therefore, more insights into epigenetic effects on complex traits are required to understand the potential of epigenetic contribution to plant adaptation. Unfortunately, such studies are so technically difficult that they are basically nonexistent. Showing causation of a single epimutation with a clear phenotype is already a hard task, doing so for putative epigenetic variants controlling quantitative traits is even harder. The only studies that have tackled this challenge and successfully identified DNA methylation variants affecting quantitative traits, so called epiQTLs, made use of A. thaliana epigenetic recombinant inbred lines epiRILs (Cortijo et al. 2014; Furci et al. 2019). Because this population harbours random DNA methylation variation in a uniform genetic background, any heritable phenotypic effect can be directly attributed to the methylation variation. Using this system Cortijo et al. 2014 managed to identify Differentially Methylated Regions explaining 60% to 90% of the flowering time and primary root length variation observed. More recently, hypomethylated DNA loci controlling quantitative resistance to the fungal pathogen Hyaloperonospora arabidopsidis to were also identified in the epiRILs (Furci et al. 2019).

Tight link between genetics and epigenetics

An important complicating factor, hindering advances in determining the importance of epigenetics in evolution, is posed by the tight link between the genome and the epigenome. In simple words, when a genetic polymorphism is controlling an epigenetic pattern, the latter will seem to be heritable and under selection, while in reality this is only true for the causal genetic polymorphism. This can be caused by trans-acting elements at the genome level, for example when a mutation affecting a methyltransferase is responsible for the induction of different DNA methylation patterns and different global DNA methylation (Dubin et al. 2015). Alternatively, this can also happen at the level of a specific locus, driven by cis-acting elements such as TE insertions causing DNA methylation of the neighbouring region. For example, in melon, a TE insertion next to the promoter of CmWIP1 causes silencing of this gene, resulting in a switch from male to female flowers (Martin et al. 2009).

Looking at this the other way around, epigenetic variation can also be in control of genetic variation, as for example TE activity is controlled by DNA methylation (Noshay et al. 2019). This link poses difficulties when studying the phenotypic effects of DNA methylation variation, as an induced change in the DNA methylation patterns (see chemical demethylation), often used to prove its contribution to phenotypic changes, may release TEs. It is therefore difficult to determine whether the phenotypic effect following a DNA methylation change is truly due to this change or due to an unseen TE insertion.

Controlling or excluding genetic variation

To overcome these difficulties, different approaches have been used and studies have found that epigenetic variation can, at least in some cases, be independent from DNA sequence variation (Cubas, Vincent, and Coen 1999; Riddle and Richards 2002; Shindo et al. 2006; Vaughn et al. 2007). These approaches are either based on excluding genetic variation, by looking at epigenetic variation in populations with the same genetic background or accounting for genetic variation. Some common approaches are listed below:

1. Excluding genetic variation: these are the most common approaches because they are simpler and normally less expensive. The downside of these methods is that they do not allow to look at epigenetic variation in natural populations of genetically variable species.

1. Using asexually reproducing species as these have nearly identical genetic background (Vanden Broeck et al. 2018; Heer 2018; Shi et al. 2019).

2. Chemical demethylation: Demethylating agents such as Zebularine and 5-azacytidine can be used to extensively, but randomly, erase DNA methylation from genomes without modifying the genetic background, except for the above-mentioned possibility of TE activation (Latzel 2016; Münzbergova et al. 2018).

3. Methylation mutants: Introducing epigenetic variation through a mutation in the epigenetic machinery and then removing the mutation (Johannes et al. 2009, Cortijo et al. 2014, Kooke el al. 2015, Zhang et al. 2019).

4. Targeted DNA methylation: Using molecular biology techniques, such as transformation with inverted repeat (IR) sequences (Mette et al. 2000; Zicola et al. 2019) or Cas9-methyltransferase complexes, It is possible to methylate specific genomic regions. While Cas9 complexes can be used for any kind of sequence, IR insertions should only be used, when targeting non-transcribed regions or they will also degrade mRNAs through post transcriptional gene silencing.

2. Accounting for genetic variation: this strategy is used to study epigenetic variation in natural populations of species harbouring genetic variation. It is based on statistical methods allowing to determine whether the observed epigenetic variation is controlled by underlying genetic polymorphisms. This approach faces what we will call the large-scale vs high-accuracy problem, meaning it requires both large-scale collections and high-accuracy genomic tools. Large-scale surveys are fundamental to capture enough variation to provide a representative sample of the adaptive ability of a certain species. High-accuracy genomic tools are required at the DNA methylation level in order to capture the majority of the variation and to draw information about a variety of factors. Moreover, high-accuracy at the DNA sequence level, and therefore a reference genome, are also required due to the tight link between genetics and epigenetics. Some papers that adopted this approach in Arabidopsis are (Dubin et al. 2015, Sakani et al. 2019, and Kawakatsu et al. 2016).

In order to determine whether epigenetic marks play a relevant role in evolution, it is crucial to understand to which extent they are stably inherited across generations. While in sexually reproducing species the concept of generation is very straightforward, it is not the same in asexually reproducing species. Therefore, when not specified, we will refer to “stability” as the ability of an epigenetic mark to be inherited through meiosis in sexually reproducing species. Alternatively, when dealing with asexually reproducing species, we will consider epigenetics marks transmittable through mitosis as “stable” as well.

Histone modifications

Histone modifications were shown to be associated with stress memory and to be, in some cases, stably inherited through mitosis in plants, yeast and animals (Shido et al. 2005; Audergon et al. 2015). Nevertheless, they seem to revert back upon meiosis and not to be transgenerationally inherited by the offspring (Pecinka and Mittelsten Scheid 2012). This is consistent with the observation that the parental H3 histone, carrying several modifications related to differential transcription, is replaced during zygote development in Arabidopsis thaliana (Ingouff et al. 2010). Although these findings do not completely rule out the possibility that locus specific histone modifications could be copy-pasted to the zygote, this was never observed so far.

Considering these observations, histone modifications do not seem to be good candidates in playing a role in the evolution of sexually reproducing species as they are not transmissible through meiosis. Nevertheless, this could be different for asexually reproducing species which do not undergo meiosis (Castonguay and Angers 2012). More on histone modifications can be found in Chapter 8.

DNA methylation

When looking at the “origin” of DNA methylation variation from this perspective, we have to consider that while stochastic epimutations appear randomly in the genome (and therefore have the same chance of appearing in different genomic elements) environmentally induced DNA methylation is “intentionally” targeting regions in which it has a function such as promoters and other regulatory regions … (Secco et al. 2015). For this reason, these two origins of DNA methylation variation should be considered separately. Several stress memory experiments were carried out to test for the heritability of environmentally induced DNA methylation and the general trend is that induced variation reverts back to its original state after one or few generations (Secco et al. 2015; Lämke and Bäuerle 2017). Stochastic epimutations, on the other hand, can be studied using Mutation Accumulation Lines with one study showing that stochastic epimutations revert back at a quicker rate than genetic mutations, but can remain stable over several generations (Becker et al. 2011; Johannes and Schmitz 2018).

Genetic variation provides the baseline for phenotypic variation on which evolutionary processes like natural selection can act (Fisher, 1930; Hughes et al., 2008). The magnitude of genetic variation within a population can be quantified in many ways, and it is a fundamental source of biodiversity (Hughes et al., 2008). Although we know relatively little about the range of potential ecological effects (Hughes et al., 2008), there is plenty of evidence that genetic diversity can affect population dynamics (Reusch et al., 2005; Johnson et al., 2006), species interactions (Kagiya et al., 2017), community composition (Booth & Grime, 2003), and ecosystem processes (Hughes & Stachowicz, 2004; Schweitzer et al., 2005; Madritch et al., 2006).

However, recent advances in molecular biology and genomics have shown that genetic variation is not the only cause of phenotypic variation among individuals (Rapp & Wendel, 2005). One of these additional sources of phenotypic plasticity is epigenetic variation (Zhang et al., 2013). Several studies have suggested that epigenetic diversity has a more significant role in phenotypic plasticity than previously thought (Bossdorf et al., 2008; Heer et al., 2018a). It can create variation (heritable or non-heritable) in ecologically important traits such as tree growth, phenology, plant defense responses to herbivory, or even niche width and habitat differentiation (for further details, see chapter 1: "Phenotypic plasticity and adaptation").

In plants, heritable epialleles frequently arise de novo through epimutations in the germline, that is, through stochastic losses or gain of DNA methylation. These heritable epimutations seem to occur mainly at CpG dinucleotides and are highly dependent on genomic context (Taudt et al., 2016), suggesting that genetic variability in plants can influence the levels of DNA methylation (Dubin et al., 2015). Therefore, high genetic diversity can potentially translate into high epigenetic diversity. On the other hand, if epigenetic variation can create heritable variation in functional traits, then epigenetic diversity can, in principle, have positive effects similar to those of DNA sequence diversity on the functioning of populations and ecosystems (Latzel et al., 2013). Epigenetic changes can also be independent of genetic structure and could, in theory, trigger the formation of novel epialleles and promote the movement of DNA transposons that are commonly found in plant genomes. Therefore, novel 'epigenetically induced' heritable phenotypes can increase the ability of plants to adapt to environmental challenges (Richards, 2006; Mirouze & Paszkowski, 2011). Despite recent advances in the field, the effects of epigenetic variation across different ecological organization levels remain poorly understood. However, thanks to modern genomic techniques becoming more affordable and accessible, new efforts have been made to understand the relationship between genetic, epigenetic, and phenotypic variation and the range of effects of epigenetic variation at ecosystem and landscape levels. This chapter will discuss the known effects of genetic and epigenetic diversity and argue that even though more research on the topic is needed, it is safe to assume that epigenetic diversity across large-scale systems may have consequences similar to those of genetic diversity.

Summary

Genetic diversity can be defined as any measure that quantifies the magnitude of genetic variability within a population. In the last two decades, it has been shown to have a strong impact on populations, communities, and entire ecosystems (Rapp & Wendel, 2005; Kagiya et al., 2017). For example, genetic diversity reduces the rate at which species diversity declines in experimental grassland communities (Booth & Grime, 2003), increases species richness, and influences community composition in arthropod communities (Johnson et al., 2005; Witham et al., 2008; Robinson et al., 2012). The genetic diversity of dominant plant species can also affect nutrient flux, for instance, via litter decomposition processes (Bandau et al., 2016). On the other hand, phenotypic plasticity is defined as the ability of one genotype to produce more than one phenotype when exposed to different environments (Kelly et al., 2012). Intraspecific trait variability is a direct result of phenotypic plasticity and contributes to amplify the functional diversity of plant communities, a key component of biodiversity with important implications for species coexistence and ecosystem functioning (Medrano et al., 2014). Therefore, genetic diversity is the baseline for phenotypic diversity on which evolutionary processes like natural selection acts (Hughes et al., 2008). However, in recent years it became evident that epigenetic variation can play a role in phenotypic plasticity (Bossdorf et al., 2008; Heer et al., 2018), and several studies have suggested that epigenetic variation can create functional diversity in populations (Latzel et al., 2013). For example, epigenetic mechanisms play a role in allelopathy, and epigenetic changes might be more determinant than genetic variability in the success of plant invasions (Pérez et al., 2012; Hoffman, 2015; Slotkin, 2016). Furthermore, as explained in previous chapters, epigenetic variation can also have a role in how plants respond to environmental stress conditions (Kinosita & Seki, 2015). Although epimutations may arise spontaneously, a significant fraction of all epigenetic variation found within a population has a genetic and environmental basis (Kawakatsu et al., 2016). It is thus reasonable to assume that epigenetic variation can also influence populations and communities, and processes at the ecosystem or landscape levels.

References

Bandau, Franziska, et al. "Genotypic Variability in Populus Tremula L. Affects How Anthropogenic Nitrogen Enrichment Influences Litter Decomposition." Plant and Soil, vol. 410, no. 1–2, Springer International Publishing, Jan. 2017, pp. 467–81, doi:10.1007/s11104-016-3033-8.

Booth, Rosemary E., and J. Philip Grime. "Effects of Genetic Impoverishment on Plant Community Diversity." Journal of Ecology, vol. 91, no. 5, Wiley/Blackwell (10.1111), Oct. 2003, pp. 721–30, doi:10.1046/j.1365-2745.2003.00804.x.

Bossdorf, Oliver, et al. "Epigenetics for Ecologists." Ecology Letters, vol. 0, no. 0, Wiley/Blackwell (10.1111), Nov. 2007, p. 071117033013002–???, doi:10.1111/j.1461-0248.2007.01130.x.

Heer, Katrin, et al. "The Diversifying Field of Plant Epigenetics." New Phytologist, vol. 217, no. 3, Wiley/Blackwell (10.1111), Feb. 2018, pp. 988–92, doi:10.1111/nph.14985.

Hofmann, Nancy R. "Epigenetic Battles Underfoot: Allelopathy among Plants Can Target Chromatin Modification." The Plant Cell, vol. 27, no. 11, American Society of Plant Biologists, Nov. 2015, p. 3021, doi:10.1105/tpc.15.00916.

Hughes, A. Randall, et al. "Ecological Consequences of Genetic Diversity." Ecology Letters, vol. 11, no. 6, Wiley/Blackwell (10.1111), June 2008, pp. 609–23, doi:10.1111/j.1461-0248.2008.01179.x.

Johnson, Marc T. J., et al. "Additive and Interactive Effects of Plant Genotypic Diversity on Arthropod Communities and Plant Fitness." Ecology Letters, vol. 0, no. 0, Wiley/Blackwell (10.1111), Oct. 2005, p. 051012084514001, doi:10.1111/j.1461-0248.2005.00833.x.

Kagiya, Shinnosuke, et al. "Does Genomic Variation in a Foundation Species Predict Arthropod Community Structure in a Riparian Forest?" Molecular Ecology, vol. 27, no. 5, Wiley/Blackwell (10.1111), Mar. 2018, pp. 1284–95, doi:10.1111/mec.14515.

Kawakatsu, T., Huang, S. shan C., Jupe, F., Sasaki, E., Schmitz, R. J. J., Urich, M. A. A., … Ecker, J. R. (2016). Epigenomic diversity in a Global Collection of Arabidopsis thaliana Accessions. Cell, 166(2), 492–505. https://doi.org/10.1016/j.cell.2016.06.044

Kinoshita, Tetsu, and Motoaki Seki. "Epigenetic Memory for Stress Response and Adaptation in Plants." Plant and Cell Physiology, vol. 55, no. 11, Oxford University Press, Nov. 2014, pp. 1859–63, doi:10.1093/pcp/pcu125.

Latzel, Vít, et al. "Epigenetic Diversity Increases the Productivity and Stability of Plant Populations." Nature Communications, vol. 4, no. 1, Nature Publishing Group, Dec. 2013, p. 2875, doi:10.1038/ncomms3875.

Medrano, Mónica, et al. "Epigenetic Variation Predicts Regional and Local Intraspecific Functional Diversity in a Perennial Herb." Molecular Ecology, vol. 23, no. 20, Wiley/Blackwell (10.1111), Oct. 2014, pp. 4926–38, doi:10.1111/mec.12911.

Pérez, Julio, y Carmen Alfonsi, y Carlos Ramos, y Juan Antonio Gómez, y Carlos Muñoz, y Sinatra K. Salazar. "How some alien species become invasive. Some ecological, genetic and epigenetic basis for bioinvasions". Interciencia, vol. 37, no. 3, 2012, pp. 238-244. Editorial Asociación Interciencia.

Rapp, Ryan A., and Jonathan F. Wendel. "Epigenetics and Plant Evolution." New Phytologist, vol. 168, no. 1, Wiley/Blackwell (10.1111), June 2005, pp. 81–91, doi:10.1111/j.1469-8137.2005.01491.x.

Robinson, K. M., Ingvarsson, P. K., Jansson, S., Albrectsen, B. R., Soares, J., & Macaya-Sanz, D. (2012). Genetic Variation in Functional Traits Influences Arthropod Community Composition in Aspen (Populus tremula L.). PLoS ONE, 7(5), e37679. https://doi.org/10.1371/journal.pone.0037679

Slotkin, R. Keith. "Plant Epigenetics: From Genotype to Phenotype and Back Again." Genome Biology, vol. 17, no. 1, BioMed Central, Dec. 2016, p. 57, doi:10.1186/s13059-016-0920-5.

Whitham, Thomas G., et al. "Extending Genomics to Natural Communities and Ecosystems." Science (New York, N.Y.), vol. 320, no. 5875, American Association for the Advancement of Science, Apr. 2008, pp. 492–95, doi:10.1126/science.1153918.

A. Epigenetic effects at the population level.

We cannot start to understand how epigenetic diversity affects populations without first talking about intraspecific diversity. A species is a group of living individuals that can exchange genes or interbreed. By definition, intraspecific diversity is the variation occurring between individuals belonging to the same species, and it is the main factor underlying diversity at the population level. While many studies on ecosystem functioning focus on the interspecific facet of diversity, diversity also includes an intraspecific facet that is defined by phenotypic, functional, and genetic diversity measured within a single species (Schaaf et al., 2003; Bolnick et al., 2011). Generally speaking, this intraspecific trait variability increases the functional diversity of plant communities, a key component of biodiversity with important implications for species coexistence and ecosystem functioning. Intraspecific genotypic and phenotypic diversity has been demonstrated to account for a big part of the total biodiversity measured in plants and animals, representing in some cases up to a quarter of the total variability measured in communities (Raffard et al. 2018, 2019). In the last couple of decades, it has also become clear that intraspecific variation exists at the epigenetic level, for example, DNA methylation.

Epigenetic variation can be under genetic control, but some parts are independent of it due to spontaneous epimutations (Becker et al., 2011; Van Der Graaf et al., 2015) or following environmental induction (Jiang et al., 2014; Quadrana & Colot, 2016). These independent epimutations hold the most potential for finding novel intraspecific phenotypic differences (Bossdorf et al., 2008; Richards et al., 2017). So far, in‐depth documentation of intraspecific epigenetic variation has been restricted to model plant species such as Arabidopsis thaliana, Oryza sativa, and Zea mays for which extensive genomic and epigenomic resources exist (Becker et al., 2011; Schmitz et al., 2013; Van Der Graaf et al., 2015; Kawakatsu et al., 2016). However, with high throughput sequencing techniques becoming more affordable and accurate every day, new plant species are being added to the pool for which genomic resources are readily available, increasing research on natural genetic and epigenetic variation in non-model species (Richards et al., 2017).

Some studies could show that variation in DNA methylation is common in natural populations exceeding DNA sequence variation when comparing populations from different ecological origins (Herrera & Bazaga, 2010; Richards et al., 2012; Schulz et al., 2014). It is, therefore, essential to consider how epigenetic diversity varies between populations. Take, for example, two populations that are separated in time and/or space and are subjected to different conditions or environmental cues. If environmental conditions can induce the occurrence of new epialleles, these two populations will, in theory, present different epigenetic marks that mirror their specific environmental conditions, and those marks will potentially be inherited by the offspring (Richards, 2006). In line with this expectation, several studies have confirmed that populations show epigenetic differentiation between habitats. For example, in 2012, Richards and colleagues found that variation in some epigenetic loci in Fallopia japonica plants might be associated with specific local conditions (Richards et al., 2012). Another study by Lira-Madeiros et al. (2010) showed that individuals from a single population had similar genetic but divergent epigenetic profiles characteristic for the population in a particular environment. The evidence in these studies suggests that epigenetic variation among populations could reflect local acclimation and thus can play a role in helping individuals to cope with different environments.

The most commonly documented ecological effects of epigenetic diversity on populations involve the productivity or fitness of the population studied, and they can occur through different mechanisms. In 2010, Bossdorf and colleagues found that experimental alteration of DNA methylation strongly affected growth, fitness, and phenology in A. thaliana individuals. In another study, Latzel et al. (2013) suggested that epigenetic variability might affect how different epiRILs ("Epigenetic Recombinant Inbred Lines" Johannes et al., 2009) respond to treatment with salicylic acid and jasmonate, which are the main hormones involved in plant defense responses. In a series of experiments, Fieldes and colleagues showed that demethylation agents affected the fitness and phenological traits of Linum usitatissimum (Fieldes 1994; Fieldes & Amyot 1999; Fieldes et al. 2005). Moreover, evidence is growing that shortfalls in genetic diversity can be balanced by epigenetic diversity and facilitate plant population success. In 2013, Rollins and colleagues used two herbaceous species (Arctotheca populifolia and Petrorhagia nanteuilii) that showed evidence of fast morphological changes to examine if high genetic diversity is necessary for invasion success. Interestingly, they found that these two species had successfully established and adapted to a new habitat range in Australia, despite having considerably low genetic diversity in their native ranges (Rollins et al., 2013). Their findings provide an example of successful colonization processes that do not depend on high genetic diversity and align with similar findings (Richards et al., 2012) and theoretical lines of thought (Geoghegan & Spencer, 2012).

Although most studies on epigenetic effects involve short-lived plant species, methylation differences can also affect morphological and physiological processes in tree species. For example, a study by Raj et al. (2011) suggests that there might be a possible epigenetic basis for differences in transcriptomic profiles between poplar trees growing in distinct environments. In addition, white mangrove trees can exhibit striking morphological differences that might be related to a higher epigenetic diversity (Lira-Madeiros et al., 2010). Furthermore, changes in epigenetic marks have been associated with a wide variety of processes, including aging, organ maturation, and bud set or bud burst (Fraga et al. 2002; Santamaria et al. 2009; Valledor et al. 2010; Carneros et al., 2017; Lu et al., 2019). In recent years, studies on tree species responses that are epigenetically regulated have become more common (reviewed in Bräutigam et al., 2013). However, due to the limitations of working with long-lived species, natural epigenetic variation in tree species remains under-explored (Heer et al., 2018b).

B. Epigenetic effects at the community level.

We have discussed how epigenetic diversity can affect intraspecific diversity and how intraspecific diversity, in turn, changes how populations respond to their environmental conditions. Populations, although, are part of ecological communities. Although several studies have explored the importance of genetic diversity at the community level (Downing et al., 2002; Helm et al., 2009; Taberlet et al., 2012, Lamy et al., 2016), given the limitations of working at large-scale levels, quite a number of these studies disregard intraspecific variation and focus instead on trait means and trait variation among species (Bolnick et al., 2011; Zeng, Durka & Fischer, 2017). Nevertheless, community phenotypes can also arise from intraspecific trait variation (Siefert et al., 2015; Des Roches, 2018; Bongers et al., 2020), and sometimes intra-and interspecific diversity can have opposing effects (Hanh et al., 2017). Little is known about the ecological implications of epigenetic diversity on plant natural populations (Herrera, 2017), and very few studies have tried to assess the contribution of epigenetic variation to ecosystem dynamics in plant communities (Balao, Paun & Alonso, 2017; Herrera, Medrano & Bazaga, 2017; Mounger et al., 2020). However, considering that genetic variation has been proven to affect community-level interactions, it is safe to assume that epigenetic variation can have similar effects on communities. This section will address the three primary biological interactions (competition, predation and symbiosis) occurring within a community and the implications of epigenetic diversity on them.

i. Herbivory

Epigenetic responses to plant herbivory have been discussed in a previous chapter, so we will not detail them here. We will instead focus on how genetic diversity and, consequently epigenetic diversity, might affect interspecific interactions at the community level. Numerous studies have found effects of plant diversity on arthropod species richness and abundance (Koricheva et al. 2000, Crutsinger et al. 2006; Haddad et al., 2009; Robinson et al., 2012) and consumptive interactions (Moreira & Mooney 2013, Abdala-Roberts et al., 2015), with the basis of such effects being variation in ecologically critical functional traits among plant species or genotypes within species (Elle & Hare 2002, Mooney & Singer, 2012; Fernandez-Conradi et al., 2017). The influence of host-plant intraspecific diversity on arthropod communities has been extensively investigated and is a perfect example of how genetic variance can affect community-level interactions (reviewed in Koricheva et al., 2018). Similarly, epigenetic mechanisms can influence how plants respond to herbivory by, for example, inducing plant defenses (Verhoeven et al, 2010, affecting within-plant herbivore distribution (Herrera et al., 2019), or changing phytohormone production and palatability (Latzel et al., 2020). Hence, epigenetic changes can directly affect arthropod communities through different processes.

In this latter example, the authors conducted a so-called Cafeteria test with Larvae from the Egyptian cotton leafworm (Spodoptera littoralis). This moth is a generalist herbivore that feeds on plants of at least 40 families. Due to this extreme polyphagy, it has been used in many bioassay experiments of leaf palatability. In this experiment, the Egyptian cotton leafworm was put on Trifolium repens plants that had been treated with various levels of 5-azacytidine application to induce various levels of methylation, as well as on normally methylated control plants. The caterpillars could then freely choose between the plants and preferred individuals that were intensively treated over control or moderately treated plants (F = 7.22, p < 0.001). To get an indication of the mechanism behind the varying preferences, Latzel et al. measured levels of phytohormones known to be involved in inducible defense production, such as jasmonic acid, which is considered the key hormone in establishing altered gene expression related to inducible defense.

Most plant diversity studies have focused exclusively on its effects within a single trophic level (herbivores), but plant diversity may also indirectly affect higher trophic levels, i.e., enemies and mutualists of herbivores (Haddad et al., 2011; Moreira & Mooney, 2013). There are two ways in which plant diversity can influence higher trophic levels. The first one is related to an increase in the number of herbivores present in a specific community. There is a positive correlation between plant species richness and the diversity of associated consumers. Approximately 90% of herbivorous insects present some degree of host specialization (Bernays & Graham, 1988). If plant species richness increases, so do the number of herbivore species, and if the number of herbivores increases, so will the abundance of mutualists and enemies of those herbivores. The second way is mediated by traits. In this case, plant diversity influences herbivore, mutualist, or enemy traits, such as herbivore susceptibility to enemies (Johnson et al., 2006; Moreira et al., 2012; Moreira & Mooney, 2013). If epigenetic diversity can increase (or decrease) plant species richness and plant diversity, it will also affect arthropod communities.

Changes in aboveground net primary productivity (ANPP) can also influence plant-herbivore interactions. Higher ANPP means more energy available for consumers, and therefore more herbivore species or a higher number of individuals per species will be able to use this energy. Consequently, a higher number of herbivores will result in a higher number of enemies or mutualists. There is evidence that epigenetic diversity positively affects biomass production in Arabidopsis (Latzel et al., 2013). Although further research is needed to assess if this positive effect also extends to other plant species, epigenetic diversity has the potential of influencing herbivore populations by increasing plant biomass production and ANPP.

ii. Competition

Competition is a long-term interaction between organisms in which both organisms suffer adverse effects. Several studies have shown that plasticity in relevant functional traits derived from genetic variation can contribute to the success of invasive plant species by increasing their fitness across a range of habitats (Richards et al., 2006; Muth & Pigliucci, 2007; Walls, 2010; Davidson et al. 2011). Invasive species can have significant effects on ecosystem functioning, and invasions can lead to a loss of plant diversity (Linders et al., 2019). As discussed above, changes in plant diversity can extend to other trophic or organizational levels. Hence the invasive capacity of certain plant species can have effects on plant populations and communities. Interestingly, many invasive plant species appear to perform well despite having low levels of genetic variation. Dlugosch & Parker (2008a,b) reported that even though several plant species had suffered substantial losses of genetic diversity when compared to source populations, only one showed a significant decline in phenotypic variance. Several authors have suggested that epigenetic diversity increases the chances of successful colonization events in invasive species (Richards et al., 2006; Loomis & Fishman, 2009; Douhovnikoff & Dodd, 2015; Slotkin, 2016; Mounger et al., 2020). Even if most of the research in this field is focused on understanding plant invasion dynamics, epigenetic diversity could help explain how pioneer plant species colonize new habitats in successional processes. A recent study by Venturelli and colleagues (2015) showed that some allelochemicals released by Triticum aestivum (and other plant species) inhibit histone deacetylases, a group of enzymes involved in chromatin modification. Even more recently, Puy et al. (2020) showed that differences in DNA methylation of parental individuals affected offspring phenotypes. In their study, offspring of plants under stronger competition presented resource-conservative phenotypes and developed faster, suggesting that transgenerational phenotypic plasticity influenced competitive plant-plant interactions (Puy et al., 2020).

iii. Symbiosis

Symbiotic relationships are long-term interactions between two organisms. The term "symbiosis" includes a broad range of biological interactions that can be classified according to the effects the relationship has on each organism (Relman, 2008). The three major types of symbiotic interactions are mutualism, commensalism, and parasitism. Mutualistic interactions have a positive effect on both organisms, commensalism has a positive effect on one individual while the other is neither benefited nor harmed and in parasitism, the parasite benefits from the host and the host is damaged in some way.

Plants harbor an extreme diversity of symbionts, and their responses to symbiotic microbes and fungal organisms are probably the most studied interactions. Substantial literature documents the range of phenotypic variants conferred by symbiotic organisms, and many examples of mutualist-induced changes that are genotype-dependent in plant traits have been reported (Vannier et al., 2015; Gilbert, Tozer & Westoby, 2019; Wen et al., 2020). Several mechanisms control these interactions, but the plant epigenome has emerged as a critical modulator of pathogenic and symbiotic interactions (Bazin et al., 2012; Yu et al., 2013; Espinas, Saze & Saijo, 2016; Zogli & Libault, 2017). Interestingly, some symbiotic interactions could be considered epigenetic phenomena because plant endophytes can alter gene expression and phenotype without causing changes in the underlying DNA structure (Rodriguez et al., 2008). A major part of the research on epigenetic control of plant-symbiont interactions focuses on mycorrhizal fungi due to the importance of these organisms for processes that can affect plant fitness (e.g., nitrogen-fixing reactions, nutrient uptake, and abiotic and biotic stress tolerance). Nevertheless, some studies have tried to determine the role of epigenetic changes in plant-pollinator interactions, another common type of mutualism, and plant-parasite interactions. For example, in 2009, Marfil and colleagues associated distinctive methylation profiles with aberrant flower phenotypes in Solanum ruiz-lealii plants. This can potentially influence pollination rates, as bumblebees, the only pollinator of Solanum plants, do not visit aberrant flowers (Marfil, Camado & Masueli, 2009). In another study by Kellenberger et al. (2016), demethylation of Brassica rapa plants resulted in reduced attractiveness of the plants to pollinator bees (Kellenberger, Schlüter and Schiestle, 2016). Quite recently, Samarth et al. (2020) have introduced the hypothesis of an "epigenetic summer memory" as a driver of mast flowering (mass synchronized flowering of perennial plants over a wide geographical area), an event that has major impacts on trophic interactions (Kelly, 1994).

Parasitic interactions have received a bit more attention, mainly due to the high economic losses plant parasites cause on crops, and recent studies have attempted to throw some light on the epigenetic regulation of plant-nematode interactions. For example, Hewezi et al. showed that cyst nematodes could induce changes in the root epigenome (Hewezi et al., 2017). In another study, Sahid et al. (2018) suggest that miRNA-mediated changes in the host gene expression might act in a way beneficial to the nematode, and Pratx and colleagues explored the role of the epigenetic machinery of the root-knot nematode Meloidogyne incognita (Pratx et al., 2018). However, despite the advances of the modern literature, to our knowledge, all the research so far has focused on understanding the epigenetic regulation of symbiotic interactions, and there are no studies on the potential effects of epigenetic diversity on community structure or dynamics.

C. Epigenetic effects at the ecosystem and landscape level

We have discussed how variation at the species level can extend into higher organizational levels. When communities are grouped with their abiotic surroundings, they are considered an ecosystem. Genetic diversity and epigenetic diversity shape the structure and dynamics of each level at the population and community levels. In contrast, landscape structure can have significant effects on the genetic diversity of populations (Münzbergová et al., 2013; González et al., 2020; Lehmair et al., 2020). The study of large-scale interactions is limited by the complexity of each system and the sheer number of different variables and processes encompassed in the system. The further up the organizational level we move, the more difficult it becomes to study possible epigenetic effects. From a traditional perspective, studies on genetic diversity at large-scale levels (i.e., over large areas and for many species) are still demanding, given the need for field sampling and the still more or less high costs of genetic analysis. (Taberlet et al., 2012).

Furthermore, plant functional traits can strongly influence ecosystem properties, acting in several contexts that include the effects of dominant or foundation species and keystone species or the interactions between the individuals of the ecosystem (Hooper et al., 2005). Regardless, several studies have shown that the genetic diversity of dominant plant species can affect essential ecosystem functions. Genotypic variation in trees of the Populus genus is a well-studied example of how genetic diversity can influence nutrient cycles and energy fluxes. Differences among several aspen (Populus tremula) genotypes can have substantial effects on litter decomposition and nutrient release (Madritch et al. 2007; Bandau et al., 2016; Hughes et al., 2018). Nutrient fluxes can also be affected by foliar secondary metabolites. For example, condensed tannins (CT) influence carbon and nitrogen cycles specifically by affecting the microbial communities that mediate these cycles. In other hybridizing cottonwoods (Populus fremontii, Populus angustifolia), the chemical composition of the leaf litter impacts the rate of decay and nutrient flux to a degree that is comparable with the effects of species diversity (Schweitzer et al. 2005). Since nutrient fluxes are ecosystem-wide processes, any change in these fluxes will affect every population or community within the ecosystem (Fischer et al., 2007).

Moreover, disparate groups of organisms demonstrate significant relationships between community composition and concentrations of CT in Populus that link above- and below-ground processes. For example, Schweitzer et al. (2008) showed that microbial community composition differs significantly in soils beneath Populus genotypes that varied in their expression of foliar CT. Thus, even when only a few traits vary between genotypes, microorganisms belonging to different functional groups can occur. In another study, Madritch and others (2007) found that the CT concentration of frass deposition of two herbivores affected below-ground respiration and extracellular enzyme activity of microbial communities (Schweitzer et al., 2008).

When including epigenetic diversity into large-scale studies, several layers of complexity are added to the mix. In contrast to genetic or genomic patterns, the strength, the effects of epigenetic diversity at a landscape level, and its evolutionary implications are poorly understood. Generally speaking, disentangling the effects of epigenetic variation from genetic variation is not a straightforward process. Some advances have been made in this regard by using clonally propagated species and common garden experiments (Bossdorf et al., 2008; Richards et al., 2017). Despite the limitations, many recent landscape-level studies have investigated the role of epigenetics in intraspecific trait variation and adaptation (Medrano et al., 2014; Dubin et al., 2015; Preite et al., 2015; Foust et al., 2016; Gugger et al., 2016; Herrera et al., 2016; Keller et al., 2016; Alakärpa et al., 2018; Gáspár, Bossdorf & Durka, 2018). These studies focus on the relationship between genetic and epigenetic variation at the landscape level, correlations between environmental variables and epigenetic marks, and correlations between epigenetic marks and plant phenotypic traits (Whipple & Holeski, 2016).

Alonso, Conchita, Ricardo Pérez, Pilar Bazaga, and Carlos M. Herrera. 2015. “Global DNA Cytosine Methylation as an Evolving Trait: Phylogenetic Signal and Correlated Evolution with Genome Size in Angiosperms.” Frontiers in Genetics 6.

Audergon, Pauline N. C. B., Sandra Catania, Alexander Kagansky, Pin Tong, Manu Shukla, Alison L. Pidoux, and Robin C. Allshire. 2015. “Restricted Epigenetic Inheritance of H3K9 Methylation.” Science 348 (6230): 132–35.

Becker, Claude, Jörg Hagmann, Jonas Müller, Daniel Koenig, Oliver Stegle, Karsten Borgwardt, and Detlef Weigel. 2011. “Spontaneous Epigenetic Variation in the Arabidopsis Thaliana Methylome.” Nature 480 (7376): 245–49.

Bossdorf, Oliver, Christina Richards, and Massimo Pigliucci. 2008. “Epigenetics for Ecologists.” Ecology Letters 11 (2): 106–15.

Cambareri, E. B., B. C. Jensen, E. Schabtach, and E. U. Selker. 1989. “Repeat-Induced G-C to A-T Mutations in Neurospora.” Science 244 (4912): 1571–75.

Castonguay, Emilie, and Bernard Angers. 2012. “The Key Role of Epigenetics in the Persistence of Asexual Lineages.” Research article. Genetics Research International. 2012.

Cortijo, Sandra, René Wardenaar, Maria Colomé-Tatché, Arthur Gilly, Mathilde Etcheverry, Karine Labadie, Erwann Caillieux, et al. 2014. “Mapping the Epigenetic Basis of Complex Traits.” Science 343 (6175): 1145–48.

Cubas, Pilar, Coral Vincent, and Enrico Coen. 1999. “An Epigenetic Mutation Responsible for Natural Variation in Floral Symmetry.” Nature 401 (6749): 157–61.

Darwin, Charles. 1859. “On the Origin of Species, 1859.” Taylor & Francis. 1859. https://www.taylorfrancis.com/books/9781134439928.

Dickins, T. E., and R. A. Barton. 2013. “Reciprocal Causation and the Proximate–Ultimate Distinction.” Biology & Philosophy 28 (5): 747–56.

Dubin, Manu J, Pei Zhang, Dazhe Meng, Marie-Stanislas Remigereau, Edward J Osborne, Francesco Paolo Casale, Philipp Drewe, et al. 2015. “DNA Methylation in Arabidopsis Has a Genetic Basis and Shows Evidence of Local Adaptation.” ELife 4.

Fisher, R. A. 1919. The Correlation between Relatives on the Supposition of Mendelian Inheritance. Earth and Environmental Science Transactions of the Royal Society of Edinburgh. 1919; 52 (02): 399–433.

Flood, Pádraic J., Joost van Heerwaarden, Frank Becker, C. Bastiaan de Snoo, Jeremy Harbinson, and Mark G. M. Aarts. 2016. “Whole-Genome Hitchhiking on an Organelle Mutation.” Current Biology 26 (10): 1306–11.

Frankham, Richard. 1996. “Introduction to Quantitative Genetics (4th Edn): By Douglas S. Falconer and Trudy F.C. Mackay Longman, 1996. £24.99 Pbk (Xv and 464 Pages) ISBN 0582 24302 5.” Trends in Genetics 12 (7): 280.

Furci, Leonardo, Ritushree Jain, Joost Stassen, Oliver Berkowitz, James Whelan, David Roquis, Victoire Baillet, Vincent Colot, Frank Johannes, and Jurriaan Ton. 2019. “Identification and Characterisation of Hypomethylated DNA Loci Controlling Quantitative Resistance in Arabidopsis.” Edited by Daniel J Kliebenstein and Christian S Hardtke. ELife 8 (January): e40655.

Futuyma, Douglas J. 1998. “Evolutionary Biology. Sunderland.” Mass.: Sinauer Associates. 2017. “Evolutionary Biology Today and the Call for an Extended Synthesis.” Interface Focus 7 (5): 20160145.

Gao, Dongying, Jinfeng Chen, Mingsheng Chen, Blake C. Meyers, and Scott Jackson. 2012. “A Highly Conserved, Small LTR Retrotransposon That Preferentially Targets Genes in Grass Genomes.” PLOS ONE 7 (2): e32010.

Gáspár, Bence, Oliver Bossdorf, and Walter Durka. 2018. “Structure, Stability and Ecological Significance of Natural Epigenetic Variation: A Large-Scale Survey in Plantago Lanceolata.” New Phytologist 0 (0).

Gugger, Paul F., Sorel Fitz‐Gibbon, Matteo Pellegrini, and Victoria L. Sork. 2016. “Species-Wide Patterns of DNA Methylation Variation in Quercus Lobata and Their Association with Climate Gradients.” Molecular Ecology 25 (8): 1665–80.

Haig, David. 2007. “Weismann Rules! OK? Epigenetics and the Lamarckian Temptation.” Biology & Philosophy 22 (3): 415–28.

Heer, Katrin, Kristian K. Ullrich, Manuel Hiss, Sascha Liepelt, Ralf Schulze Brüning, Jiabin Zhou, Lars Opgenoorth, and Stefan A. Rensing. 2018. “Detection of Somatic Epigenetic Variation in Norway Spruce via Targeted Bisulfite Sequencing.” Ecology and Evolution 8 (19): 9672–82.

Herrera, Carlos M., and Pilar Bazaga. 2013. “Epigenetic Correlates of Plant Phenotypic Plasticity: DNA Methylation Differs between Prickly and Nonprickly Leaves in Heterophyllous Ilex Aquifolium (Aquifoliaceae) Trees.” Botanical Journal of the Linnean Society 171 (3): 441–52.

Huxley, Julian Sorell. 1943. “Evolution, the Modern Synthesis.” London Georg Allen Unwin.

Ingouff, Mathieu, Svenja Rademacher, Sarah Holec, Lucija Šoljić, Nie Xin, Anne Readshaw, Shi Hui Foo, Benoît Lahouze, Stefanie Sprunck, and Frédéric Berger. 2010. “Zygotic Resetting of the HISTONE 3 Variant Repertoire Participates in Epigenetic Reprogramming in Arabidopsis.” Current Biology 20 (23): 2137–43.

Johannes, Frank, Emmanuelle Porcher, Felipe K. Teixeira, Vera Saliba-Colombani, Matthieu Simon, Nicolas Agier, Agnès Bulski, et al. 2009. “Assessing the Impact of Transgenerational Epigenetic Variation on Complex Traits.” PLOS Genetics 5 (6): e1000530.

Johannes, Frank, and Robert J. Schmitz. 2018. “Spontaneous Epimutations in Plants.” New Phytologist, September.

Kalisz, Susan, and Michael D. Purugganan. 2004. “Epialleles via DNA Methylation: Consequences for Plant Evolution.” Trends in Ecology & Evolution 19 (6): 309–14.

Kawakatsu, Taiji, Shao-shan Carol Huang, Florian Jupe, Eriko Sasaki, Robert J. Schmitz, Mark A. Urich, Rosa Castanon, et al. 2016. “Epigenomic Diversity in a Global Collection of Arabidopsis Thaliana Accessions.” Cell 166 (2): 492–505.

Kinoshita, Tetsu, and Motoaki Seki. 2014. “Epigenetic Memory for Stress Response and Adaptation in Plants.” Plant and Cell Physiology 55 (11): 1859–63.

Koo, Dal-Hoe, William T. Molin, Christopher A. Saski, Jiming Jiang, Karthik Putta, Mithila Jugulam, Bernd Friebe, and Bikram S. Gill. 2018. “Extrachromosomal Circular DNA-Based Amplification and Transmission of Herbicide Resistance in Crop Weed Amaranthus Palmeri.” Proceedings of the National Academy of Sciences 115 (13): 3332–37.

Kooke, Rik, Frank Johannes, René Wardenaar, Frank Becker, Mathilde Etcheverry, Vincent Colot, Dick Vreugdenhil, and Joost J. B. Keurentjes. 2015. “Epigenetic Basis of Morphological Variation and Phenotypic Plasticity in Arabidopsis Thaliana.” The Plant Cell, February, tpc.114.133025.

Laland, Kevin N., Tobias Uller, Marcus W. Feldman, Kim Sterelny, Gerd B. Müller, Armin Moczek, Eva Jablonka, and John Odling-Smee. 2015. “The Extended Evolutionary Synthesis: Its Structure, Assumptions and Predictions.” Proc. R. Soc. B 282 (1813): 20151019.

Lämke, Jörn, and Isabel Bäurle. 2017. “Epigenetic and Chromatin-Based Mechanisms in Environmental Stress Adaptation and Stress Memory in Plants.” Genome Biology 18 (1): 124.

Law, Julie A., and Steven E. Jacobsen. 2010. “Establishing, Maintaining and Modifying DNA Methylation Patterns in Plants and Animals.” Nature Reviews Genetics 11 (3): 204–20.

Lira-Medeiros, Catarina Fonseca, Christian Parisod, Ricardo Avancini Fernandes, Camila Souza Mata, Monica Aires Cardoso, and Paulo Cavalcanti Gomes Ferreira. 2010. “Epigenetic Variation in Mangrove Plants Occurring in Contrasting Natural Environment.” PLOS ONE 5 (4): e10326.

Manning, Kenneth, Mahmut Tör, Mervin Poole, Yiguo Hong, Andrew J. Thompson, Graham J. King, James J. Giovannoni, and Graham B. Seymour. 2006. “A Naturally Occurring Epigenetic Mutation in a Gene Encoding an SBP-Box Transcription Factor Inhibits Tomato Fruit Ripening.” Nature Genetics 38 (8): 948–52.

Martin, Antoine, Christelle Troadec, Adnane Boualem, Mazen Rajab, Ronan Fernandez, Halima Morin, Michel Pitrat, Catherine Dogimont, and Abdelhafid Bendahmane. 2009. “A Transposon-Induced Epigenetic Change Leads to Sex Determination in Melon.” Nature 461 (7267): 1135–38.

Masel J (October 2011). "Genetic drift". Current Biology. Cell Press. 21 (20): R837–8.

Mendel, Gregor. 1865. “Versuche Über Pflanzen-Hybriden. Verhandlungen Des Naturforschenden Vereines in Brünn.,” no. 4: 3–47.

Mesoudi, Alex, Simon Blanchet, Anne Charmantier, Étienne Danchin, Laurel Fogarty, Eva Jablonka, Kevin N. Laland, et al. 2013. “Is Non-Genetic Inheritance Just a Proximate Mechanism? A Corroboration of the Extended Evolutionary Synthesis.” Biological Theory 7 (3): 189–95.

Mette, M.f., W. Aufsatz, J. van der Winden, M.a. Matzke, and A.j.m. Matzke. 2000. “Transcriptional Silencing and Promoter Methylation Triggered by Double-Stranded RNA.” The EMBO Journal 19 (19): 5194–5201.

Mittelsten Scheid, Ortrun, Karin Afsar, and Jerzy Paszkowski. 2003. “Formation of Stable Epialleles and Their Paramutation-like Interaction in Tetraploid Arabidopsis Thaliana.” Nature Genetics 34 (4): 450–54.

Müller, Gerd B. 2007. “Evo–Devo: Extending the Evolutionary Synthesis.” Nature Reviews Genetics 8 (12): 943–49.

Münzbergová, Zuzana, Vít Latzel, Maria Šurinová, and Věroslava Hadincová. 2019. “DNA Methylation as a Possible Mechanism Affecting Ability of Natural Populations to Adapt to Changing Climate.” Oikos 128 (1): 124–34.

Noshay, Jaclyn M., Sarah N. Anderson, Peng Zhou, Lexiang Ji, William Ricci, Zefu Lu, Michelle C. Stitzer, et al. 2019. “Monitoring the Interplay between Transposable Element Families and DNA Methylation in Maize.” PLOS Genetics 15 (9): e1008291.

Ong-Abdullah, Meilina, Jared M. Ordway, Nan Jiang, Siew-Eng Ooi, Sau-Yee Kok, Norashikin Sarpan, Nuraziyan Azimi, et al. 2015. “Loss of Karma Transposon Methylation Underlies the Mantled Somaclonal Variant of Oil Palm.” Nature 525 (7570): 533–37.

Pecinka, Ales, and Ortrun Mittelsten Scheid. 2012. “Stress-Induced Chromatin Changes: A Critical View on Their Heritability.” Plant and Cell Physiology 53 (5): 801–8.

Pfeifer, G. P. 2006. “Mutagenesis at Methylated CpG Sequences.” In DNA Methylation: Basic Mechanisms, edited by Walter Doerfler and Petra Böhm, 259–81. Current Topics in Microbiology and Immunology. Berlin, Heidelberg: Springer Berlin Heidelberg.

Pigliucci, Massimo. 2009. “An Extended Synthesis for Evolutionary Biology.” Annals of the New York Academy of Sciences 1168 (1): 218–28.

Pigliucci, Massimo, and Leonard Finkelman. 2014. “The Extended (Evolutionary) Synthesis Debate: Where Science Meets Philosophy.” BioScience 64 (6): 511–16.

Provine, W. B. 1978. “The Role of Mathematical Population Geneticists in the Evolutionary Synthesis of the 1930s and 1940s.” Studies in History of Biology 2: 167–92.

Rangwala, Sanjida H., Rangasamy Elumalai, Cheryl Vanier, Hakan Ozkan, David W. Galbraith, and Eric J. Richards. 2006. “Meiotically Stable Natural Epialleles of Sadhu, a Novel Arabidopsis Retroposon.” PLOS Genetics 2 (3): e36.

Richards, Christina L., Conchita Alonso, Claude Becker, Oliver Bossdorf, Etienne Bucher, Maria Colomé-Tatché, Walter Durka, et al. 2017. “Ecological Plant Epigenetics: Evidence from Model and Non-Model Species, and the Way Forward.” Ecology Letters 20 (12): 1576–90.

Richards, Eric J. 2006. “Inherited Epigenetic Variation — Revisiting Soft Inheritance.” Nature Reviews Genetics 7 (5): 395–401.

Riddle, Nicole C., and Eric J. Richards. 2002. “The Control of Natural Variation in Cytosine Methylation in Arabidopsis.” Genetics 162 (1): 355–63.

Secco, David, Chuang Wang, Huixia Shou, Matthew D Schultz, Serge Chiarenza, Laurent Nussaume, Joseph R Ecker, James Whelan, and Ryan Lister. 2015. “Stress Induced Gene Expression Drives Transient DNA Methylation Changes at Adjacent Repetitive Elements.” ELife 4.

Shindo, Chikako, Maria Jose Aranzana, Clare Lister, Catherine Baxter, Colin Nicholls, Magnus Nordborg, and Caroline Dean. 2005. “Role of FRIGIDA and FLOWERING LOCUS C in Determining Variation in Flowering Time of Arabidopsis.” Plant Physiology 138 (2): 1163–73.

Shindo, Chikako, Clare Lister, Pedro Crevillen, Magnus Nordborg, and Caroline Dean. 2006. “Variation in the Epigenetic Silencing of FLC Contributes to Natural Variation in Arabidopsis Vernalization Response.” Genes & Development 20 (22): 3079–83.

Smocovitis, Vassiliki Betty. 1997. “G. Ledyard Stebbins, Jr. and the Evolutionary Synthesis (1924–1950).” American Journal of Botany 84 (12): 1625–37.

Stebbins, C. L. 1950. “Variation and Evolution in Plants.” Variation and Evolution in Plants. https://www.cabdirect.org/cabdirect/abstract/19510302133.

Vanden Broeck, An, Karen Cox, Rein Brys, Stefano Castiglione, Angela Cicatelli, Francesco Guarino, Berthold Heinze, Marijke Steenackers, and Kristine Vander Mijnsbrugge. 2018. “Variability in DNA Methylation and Generational Plasticity in the Lombardy Poplar, a Single Genotype Worldwide Distributed Since the Eighteenth Century.” Frontiers in Plant Science 9.

Vaughn, Matthew W., Miloš Tanurdžić, Zachary Lippman, Hongmei Jiang, Robert Carrasquillo, Pablo D. Rabinowicz, Neilay Dedhia, et al. 2007. “Epigenetic Natural Variation in Arabidopsis Thaliana.” PLOS Biology 5 (7): e174.

Wallace, Alfred Russel. 1858. Contributions to the Theory of Natural Selection: A Series of Essays. 2nd ed. Cambridge: Cambridge University Press.

Whitelaw, Nadia C., and Emma Whitelaw. 2006. “How Lifetimes Shape Epigenotype within and across Generations.” Human Molecular Genetics 15 (suppl_2): R131–37.

Wilkins, Adam S. 2008. “Waddington’s Unfinished Critique of Neo-Darwinian Genetics: Then and Now.” Biological Theory 3 (3): 224–32.

Zhang, Jianmin, Zhen Yang, Pei Feng, Xiao Zhong, Qiqi Ma, Qi Su, Xiangping Wang, Chuanren Li, and Yazhen Yang. 2019. “Identification and the Potential Roles of Long Non-Coding RNAs in Cotton Leaves Damaged by Aphis Gossypii.” Plant Growth Regulation 88 (3): 215–25.

Zicola, Johan, Liangyu Liu, Petra Tänzler, and Franziska Turck. 2019. “Targeted DNA Methylation Represses Two Enhancers of FLOWERING LOCUS T in Arabidopsis Thaliana.” Nature Plants 5 (3): 300–307.

Díaz, S. M., Settele, J., Brondízio, E., Ngo, H., Guèze, M., Agard, J., ... & Zayas, C. (2019). The global assessment report on biodiversity and ecosystem services: Summary for policy makers.

Frankham, R. (1995). Conservation genetics. Annual review of genetics, 29(1), 305-327.

Hampe, A., & Petit, R. J. (2005). Conserving biodiversity under climate change: the rear edge matters. Ecology letters, 8(5), 461-467.

Hazarika, R. R., Serra, M., Zhang, Z., Zhang, Y., Schmitz, R. J., & Johannes, F. (2022). Molecular properties of epimutation hotspots. Nature Plants, 1-11.

Herrera, C. M., Medrano, M., & Bazaga, P. (2017). Comparative epigenetic and genetic spatial structure of the perennial herb Helleborus foetidus: Isolation by environment, isolation by distance, and functional trait divergence. American Journal of Botany, 104(8), 1195-1204.

Holderegger, R., Balkenhol, N., Bolliger, J., Engler, J. O., Gugerli, F., Hochkirch, A., ... & Zachos, F. E. (2019). Conservation genetics: Linking science with practice.

Johannes, F., & Schmitz, R. J. (2019). Spontaneous epimutations in plants. New Phytologist, 221(3), 1253-1259.

Menges, E. S. (1991). The application of minimum viable population theory to plants. Genetics and conservation of rare plants, 45, 158-164.

Richards, C. L., Alonso, C., Becker, C., Bossdorf, O., Bucher, E., Colomé‐Tatché, M., ... & Verhoeven, K. J. (2017). Ecological plant epigenetics: Evidence from model and non‐model species, and the way forward. Ecology letters, 20(12), 1576-1590.

Soulé, M. E. (1985). What is conservation biology?. BioScience, 35(11), 727-734.

Yao, N., Schmitz, R. J., & Johannes, F. (2021). Epimutations define a fast-ticking molecular clock in plants. Trends in Genetics, 37(8), 699-710.

Xie, H., Konate, M., Sai, N., Tesfamicael, K. G., Cavagnaro, T., Gilliham, M., ... & Lopez, C. M. (2017). Global DNA methylation patterns can play a role in defining terroir in grapevine (Vitis vinifera cv. Shiraz). Frontiers in plant science, 8, 1860.

As genetic diversity in conservation genetics, epigenetic diversity will be at the core of any conservation epigenetic approach. If it will prove to be relevant in various taxa, functional epigenetic diversity would be bound to environmental priming. Consequently, one essential principle to sustain epigenetic functional diversity would be to focus on environmental heterogeneity in conservation strategies. This principle is scalable, meaning that it would refer on the one side to range-wide heterogeneity e.g. strengthening rear edges conservation analogous to the genetic diversity (Hampe and Petit 2005). But it would also refer to microenvironmental heterogeneity. In other words, conservation epigenetics should aim to diversify habitat heterogeneity and welcome stress as well as disturbances as important factors to strengthen or sustain the acclimation potential of a population or species. This principle could also be reflected in ex-situ and breeding strategies. Especially for the latter, it would mean a clear deviation from current management strategies. For example, tree seeds are normally harvested in orchards or from so called plus-trees. The former normally are placed under ideal habitat situations in productive sites. In both cases seed donors normally have favorable phenotypes. If epigenetic diversity is the ultimate goal, it would mean that seeds should also come from edge populations, stressed populations, and not-perfect phenotypes, in other words selecting seeds to increase plasticity.

The human-induced 6th mass extinction threatens more species with global extinction than ever before with an average of around 25 % of species in assessed animal and plant groups being threatened. That means, that up to one million species already face extinction, many of which in the next decades, unless action is taken to reduce the intensity of drivers of biodiversity loss (Díaz et al. 2019). Conservation biology has been the research field that deals with this biodiversity crisis. It studies the conservation of biodiversity by investigating the biology of species, communities, and ecosystems that are directly or indirectly threatened by human activities or other agents. Its goal is to provide principles and tools for preserving all levels of biodiversity including its evolutionary potential (Soulé 1985). Conservation genetics in turn, is a well-established scientific field within conservation biology and is dedicated to shedding light on the evolutionary dimension of conservation biology. Large numbers of scientific studies are published each year in dedicated scientific journals as well as journals targeting broader scientific audiences (Holderegger et al. 2019). Conservation genetics has impacted conservation on two levels. First, it established mechanistic principles that have acted as conservation guidelines already for decades – from the core principle that genetic diversity is crucial for the well-being of populations to central concepts such as inbreeding depression, accumulation and loss of deleterious mutations, loss of genetic variation in small populations, genetic adaptation to captivity and its effect on reintroduction success, outbreeding depression (Frankham 1996), minimum viable populations (Menges 1991), and the distinctness of rear edge populations in conservation (Hampe and Petit 2005). Besides such textbook principles, the availability of conservation genetic tools directly empowers conservation managers on the ground including barcoding and metabarcoding for species identification or monitoring or with genetic studies on gene flow to assess fragmentation and connectivity to assess the success of connectivity measures (Holderegger et al. 2019). Still, Holderegger and co-workers attest that conservation genetics still remains a largely academic field and that real‐world examples with a clear focus on application would largely be restricted to emblematic vertebrate fauna. They observe, that this gap between conservation genetic sciences and the practical conservation yet increases through the rapid development of high‐throughput genotyping technologies and thus the use of genomic information as new challenges emerge with regards to data analyses. Undeniably, despite these initial gaps between science and management, the genomic revolution will provide valuable information for conservation genetic approaches, as it will allow to focus on a species’ or population’s adaptive potential to respond to stress by means of molecular changes (Eizaguirre & Baltazar-Soares, 2014).

In a recent review, Rey et al. propose to establish yet another research field that will contribute to conservation biology, namely conservation epigenetics. They argue that epigenetic marks – more particularly DNA methylation – and developmental reprogramming should be considered as an additional conservation level stating that DNA methylation is sensitive to the environment and is involved in organisms' plastic and adaptive responses to changing environments. As we have seen in previous chapters, epigenetic elements can act in conjunction with genetic information to modulate phenotypes during development (Allis & Jenuwein, 2016). Moreover, while some epigenetic patterns are under genetic determinism, some others are directly modulated by the surrounding environmental conditions (Feil & Fraga, 2012), particularly that of DNA methylation. Thus, DNA methylation induced by environmental stressors during development that produces maladaptive phenotypes can have negative consequences in populations (Piferrer, 2016). Rey et al. argue, that accounting for such epigenetic trap effect faced by some populations could be useful in a conservation context. Specifically, at the population level, modifications of DNA methylation patterns among individuals in response to changing environment could be associated with a phenotypic shift from suboptimal to optimal value in the resulting environment, hence leading to adaptive phenotypic plasticity corresponding to the environmentally induced phenotype variation (EPV) (Rey et al. 2019 and references therein). Alternatively, environmental changes could potentially induce spontaneous and random modifications in DNA methylation patterns potentially resulting in the broadening of phenotypic values around the original mean phenotype within populations corresponding to the stochastic developmental phenotype variation (SPV) (Rey et al. .2019 and references therein). They further argue, that those two processes can lead to phenotypic diversification, with EPV being expected to be selected when environmental changes are predictable allowing organisms to quickly respond and adjust their phenotypes to maximize their fitness while SPV could be considered a random and non-directional flexibilization of genome expression to new and/or unpredictable environments constituting a bet-hedging strategy resulting in the maintenance of few individuals harboring optimal phenotypes and most individuals expressing suboptimal phenotypes in the new environment (Rey et al. 2019 and references therein).

However, as we have seen throughout this book, the ecological importance of variation in DNA methylation relative to genetic variation has only been established in a few individual case studies and still needs to be empirically quantified in non-model plant species (Richards et al. 2017). That said, the growing body of literature shows that the distribution and especially function of DNA methylation variation varies strongly among taxa and its role in acclimation is not as straight forward as seen for genetic variation in adaptation. Therefore, it will likely take at least another decade to establish sound evidence for a large variety of conservation relevant taxa for conservation epigenetics to become effective. And given that sound epigenetic research is necessarily based on omics resources, a conservation epigenetic toolbox will likely not be available for conservation practitioners in the near future. As with the beginnings of conservation genetics decades ago, our current understanding of plant epigenetics might still be sufficient to establish basic principles for conservation management.

Imprinting is a preferential expression pattern of genes according to their maternal and paternal allele origin. When the preferential expression comes from the mother, the genes are called maternally expressed genes (MEG), when from the father, paternally expressed genes (PEG) (Batista & Köhler, 2020). Epigenetic modifications in DNA methylation, histone modifications, or chromatin composition might be directly favoring the activity of one allele over another (Dong et al., 2018; P. R. V. Satyaki & Gehring, 2017). This epigenetic phenomenon is exclusive for flowering plants, suggesting an independent evolution among plant species of different periods in time (Batista & Köhler, 2020).

Imprinting is detected mostly in the endosperm, an analog of the placenta of mammals. Even though the endosperm surrounds the embryo and supplies nutrients to it from the maternal parent, little is known of imprinted genes in the embryo (Fig. 8) (Batista & Köhler, 2020; Law & Jacobsen, 2010; P. R. V. Satyaki & Gehring, 2017). What we know is that a small number of genes are differentially methylated and silenced in male and female tissues. This is regulated by de novo methylation, maintenance methylation, and demethylation, with demethylation dominating the process (Batista & Köhler, 2020; P. R. V. Satyaki & Gehring, 2017). This is suggested by DME activity and the presence of DML2-3 and ROS1 in the central cell and the vegetative nucleus of the male and female gametophyte in Arabidopsis and rice (Batista & Köhler, 2020). The reason for this massive active demethylation in vegetative gametophyte tissue may be the protection through hypermethylation of the DNA in the haploid egg and sperm cells. In the germline, active transposons could produce much damage. The active demethylation of TEs leads to their transcription and the production of small interfering RNAs (siRNAs) in the tissue surrounding the egg and sperm cells. From their, siRNAs are thought to be transported into the egg and sperm cells, leading to hypermethylation of their homolog sequences throughout the RdDM pathway and thereby effectively hindering the activation of TEs in the germline (Fig. 8B) (Batista & Köhler, 2020; Law & Jacobsen, 2010). However, how the siRNAs are exported to adjacent compartments is yet unknown (Law & Jacobsen, 2010).

So, hypermethylation of the embryo DNA is most likely caused by the demethylation of the surrounding vegetative tissue. However, this is not yet imprinting because it does not yet include a preference of the maternal of the parental allele. What is needed here, ist that not only the egg but also the endosperm is fertilized, which is the case in flowering plants (Fig. 8). Now it is principally possible that only the male or the female allele is transcribed in the endosperm, leading to allele-specific gene expression in the endosperm. This is realized, for example, via the accumulation of the histone H3K27me3 on the maternal allele of the MADS-box transcription factor PHERES 1, after demethylation through DME in the central cell of the endosperm. The maternal allele is silenced through this accumulation. In the parental allele, a 3’ sequence is methylated, which is thought to prevent H3K27me3 accumulation. Consequently, only the parental allele of PHERES 1 can be expressed in the endosperm (Batista & Köhler, 2020).

In many plant species such as Arabidopsis thaliana, maize, rice, and sorghum between the 40 to 50% of maternal expressed imprinted genes (MEGs) and 60% of parental expressed imprinted genes (PEGs) are associated with epigenetic marks in gene bodies and flanking regions in the endosperm (Batista & Köhler, 2020; Satyaki & Gehring, 2017). There is little information about well-identified imprinted genes and their regulation mechanism in Plants. One of the most studiest genes is the FLOWERING WAGENINGEN (FWA) gene in Arabidopsis, which encodes a transcription factor related to delayed flowering. FWA is tissue-specific activated by DNA demethylation in the female gamete and endosperm (Fujimoto et al., 2008; Meyer, 2011). The FWA gene is rich in tandem repeats and a SINE-related sequence which are direct targets for the methylation machinery, and it is sufficient for imprinting and vegetative silencing (Fujimoto et al., 2008; Meyer, 2011).

To summarize, a combination of epigenetic mechanisms is responsible for parent-of-origin expressed genes. However, it seems so far that plant species contain differ in their regulating systems hampering the general understanding of this phenomenon. Therefore, the development of new molecular tools probably is needed.

DNA methylation through the years has become an evolutionary preserved mechanism that has contributed to the divergence of prokaryotes and eukaryotes, helping them adapt to different environmental conditions over the years (Bewick & Schmitz, 2017). Epigenetic modifications, including DNA methylation, histone modifications, and the expression of non-coding RNAs (ncRNA), influence the chromatin structure and alter the accessibility of genomic regions for interacting enzymes (H. Zhang, Lang, & Zhu, 2018). This means that DNA methylation is essential to many biological processes directly modifying the genome architecture, the definition of Euchromatin and Heterochromatin, and the control of gene expression (Bewick & Schmitz, 2017; H. Zhang et al., 2018). Dysfunctions of DNA methylation can lead to abnormalities in plants, such as failure in tomato and orange fruit ripening or vice versa promote early strawberry fruit ripening (Cheng et al., 2018; Huang et al., 2019; H. Zhang et al., 2018).

The genomic location of DNA methylation can display different roles in the maintenance of the structure and integrity of the genome and gene expression regulation (Bewick & Schmitz, 2017). In combination with histone modifications and other interacting proteins, DNA methylation defines the chromatin structure and, through this, the accessibility of the DNA. DNA methylation helps to regulate gene expression, transposon silencing, chromosome interactions (Fig. 1) (Zhang et al., 2018). To Illustrate, in plants, DNA methylation is distributed at the body of genes (where the function is often unclear) and at repetitive regions, where it restricts the expression of TEs, which represent in some plant species more than 80% of the genome, e.g., barley, sunflower, and maize. (Meyer, 2011; Vitte, Fustier, Alix, & Tenaillon, 2014). The function of the variability of DNA methylation in some genomic regions in some plant species remains a mystery (Zilberman, 2017). For example, five different apple cultivars showed differential methylation patterns in promoter regions of genes that regulate the anthocyanin pathway. Even though changes in the transcription of these genes generate different red-skin phenotypes, methylation was not the main factor to alter their expression (Jiang et al., 2019).

In plants, methylation can occur in distinct site classes based on the sequence context with which the methylated cytosine (mC) is accompanied. There are three sequence contexts, including the dinucleotide CpG or CG sites (mCG, the p marks the phosphate), and the trinucleotides CHG and CHH (mCHG and mCHH), where H can be adenine (A), cytosine (C), or thiamine (T) (Bewick & Schmitz, 2017). This contrasts with mammals, where methylation is primarily found in the CG context and, in specific cell types, also in the CH context (Chad E. & Schmitz, 2017).

The most common type of methylation across the different kingdoms, CG methylation, is also the most predominant in plants (Chad E. & Schmitz, 2017). One reason for this wide distribution is probably found in the relatively simple maintenance mechanism after DNA replication. Because CG and CHG contexts are symmetrical, due to the obligatory cytosine on the complementary strand (i.e., complementary to G), cytosine methylation can be copied from the old strand to the newly synthesized strand. In comparison, CHH sites are asymmetrical (i.e., the complementary strand does not contain C) and require special machinery to be maintained during DNA replication since no complementary methylated sequence is available to guide the re-methylation (Foyer & Noctor, 2013). In summary, cytosine methylation is the primary epigenetic mark, partly because it is technically easy to access (see Chapter 11). However, the consequences of cytosine methylation at a specific site depend on the position and the sequence context and are not easily interpreted.

Summary

DNA methylation plays a crucial role in the regulation of gene expression, in the activity of transposable elements, in the defense against foreign DNA, and even in the inheritance of specific gene expression patterns (Xu, Tanino, & Robinson, 2016; Finnegan, Genger, Peacock, & Dennis, 1998; Xu, Tanino, Horner, & Robinson, 2016). DNA methylation refers to the cytosine methylation process through the covalent enzyme-catalyzed transfer of a methyl group from S-adenosylmethionine to the 5’ position of cytosine, thus converting cytosine to 5-methylcytosine (5mC) (Pikaard et al., 2014; Sahu et al., 2013). DNA methylation in plants is species-, tissue-, organelle-, and age-specific. It is controlled by phytohormones, changes during plant development, and biotic and abiotic stress conditions (Finnegan et al., 1998). This epigenetic mark can be accumulated during plant vegetative phases and, principally, be passed on to the next generations. DNA cytosine methylation appears in three contexts, CG, CHG, and CHH, where H can be A, C, or T (Sahu et al., 2013). It predominantly occurs on transposons and other repetitive DNA elements in the genome. DNA methylation patterns must be stably maintained to ensure that transposons remain silenced and preserve cell-type identity. Three different pathways maintain DNA methylation differing in their central enzyme: the DNA METHYLTRANSFERASE 1 (MET1) maintains CG methylation, the CHROMOMETHYLASE (CMT3), a plant-specific DNA methyltransferase, maintains CHG methylation, and the DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) maintains the asymmetric CHH methylation through de novo methylation (Law & Jacobsen, 2011). Although DNA methylation is a stable epigenetic mark in most cases, reduced levels of methylation are observed during plant development. Methylation loss can either occur passively via replication without functional maintenance methylation pathways or actively by removing methylated cytosines with DNA glycosylase activity. The symmetrical CG or CHG methylation is inherited during the DNA replication in the form of hemimethylated sequences. It provides the memory of the methylation imprint present in the parental DNA, suggesting a role in stress protection memory (Suzuki & Bird, 2008). On the contrary, the asymmetrical cytosine methylation must be reestablished de novo after each replication cycle. DNA methylation in plants is closely associated with histone modifications, and it affects the binding of specific proteins to DNA and the formation of respective transcription complexes in the chromatin (Pikaard et al., 2014; Zamir, 2001). Those epigenetic marks trigger chromatin remodeling, which plays a crucial role not only in transcriptional regulation but also in DNA repair and replication (Kim et al., 2019). It has been proposed that MET1 and DDM1 could be involved in the DNA damage response reducing chromatin density (Kim et al., 2019; Shaked, Avivi-ragolsky, & Levy, 2006). DDM1 mutations generate a strong alteration in the nuclear organization and chromatin structure, particularly in the centromeric and pericentromeric regions, resulting in the impediment of the DNA repair machinery that loses its access to the damaged sequences. This emphasizes the broad involvement of recombination and DNA repair proteins in plant genome maintenance and the link between epigenetic and genetic processes.

References

Kim, J. H. (2019). Chromatin remodeling and epigenetic regulation in plant DNA damage repair. International Journal of Molecular Sciences, 20(17). https://doi.org/10.3390/ijms20174093

Finnegan, E. J., Genger, R. K., Peacock, W. J., & Dennis, E. S. (1998). DNA METHYLATION IN PLANTS.

Law, J. A., & Jacobsen, S. E. (2011). patterns in plants and animals, 11(3), 204–220. https://doi.org/10.1038/nrg2719.Establishing

Pikaard, C. S., Scheid, O. M., Kingston, R. E., Tamkun, J. W., Baulcombe, D. C., & Dean, C. (2014). Epigenetic Regulation in Plants Epigenetic Regulation in Plants, 1–31. https://doi.org/10.1101/cshperspect.a019315

Sahu, P. P., Pandey, G., Sharma, N., Puranik, S., Muthamilarasan, M., & Prasad, M. (2013). Epigenetic mechanisms of plant stress responses and adaptation. Plant Cell Reports, 32(8), 1151–1159. https://doi.org/10.1007/s00299-013-1462-x

Shaked, H., Avivi-ragolsky, N., & Levy, A. A. (2006). Involvement of the Arabidopsis SWI2/SNF2 Chromatin Remodeling Gene Family in DNA Damage Response and Recombination, 2(June), 985–994. https://doi.org/10.1534/genetics.105.051664

Suzuki, M. M., & Bird, A. (2008). DNA methylation landscapes: provocative insights from epigenomics, 9(June), 465–476. https://doi.org/10.1038/nrg2341

Xu, J., Tanino, K. K., Horner, K. N., & Robinson, S. J. (2016). Quantitative trait variation is revealed in a novel hypomethylated population of woodland strawberry (Fragaria vesca). BMC Plant Biology, 16(1), 1–17. https://doi.org/10.1186/s12870-016-0936-8

Xu, J., Tanino, K. K., & Robinson, S. J. (2016). Stable Epigenetic Variants Selected from an Induced Hypomethylated Fragaria vesca Population. Frontiers in Plant Science, 7(November), 1–14. https://doi.org/10.3389/fpls.2016.01768

Zamir, D. (2001). Improving plant breeding with exotic genetic libraries. Nature Reviews Genetics, 2(12), 983–989. https://doi.org/10.1038/35103590

Plant genomes are dynamic and differ strongly in size due to differences in gene content, the number of transposons, or other repetitive sequences, which influence the diversification of DNA methylation mechanisms (Chad E. & Schmitz, 2017; Pellicer, Hidalgo, Dodsworth, & Leitch, 2018). Several genome-wide methylome studies applying bisulfite sequencing (BiSeq) (see Chapter 11) demonstrated that plants have a higher epigenome diversity among species than animals (Fig. 6). This can be attributed to genetic variation, for example, large differences in the amount of heterochromatin and to the three different cytosine contexts in plants (Jones, 2012; Niederhuth et al., 2016; Pellicer et al., 2018; Yi, 2017).

Early papers reported cytosine methylation is mainly restricted to the nuclear DNA, suggesting DNA methylation in plastid genomes does not play a role in controlling gene activity (Ahlert, Stegemann, Kahlau, Ruf, & Bock, 2009; Finnegan, Genger, Peacock, & Dennis, 1998). However, high levels of N6-methyladenosine (m6A) methylation findings in the chloroplast and mitochondria propose the presence of methylation machinery inside these organelles. In addition, RNA methylation was identified, although its role in plant organelles is yet not fully understood (Manduzio & Kang, 2021).

The distribution of methylcytosines over the nuclear genome varies among species. Generally, it is concentrated in regions rich in repeated sequences, which include the centromere surrounding DNA and telomeres, or in genome regions containing many transposons (Finnegan, Genger, Peacock, & Dennis, 1998).

Such patterns of context-dependent accumulation of cytosine methylation can also be seen on a smaller scale. When we average methylation frequency across genes, transcription start sites, or transposons, as shown in the next infographic (Fig. 7).

The comparative epigenomic analysis identifies how dynamic the methylomes between flowering plant species can be. Families such as Brassicaceae and most Poaceae showed globally lower mCHG and mCHH methylation than other plants. One reason for these pronounced differences is that large genomes, like Zea mays, contain much higher numbers of repetitive sequences and transposons than smaller genomes, like strawberry. These sequences are commonly characterized by higher methylation levels. On the other hand, the gbM of ortholog genes showed a conserved pattern across species (Niederhuth et al., 2016). In conclusion, the variation in DNA methylation between plant species opens new areas of study to understand the role of DNA methylation and their correlation with evolutionary distance as well as biological diversity.

Maintenance of DNA methylation patterns

It is essential for plants to maintain the global DNA methylation patterns to keep transposons silenced, preserve cell type identity, or establish an epigenetic memory against environmental stresses (Law & Jacobsen, 2010). According to the sequence context of the cytosine sites (CG, CHG, and CHH), DNA methyltransferases control and regulate their methylation due to cooperative or competing interactions by different mechanisms (Meyer, 2011; H. Zhang et al., 2018). The efficiency of the different DNA methyltransferases is reflected in the methylation level at their preferred target sites. Symmetric methylation (CG and CHG) is most efficient, with most sites being 80–100% methylated, while methylation levels at nonsymmetric (CHH) target sites vary between 10% to 20% (Martínez-García et al., 2010; P. R. V. Satyaki & Gehring, 2017a).

The sequence context further influences DNA methylation in plants even beyond the mentioned dinucleotide and trinucleotide sites. For example, CG sites are undermethylated when the exact four-base context is ACGT. Also, CHG and CHH sites are less efficiently methylated when another cytosine follows the C, as in CCG instead of CAG or CTG. These more specific sequence contexts seem to reduce the efficiency of the methyltransferases. However, the reason is yet unknown. For example, in Arabidopsis thaliana, some methyltransferases share similar sequence specificities to probably provide a methylation backup system and avoid harmful alterations in the plant genome (Law & Jacobsen, 2010; Li et al., 2018; Meyer, 2011).

Maintenance of CG methylation

In plants, genetic analyses have demonstrated that CG methylation is regulated like in mammals through homolog proteins. CG cytosine methylation is maintained by the DNA METHYLTRANSFERASE 1 (MET1), which recognizes hemimethylated CG dinucleotides following DNA replication and methylates the unmodified cytosine in the new complementary strand (Fig. 2). To do this, MET1 probably is recruited to the replication complex by the VARIANT IN METHYLATION (VIM) protein family of SRA (SET- and RING-associated) domain proteins (Law & Jacobsen, 2010; Zhang et al., 2018). This interaction with VIM was proposed because VIM1 loss of function mutants of Arabidopsis thaliana lose the DNA methylation of their centromeres (J. Kim, Kim, Richards, Chung, & Woo, 2014).

Prevalence of non-CG methylation in plants

Non-CG sites refer to symmetrical trinucleotide CHG sites and to asymmetrical CHH trinucleotide sites. Non-CG methylation is maintained by plant-specific enzymes such as the CHROMOMETHYLASE (CMT) family (Kenchanmane Raju, Ritter, & Niederhuth, 2019). CHG methylation is maintained via a reinforcing loop, which involves histone and DNA methylation (Fig. 3) (Law & Jacobsen, 2010). The DNA methyltransferases CHROMOMETHYLASE 3 (CMT3) and, to less extent, CMT2 catalyze the maintenance of CHG methylation in A. thaliana (Fig. 3A) (Zhang et al., 2018). However, they require a whole complex of other proteins. For example, the histone methyltransferases SUPPRESSOR OF VARIEGATION 3-9 HOMOLOGs: SUVH4 (also known as KRYPTONITE (KYP)), SUVH5, and SUVH6 bind non-CG methylated regions and methylate H3K9 (Fig. 3B)(Kenchanmane Raju et al., 2019). SUVH4, primarily responsible for H3K9 demethylation, is an essential partner, as suggested by a dramatic decrease of DNA CHG methylation in loss of function mutants (Law & Jacobsen, 2010). Law & Jacobsen (2010) suggested that the reason for the interdependence of CMT3 and SUVH4 could be found in their multidomain structure. Because CMT3 can bind to the methylated histone H3 with a chromodomain, SUVH4 can bind to CHG methylated DNA with its SRA domain (Law & Jacobsen, 2010; Zhang et al., 2018). Interestingly, CMT3 appears to have a role also in gene-body methylation (gbM) in several plant species (Wendte et al., 2019). In Eutrema salsugineum, a plant that naturally lacks gbM, the gain of CMT3 triggered a new establishment of gbM in genes homolog to naturally methylated genes in A. thaliana. The gained gbM was maintained even in following generations (Wendte et al., 2019). However, this observation is still a very recent one.

Asymmethric DNA methylation

CHH methylation cannot be copied from the old strand, and therefore is maintained by constant de novo methylation. De novo DNA methylation is a process that involves plant-specific pathways and enzymes. For instance, the methyltransferases DRM2 and CMT2 catalyze this reaction, depending on the broader sequence context (Henderson et al., 2010). DRM2 is part of the RNA-directed DNA methylation (RdDM) pathway (described in more detail in Chapter 3) that guides DRM2 to small-RNA target regions rich in transposons or other repeat sequences primarily located in the Euchromatin. On the contrary, CTM2 targets H1-containing Heterochromatin sites (Zhang et al., 2018). While the Euchromatine is the region containing most of the protein-coding genes, Heterochromatin contains most of the transposons and other repeat sequences. Other enzymes may affect the maintenance of asymmetric CHH methylation, like MET1, CMT3, SuvH2, and SuvH9, so there is also some cross-talk with H3K9 methylation. However, only CTM2 and DRM2 can catalyze this reaction. But in turn, these two enzymes can also de novo methylate cytosines in other sequence contexts (Zhang et al. 2018).

Notably, CMT2 maintained DNA methylation is reduced by mutations in DECRESED DNA METHYLATION 1 (DDM1), a chromatin-remodeling protein (Zhang et al., 2018). This association was used in a pivotal study to produce a population of Arabidopsis thaliana recombinant inbred lines with a nearly identical genomic sequence but different cytosine methylation patterns, called EpiRiLs (Johannes et al., 2009). The study demonstrated for the first time that differences only in DNA methylation produced phenotypic variation.

DNA demethylation

It is known that DNA methylation is a stable epigenetic mark across species; however, during different plant lifetime processes, a decrease in global methylation has been observed (Law & Jacobsen, 2010). A reduction of methyltransferases activity or low levels of methyl donors present may result in failure to conserve methylation during DNA replication , which is known as passive DNA demethylation. On the contrary, removing the methyl group by an enzymatic process is described as active DNA demethylation (Zhang et al., 2018).

Active demethylation

Active demethylation implies a family of DNA demethylases, 5-mC DNA glycosylases–apurinic/apyrimidinic lyases, which drive the base excision repair (BER) pathway (Law & Jacobsen, 2010; Zhang et al., 2018). In A. thaliana, the group of glycosylases involves REPRESSOR OF SILENCING 1 (ROS1), TRANSCRIPTIONAL ACTIVATOR DEMETER (DME), DEMETER-LIKE PROTEIN 2 and 3 (DML2 and DML3), which are able to identify and extract 5-mC from all cytosines in a double-stranded DNA sequence (Law & Jacobsen, 2010; Zhang et al., 2018).

During DNA demethylation, DME/ROS1 behave as glycosylases hydrolyzing the glycosylic bond between the cytosine and the deoxyribose molecule, then apurinic or apyrimidinic lyases cut the DNA backbone and produce an excision of the methyl-cytosine base, which will be filled through a DNA polymerase and ligase enzymes with a non-methylated cytosine base (Figure 4). However, the exact process, how the reposition of the unmethylated cytosine occurs is still unidentified (Parrilla-Doblas, Roldán-Arjona, Ariza, & Córdoba-Cañero, 2019; H. Zhang et al., 2018).

The glycosylases ROS1, DML2, and DML3 are expressed preferably in vegetative tissues, possibly to counteract robust DNA methylation by the RdDM pathway. Law & Jacobsen (2010) suggest that the shared target sequences of these glycosylases with the RdDM pathway regulate gene expression activity by removing methylation cytosines in genes and preventing their silencing (Fig.5A). However, these enzymes also maintain an adaptable inactive state to keep silenced transposons. In addition, it has been shown that the activity of genes nearby transposons is negatively affected by the RdDM pathway, which maintains the methylation in Euchromatic regions (Fig. 5B) (M. Y. Kim & Zilberman, 2014; Law & Jacobsen, 2010; H. Zhang et al., 2018).

Unlike ROS1 and DML, DME has been observed preferentially expressed in plant gamete cells and being involved in imprinting, as will be further outlined in the section DNA methylation and imprinting (Wöhrmann et al., 2012).

Passive demethylation in plants

Passive demethylation has been characterized in plant endosperm where MET1 activity decreases during female gametogenesis and results in a drop of the global methylation (Jullien et al., 2008). The reduction of MET1 expression levels activates demethylation pathways directed by DME. Together this activates genes that are expressed in a parent-of-origin-specific manner (Law & Jacobsen, 2010).

Ecological systems are defined by the interactions of organisms and the physical processes affecting those organisms through space and time. As a result, ecological systems are complex, and to simplify the study of specific processes or interactions, ecologists have organized them in hierarchical levels (Lidicker, 2007). The lowest level of ecological organization is the population, defined as individuals of the same species that inhabit a given area simultaneously (Hannan & Freeman, 1977; Berryman, 2002). A collection of local populations that interact with each other and the larger area or region with a balance between extinction and recolonization rates is called a metapopulation (Hastings & Harrison, 1994). Individual populations are further organized into ecological communities. A community is an assemblage of populations of different species that live in a shared environment and interact with one another, forming together a distinctive system (Fauth et al., 1996).

When several communities are grouped in their abiotic surroundings, they are considered an ecosystem (Tansley, 1935). Furthermore, all communities exist in the broader spatial context of the landscape. Each landscape is composed of unique communities and ecosystems, and the broad-scale of geological and climatic patterns occurring around the globe give rise to regional patterns in the geographic distribution of ecosystems. Geographic regions that have similar geographical and climatic conditions support similar types of communities and ecosystems. These broad-scale regions dominated by similar types of ecosystems are referred to as biomes (Clements, 1916), and biomes are further grouped in what we call the biosphere, the sum of all the organisms on the planet and their environment, considered as a system of interacting components.

Two components, structure and dynamics, characterize every ecological level. The system structure results from demographic parameters such as the density of individuals that belong to specific populations or communities, the relative abundance of those individuals, or the number of species found in any given community. This species richness is the simplest measure of community structure, but not all species are equally abundant or affect the community in the same way. Foundation species are the base of a community and play a significant role in defining its structure (Ellison et al., 2005). When a single or a few species predominate within a community, those species are referred to as dominants (Grime, 1987), while a less abundant species with a disproportionate impact on community structure and dynamics are called keystone species (Holling, 1992). Population and community dynamics result from the interactions between the individual organisms and their environment and the individuals' interactions. Dominant or keystone species are essential within a community because small changes affecting this species can affect the whole system. This chapter will focus on how epigenetic diversity can influence ecological processes and interactions at many different organizational levels and how changes in lower levels can affect higher levels.

Aside from functional epigenetic diversity, conservation management could profit from epimutation markers as a new tool that could potentially replace SSR markers for some questions. The background is that heritable gains or losses of cytosine methylation can arise stochastically in plant genomes independently of DNA sequence changes. These are called ‘spontaneous epimutations’ and can be inherited across mitotic and meiotic cell divisions. They occur at rates four to five orders of magnitude higher than the DNA mutation rate per unit time and are neutral at the genome-wide scale accumulating in plant genomes in a ‘clock-like’ fashion (Yao et al. 2021). Emerging evidence indicates that these properties can be exploited for reconstructing and timing recent evolutionary events and for age dating long-lived perennials (see figure 1).

Figure 1: Epimutations have Clock-like Properties and can be used for example to date long-lived plants. Source: Yao et al. 2021.

Consequently, spontaneous epimutations can be used for the development of biomarkers to study wild populations' ecological structuring, and the study of landscape connectivity (Rey et al. 2019), in relation with conservation efforts of clonal plants, with reconstructing the dispersal of invasive clonal plant species and also help to differentiate between naturally dispersed plants and escaped garden plants.

One recent example of the usage of epigenetic biomarkers was shown in the pruning systems used in vineyards that induced detectable DNA methylation signatures in vines even at narrow geographical scales (Xie et al., 2017). In a conservation perspective, this example illustrates how methylation markers could be used to determine conservation units accounting not only for the long-term evolutionary history of organisms but also for some important fractions of their current ecological context.

Another example stems from well-illustrated populations of the perennial herb Helleborus foetidus of south-eastern Spain. Here Herrera and coworkers (2017) established the genetic, epigenetic and phenotypic structures of subpopulations on 10 geo-graphically distant sites. These sites had diverging environmental conditions and the genetic structure followed a classical isolation-by-distance pattern. The epigenetic structure in contrast, clearly followed an isolation-by-environment pattern, better reflecting the ecological processes that have shaped population phenotypic differentiation (Herrera et al., 2017).

It is very likely that in plants conservation epigenetics will never take on a central role in conservation biology as conservation genetics has. However, we do think that plant epigenetics will be a valuable complementation to conservation genetics possibly helping to put focus on the importance of microenvironmental heterogeneity in conservation and providing valuable tools based on spontaneous epimutations. It might be a stretch though, to invent yet another research field with the term Conservation Epigenetics.

Clements, F. E. (1916). Plant succession: an analysis of the development of vegetation. Carnegie Institution of Washington.

Grime, J. P. (1987). Dominant and subordinate components of plant communities: implications for succession, stability and diversity. Symposium of the British Ecological Society.

Madritch, M., Donaldson, J. R., & Lindroth, R. L. (2006). Genetic identity of Populus tremuloides litter influences decomposition and nutrient release in a mixed forest stand. Ecosystems, 9(4), 528–537.

Mooney, K. A., & Singer, M. S. (2012). Plant effects on herbivore-enemy interactions in natural systems. Trait-Mediated Indirect Interactions: Ecological and Evolutionary Perspectives, 107–130.

In eukaryotes, DNA is tightly packed and functionally organized inside the nucleus. From a megabase to a local scale, the hierarchical chromatin organization influences several fundamental cellular processes, including transcription, replication, and DNA repair. At the base of this hierarchical organization is the nucleosome, which consists of 147 bp of DNA wrapped around an octamer of histone proteins. Chromatin accessibility at the nucleosomal level is regulated by the combined action of two classes of proteins: histone-modifying enzymes and chromatin remodelers. Together, they coordinate the cellular processes dictated by the genome in response to developmental or environmental stimuli.

Histone-modifying enzymes post-translationally modify the N-terminal tails of histone proteins through acetylation, phosphorylation, ubiquitination, ADP-ribosylation, or methylation. These histone modifications can be reversed by a set of antagonistic enzymes catalyzing the removal of chemical modifications. The combinatorial arrangement of histone marks defines and partitions the genome into functional domains, such as transcriptionally silent heterochromatin and transcriptionally active euchromatin. For instance, histone acetylation has generally been associated with transcriptional activation, whereas histone ubiquitination most frequently plays a role in transcriptional repression and histone methylation has been involved in both, transcriptional activation and silencing. The local combination of inter-related histone modifications forms the “histone code”, which is interpreted by effector proteins or “readers” that recognize and bind to modifications through specific domains to regulate transcription.

The second class of chromatin-modifying factors are the ATP-dependent chromatin-remodeling enzymes. ATP-dependent chromatin remodelers alter nucleosomal structure and DNA accessibility. They mediate gene activation by repositioning (slide, twist, or loop) of nucleosomes along the DNA, evicting histones, or facilitating exchange of histone variants. ATP chromatin remodelers are categorized according to their biochemical activities and ATPase subunit similarity into four different subfamilies: Switching Defective/Sucrose Non-Fermenting (SWI/SNF), Imitation Switch (ISWI), Chromatin Helicase DNA-Binding (CHD) and Inositol Requiring (INO80). ISWI and CHD families participate in nucleosome spacing in chromatin assembly after replication; SWI/SNF subfamilies are important for nucleosomal disassembly, and INO80 and SWR1 complexes have opposite roles in histone variant exchange.

In this chapter we will describe the current knowledge about chromatin organization, from higher order chromatin structures to nucleosomal local scale. We will then describe the different histone modifications and chromatin remodelers and their role in regulating chromatin modifications and transcription to control cellular and physiological responses in plants.

Overview

Though it is by no means the only biological mechanism within the domain of epigenetics, DNA methylation is among the most prevalent and the most studied throughout the field. In earlier chapters, we have discussed underlying molecular processes and the ecological consequences of differential methylation within species- but how exactly do we detect these differences? One technique that has emerged at the forefront of epigenetic research is bisulfite sequencing: a distinct adaptation of next-generation sequencing which produces genome-wide methylation profiles at a nucleotide-level resolution.

The technique, devised by Frommer et al. [Frommer 1992] and refined for modern sequencing techniques by Lister et al. [Lister 2008] and Cokus et al. [Cokus 2008], involves the treatment of extracted DNA from test samples with sodium bisulfite, a deaminating agent which mediates the conversion of unmethylated cytosine nucleotides into uracil. Cytosine bases that carry methyl groups (e.g. 5-methylcytosine, 5-hydroxymethylcytosine) are left unaffected by the treatment and remain in their original unconverted state. As uracil residues are subsequently converted to thymine during the PCR step of standard DNA sequencing, these bisulfite-treated samples can be subjected to standard sequencing protocols and used to generate sequencing reads, which carry epigenetic information. Once treated, the reads effectively reframe the research question from a biological to a computational, algorithmic concern, at least until the results require interpretation.

In standard sequencing, the next step is to follow workflows for read alignment of the sequencing reads to a reference genome assembly. Aligning sequencing reads uses overlapping sequences to puzzle as many reads as possible into long sequence fragments, so-called contigs. The alignment presents some issues when handling bisulfite data, as thymine residues can no longer be considered as entirely independent entities to cytosine. Read alignment algorithms usually operate based on scoring matrixes, which assign an overall probability for the alignment of two sequences based on the number and position of matches, mismatches, insertions, and deletions between nucleotides. The problem arises in that reference cytosines can conceptually match thymines in bisulfite-treated reads, but not vice versa. Existing algorithms are often not built to handle this asymmetry between bases, so the solution is either to adapt these tools in some way further or to operate specifically with algorithms designed for bisulfite data. Several tools now exist in representation of either category, including notably Bismark [Krueger 2011] and BWA-meth [Pederson 2014], which adapt the popular standard aligners bowtie [Langmead 2013] and BWA [Li 2010], and software such as segemehl [Otto 2012] or ERNE-BS5 [Prezza 2012] which are capable of interpreting bisulfite reads in their own right.

The principles of bisulfite sequencing notwithstanding, another important consideration when designing such an experiment involves the chosen strategy for library preparation. Like generally in next-generation-sequencing, sequencing depth and coverage are also very important for bisulfite-sequencing, as we seek to maximize sequencing coverage regarding the scope of the questions we are looking to answer and the practical limitations of the study, such as cost and time. For example, does your study seek to investigate genome-wide methylation patterns, or is it enough to focus on a reduced subset of the DNA? Herein, we will consider Whole-Genome Bisulfite Sequencing (WGBS) applications, Reduced-Representation Bisulfite Sequencing (RRBS), and variations of these methods. In particular, what implications have such protocols on data robustness, and how should we adjust the procedure to improve quality control during the downstream analyses?

The chapter covers various technical concerns of bisulfite sequencing, from DNA extraction and library preparation to sequencing itself and the downstream extraction of methylated positions. The bioinformatic principles determine the data validity for answering the questions posed by the study, and an a priori consideration, therefore, is fundamental to the successful outcome of any such experiment. Finally, we discuss the rigid limitations of bisulfite sequencing and give brief suggestions for alternative methods that might be used to address these issues.

Whole Genome Bisulfite Sequencing (WGBS)

WGBS is the practice of applying bisulfite sequencing on a genome-wide scale, capturing all regions, and attempting to define global methylation patterns in each sample. This method is appropriate when the study question is broad in scope or if prior information on the genomic regions of interest is limited. It can be considered the "go-to" approach when other methods for concentrating the sequencing on reduced subsets of the genome are either unavailable or inappropriate for the study in question.

There are two main variations of WGBS library preparation, known as BS-Seq [Cokus 2008] and MethylC-Seq [Lister 2008] (Figure 2). In terms of the protocol, they differ primarily in the number of PCR steps and when the ligation of sequencing adaptors occurs relative to the treatment with sodium bisulfite. Many sequencing technologies require specific sequencing adaptors to facilitate base calling on selected DNA fragments. In the case of Illumina, these adaptors are bound to complementary sequences on the flow cell, forming clusters to be sequenced by synthesis. If the adaptor is not present, the DNA molecule is simply washed off the cell, and no information is retrieved. The issue here is that the bisulfite treatment alters the sequence of these adaptors wherever there are unmethylated cytosines present, rendering them incompatible with the complementary sequences on the flow cell. MethylC-Seq addresses this by using custom cytosine-methylated adaptors that remain unaffected by sodium bisulfite. In contrast, BS-Seq circumvents the issue by ligating the adaptors only after the bisulfite treatment.

In principle, the approach of BS-Seq seems more straightforward. However, ligating the sequencing adaptors after the bisulfite treatment presents another problem. The two strands of DNA are no longer complementary to each other and hence remain in a single-stranded state. This is a problem because sequencing adaptor ligation requires duplex DNA. Therefore an additional round of PCR is necessary before adaptor ligation can occur. This PCR step generates reverse complementary strands to both the bisulfite-treated Watson (+FW) and Crick (-FW) strands of the original DNA, which are themselves distinct sequences (+RC and -RC, respectively). The result is a set of four sequences where both the FW and RC strands are indistinguishable from each other by the sequencer. Strand-specificity is therefore lost, and additional bioinformatic processing is required to resolve which reads belong to which strand. In MethylC-Seq, only the +FW and -FW sequences are present, and strand-specificity is cleanly preserved during sequencing, though with paired-end data, it becomes more complex as the +RC and -RC strands are present as well.

A more recent variation of these approaches has also been developed, known as post-bisulfite adaptor tagging (PBAT) [Miura 2012]. In this case, the bisulfite conversion process itself is first used to fragment the genomic DNA. Adaptor ligation is then facilitated by two rounds of random priming extension in place of PCR, thereby maintaining strand-specificity while avoiding any denaturation of adaptor-ligated DNA. The real advantage of this method, however, is its sensitivity in handling sub microgram quantities of DNA without the need for additional amplification, contrary to MethylC-Seq, where the bisulfite treatment often fragments adaptor-ligated DNA templates, which then cannot be used during sequencing. In such a case, the remaining DNA may need to be amplified to achieve a reasonable DNA mass for sequencing, but this amplification risks inducing PCR artifacts. The approach of PBAT can circumvent the need for PCR amplification on sub microgram quantities of DNA. Still, it should be noted that random primer extension is subject to its own biases. Sequence-specific site preferences can give rise to "pile-ups" of reads, and differential priming between methylated and unmethylated alleles has been hypothesized. Therefore, it may be preferable to run MethylC-Seq with a very low number of PCR amplification cycles (e.g., ~ 4) in cases where sample availability is not strictly limited. Last but not least, an even newer approach (TAPS) was published recently that avoids using the bisulfite conversion altogether, allowing for higher mapping quality and nonfragmented duplex DNA after the conversion of methylated cytosines into thymines (Yibin et al. 2019). The downside of TAPS, it is too new to be available as a commercial kit. TAPS is very new and promising, but experiences with this method are yet scars, and so we do not discuss it further in this chapter.

Regardless of the approach selected, at least two cycles of post-bisulfite PCR are necessary to facilitate the conversion of uracil to thymine before sequencing can occur. For these PCRs, the presence of uracil in the sequence precludes the use of many standard, high-fidelity polymerase enzymes with proofreading mechanisms such as Phusion (Thermo Scientific) or KAPA HiFi (Roche). On encountering uracil, these enzymes stall as they await base excision repair [Lindahl 1999, Greagg 1999]. Fortunately, there are alternatives available, such as PfuTurbo Cx (Agilent) or KAPA HiFi uracil+ (Roche), specifically designed for bisulfite sequencing. Once a library has been prepared, it is standard practice to perform library quantification and normalization, using, for example, Qubit / PicoGreen assay or qPCR measurement. It should be noted during this step that methods that estimate only the total quantity of DNA may fail to give an accurate representation of the adapter-ligated DNA, particularly in MethylC-Seq libraries due to the aforementioned fragmentation caused by the bisulfite treatment. For this reason, it is recommended to use a BioAnalyzer for sizing only and qPCR to quantify the final library for bisulfite sequencing.

Several commercial kits are readily available for carrying out bisulfite conversion itself. Depending on your sample DNA quantity and library preparation methodology, the aim is to achieve maximum conversion efficiency relative to optimal DNA recovery. High temperature, high bisulfite molarity, and long incubation times are more likely to yield complete bisulfite conversion but degrade much of the DNA in the process. Incomplete conversion, however, leads to an overestimation of methylation levels on unconverted cytosines. With this trade-off in mind, a good evaluation of modern kits was provided by Kint et al. [Kint 2018], where EpiTect Bisulfite (Qiagen), EZ DNA Methylation-Gold (Zymo Research), and EZ DNA Methylation-Lightning (Zymo Research) kits were each cited for high performance with regards to several study-dependent factors.

To estimate conversion efficiency within bisulfite-treated samples, it is typical to have a control consisting of a known quantity of unmethylated DNA within the sample. Historically the conversion rate was estimated from non-CG cytosines in mammals [Hodges 2009], which is inappropriate for plants where DNA methylation occurs in the CHG and the CHH contexts. Alternatively, the mitochondria or chloroplast genomes were used, as both organelles are widely considered to escape DNA methylation [Marano 1991, Vanyushin 1988]. However, this may not be entirely reliable as conflicting evidence of DNA methylation has since been reported in both [Šimková 1998, Fojtová 2001]. Therefore, the most reliable method in plants is to use a "spike-in" of DNA from another source. The enterobacteria phage Lambda (~ 0.1% w/w) is often used, which is shown to be virtually devoid of 5mC when propagated on mutant bacteria strains lacking DNA methylase activity [Hattman 1973]. Reads aligning to the Lambda genome can then indicate the level of bisulfite conversion, as in theory, all cytosines should have been replaced with thymines.

In addition to Lambda, as noted earlier, the bacteriophage Phi X is commonly used as a "spike-in" to balance base proportions [Raine 2018, Illumina bulletin]. During the initial cycles of Illumina sequencing, the phasing/pre-phasing, color matrix corrections, and pass filter calculations are influenced by the flow cell imaging [Illumina bulletin]. In bisulfite-treated DNA, there is a notable deficiency in cytosine bases and the fluorescent color associated with it, which can skew the base-calling algorithm during this normalization process. Adding the well-balanced Phi X DNA [Sanger 1977] or any other well-balanced DNA to the sequencing library allows the Illumina sequencing to proceed unaffected. Another interesting possibility is to multiplex a bisulfite-treated library with an untreated library with each DNA fragment containing an identifying adapter sequence indicating which library it belongs to. This way, spiking can be omitted, and cytosine methylation and single nucleotide polymorphisms (SNPs) can be obtained from one sequencing run.

Reduced Representation Bisulfite Sequencing (RRBS)

RRBS is similar to WGBS in many ways but differs primarily by adding an initial selection procedure at the beginning of the library preparation. It was developed by Meissner et al. [2005] to generate large-scale sequencing data with a lower resolution than WGBS and still evenly representing the genome, though with the option to focus either on Eu- or Heterochromatin. This reduces the sequencing cost compared to WGBS but results in the loss of much sequence content that could otherwise be relevant. In cases where this technique was employed, the enriched fraction was frequently reported to be less than 1% of the whole genome size.

Sample DNA is first subjected to a restriction endonuclease that targets a specific sequence context depending on the local cytosine methylation status. A typical enzyme used is MspI, which targets CG sites in the specific sequence 5’-CCGG-3'. MspI can not cleave when this specific recognition sequence is symmetrically methylated, thus focuses on weakly methylated euchromatin rather than the heavily methylated heterochromatin in the chromosomes. Different sequence contexts require different enzymes, although this application has not been broadly applied in non-CG contexts. The enzymatic digestion produces fragments that can be size selected, usually following some additional end repair and A-tailing depending on which enzyme was used. The rest of the library preparation follows closely with that which was outlined previously for WGBS and unfortunately suffers from the same loss of strand-specificity as BS-Seq. It must be noted here that the recognition site's methylation is not the main focus of the study. Instead, the sequence flanking the recognition site is sequenced, providing information on the methylation status of many cytosines, which can principally be in all three sequence contexts.

Target capture bisulfite-sequencing

Reduced representation bisulfite sequencing is a beneficial technique when the aim is to sequence many biological samples, for example, to study population genetics or when the studied organism has a very large genome, like in many coniferous trees, for example. The technique further allows to roughly direct the analysis either to heterochromatin or euchromatin or, depending on the genome in question, enrich promotor or gene-body sites by choosing the appropriate cleavage enzyme. However, besides this possibility of setting a rough focus of the study, the idea is to provide a valid representation of the genome through a sample of random sequence reads scattered across the genome. Though, it may be desirable in a project to set the target more specifically to a particular region in the genome. This can be achieved through target capturing, which can be applied before or after bisulfite conversion (Wreczycka et al. 2017). Different techniques usually involving the hybridization of genomic DNA with the complementary of a known piece of the target sequence, combined with bisulfite conversion and followed by the above-described processing of converted DNA, enable the inference of the methylation status of a specific target location in the genome of interest. Such techniques may be helpful when, for example, unraveling the methylation status of a known promotor region is the aim of the investigation.

Ahlert, D., Stegemann, S., Kahlau, S., Ruf, S., & Bock, R. (2009). Insensitivity of chloroplast gene expression to DNA methylation. Molecular Genetics and Genomics, 282(1), 17–24. https://doi.org/10.1007/s00438-009-0440-z

Batista, R. A., & Köhler, C. (2020). Genomic imprinting in plants-revisiting existing models. Genes & Development, 34(1–2), 24–36. https://doi.org/10.1101/gad.332924.119

Bewick, A. J., & Schmitz, R. J. (2017). Gene body DNA methylation in plants. Current Opinion in Plant Biology, 36, 103–110. https://doi.org/10.1016/j.pbi.2016.12.007

Chad E., N., & Schmitz, R. J. (2017). Putting DNA methylation in context: from genomes to gene expression in plants, 4(11), 149–156. https://doi.org/10.1016/S2214-109X(16)30265-0.Cost-effectiveness

Cheng, J., Niu, Q., Zhang, B., Chen, K., Yang, R., Zhu, J. K., … Lang, Z. (2018). Downregulation of RdDM during strawberry fruit ripening. Genome Biology, 19(1), 1–14. https://doi.org/10.1186/s13059-018-1587-x

Dong, X., Chen, J., Li, T., Li, E., Zhang, X., Zhang, M., … Lai, J. (2018). Parent-of-origin-dependent nucleosome organization correlates with genomic imprinting in maize. Genome Research, 28(7), 1020–1028. https://doi.org/10.1101/gr.230201.117

Finnegan, E. J., Genger, R. K., Peacock, W. J., & Dennis, E. S. (1998). DNA methylation in plants. Annual Review of Plant Biology, 49(1), 223–247. https://doi.org/10.1146/annurev.arplant.49.1.223

Foyer, C. H., & Noctor, G. (2013). Redox Signaling in Plants. Antioxidants & Redox Signaling, 18(16), 2087–2090. https://doi.org/10.1089/ars.2013.5278

Fujimoto, R., Kinoshita, Y., Kawabe, A., Kinoshita, T., Takashima, K., Nordborg, M., … Kakutani, T. (2008). Evolution and control of imprinted FWA genes in the genus Arabidopsis. PLoS Genetics, 4(4). https://doi.org/10.1371/journal.pgen.1000048

Henderson, I. R., Deleris, A., Wong, W., Zhong, X., Chin, H. G., Horwitz, G. A., … Jacobsen, S. E. (2010). The De novo cytosine methyltransferase DRM2 requires intact UBA domains and a catalytically mutated paralog DRM3 during RNA-directed DNA methylation in arabidopsis thaliana. PLoS Genetics, 6(10), 1–11. https://doi.org/10.1371/journal.pgen.1001182

Huang, H., Liu, R., Niu, Q., Tang, K., Zhang, B., Zhang, H., … Lang, Z. (2019). Global increase in DNA methylation during orange fruit development and ripening. Proceedings of the National Academy of Sciences, 116(4), 1430 LP – 1436. https://doi.org/10.1073/pnas.1815441116

Jiang, S. H., Sun, Q. G., Chen, M., Wang, N., Xu, H. F., Fang, H. C., … Chen, X. Sen. (2019). Methylome and transcriptome analyses of apple fruit somatic mutations reveal the difference of red phenotype. BMC Genomics, 20(1), 1–13. https://doi.org/10.1186/s12864-019-5499-2

Johannes, F., Porcher, E., Teixeira, F. K., Saliba-Colombani, V., Simon, M., Agier, N., … Colot, V. (2009). Assessing the impact of transgenerational epigenetic variation on complex traits. PLoS Genetics, 5(6). https://doi.org/10.1371/journal.pgen.1000530

Jones, P. A. (2012). Functions of DNA methylation: Islands, start sites, gene bodies and beyond. Nature Reviews Genetics, 13(7), 484–492. https://doi.org/10.1038/nrg3230

Jullien, P. E., Mosquna, A., Ingouff, M., Sakata, T., Ohad, N., & Berger, F. (2008). Retinoblastoma and Its Binding Partner MSI1 Control Imprinting in Arabidopsis. PLoS Biology, 6(8), e194. https://doi.org/10.1371/journal.pbio.0060194

Kenchanmane Raju, S. K., Ritter, E. J., & Niederhuth, C. E. (2019). Establishment, maintenance, and biological roles of non-CG methylation in plants. Essays in Biochemistry, 63(6), 743–755. https://doi.org/10.1042/EBC20190032

Kim, J., Kim, J. H., Richards, E. J., Chung, K. M., & Woo, H. R. (2014). Arabidopsis VIM proteins regulate epigenetic silencing by modulating DNA methylation and histone modification in cooperation with MET1. Molecular Plant, 7(9), 1470–1485. https://doi.org/10.1093/mp/ssu079

Kim, M. Y., & Zilberman, D. (2014). DNA methylation as a system of plant genomic immunity. Trends in Plant Science, 19(5), 320–326. https://doi.org/10.1016/j.tplants.2014.01.014

Law, J. A., & Jacobsen, S. E. (2010). Establishing, maintaining and modifying DNA methylation patterns in plants and animals. Nature Reviews Genetics, 11(3), 204–220. https://doi.org/10.1038/nrg2719

Li, X., Jake Harris, C., Zhong, Z., Chen, W., Liu, R., Jia, B., … Du, J. (2018). Mechanistic insights into plant SUVH family H3K9 methyltransferases and their binding to context-biased non-CG DNA methylation. Proceedings of the National Academy of Sciences of the United States of America, 115(37), E8793–E8802. https://doi.org/10.1073/pnas.1809841115

Manduzio, S., & Kang, H. (2021). RNA methylation in chloroplasts or mitochondria in plants. RNA Biology, 160. https://doi.org/10.1080/15476286.2021.1909321

Martínez-García, J. F., Galstyan, A., Salla-Martret, M., Cifuentes-Esquivel, N., Gallemí, M., & Bou-Torrent, J. (2010). Regulatory Components of Shade Avoidance Syndrome. Advances in Botanical Research, 53(November 2015), 65–116. https://doi.org/10.1016/S0065-2296(10)53003-9

Meyer, P. (2011). DNA methylation systems and targets in plants. FEBS Letters, 585(13), 2008–2015. https://doi.org/10.1016/j.febslet.2010.08.017

Niederhuth, C. E., Bewick, A. J., Ji, L., Alabady, M. S., Kim, K. Do, Li, Q., … Schmitz, R. J. (2016). Widespread natural variation of DNA methylation within angiosperms. Genome Biology, 17(1), 1–19. https://doi.org/10.1186/s13059-016-1059-0

Parrilla-Doblas, J. T., Roldán-Arjona, T., Ariza, R. R., & Córdoba-Cañero, D. (2019). Active DNA Demethylation in Plants. International Journal of Molecular Sciences, 20(19), 1–19. https://doi.org/10.3390/ijms20194683

Pellicer, J., Hidalgo, O., Dodsworth, S., & Leitch, I. J. (2018). Genome size diversity and its impact on the evolution of land plants. Genes, 9(2). https://doi.org/10.3390/genes9020088

Satyaki, P. R. V., & Gehring, M. (2017). DNA methylation and imprinting in plants: machinery and mechanisms. Critical Reviews in Biochemistry and Molecular Biology, 52(2), 163–175. https://doi.org/10.1080/10409238.2017.1279119

Schmitz, R. J., Schultz, M. D., Urich, M. A., Nery, J. R., Pelizzola, M., Libiger, O., … Ecker, J. R. (2013). Patterns of population epigenomic diversity. Nature, 495(7440), 193–198. https://doi.org/10.1038/nature11968

Vitte, C., Fustier, M. A., Alix, K., & Tenaillon, M. I. (2014). The bright side of transposons in crop evolution. Briefings in Functional Genomics and Proteomics, 13(4), 276–295. https://doi.org/10.1093/bfgp/elu002

Wendte, J. M., Zhang, Y., Ji, L., Shi, X., Hazarika, R. R., Shahryary, Y., … Schmitz, R. J. (2019). Epimutations are associated with CHROMOMETHYLASE 3-induced de novo DNA methylation. ELife, 8, 1–27. https://doi.org/10.7554/eLife.47891

Wöhrmann, H. J. P., Gagliardini, V., Raissig, M. T., Wehrle, W., Arand, J., Schmidt, A., … Grossniklaus, U. (2012). Identification of a DNA methylationindependent imprinting control region at the Arabidopsis MEDEA locus. Genes and Development, 26(16), 1837–1850. https://doi.org/10.1101/gad.195123.112

Yi, S. V. (2017). Insights into epigenome evolution from animal and plant methylomes. Genome Biology and Evolution, 9(11), 3189–3201. https://doi.org/10.1093/gbe/evx203

Zhang, H., Lang, Z., & Zhu, J.-K. (2018). Dynamics and function of DNA methylation in plants. Nature Reviews Molecular Cell Biology, 19(8), 489–506. https://doi.org/10.1038/s41580-018-0016-z

Zhang, X., Yazaki, J., Sundaresan, A., Cokus, S., Chan, S. W. L., Chen, H., … Ecker, J. R. R. (2006). Genome-wide High-Resolution Mapping and Functional Analysis of DNA Methylation in Arabidopsis. Cell, 126(6), 1189–1201. https://doi.org/10.1016/j.cell.2006.08.003 Zilberman, D. (2017). An evolutionary case for functional gene body methylation in plants and animals. Genome Biology, 18(1), 17–19. https://doi.org/10.1186/s13059-017-1230-2

As with many ecological studies involving next-generation sequencing, the experimental design is often based heavily around one fundamental trade-off. This trade-off is paramount to achieving a level of statistical power appropriate for the study's aim. In the previous chapter (chapter NGS sequencing), we discussed sequencing depth and coverage; this applies here as we seek to maximize sequencing coverage with regards to the scope of the questions we are looking to answer and the practical limitations of the study, such as cost and time. Next-generation sequencing is expensive, with costs driven by the quantity of material to be sequenced. This can be delineated to the total genome size, number of replicates, level of sequence coverage, and sequencing technology itself. Therefore, an ideal study seeks to define and optimize these factors a priori to carrying out the experiment and the subsequent downstream analyses.

Next-generation sequencing is a fast-developing field, with several commercial technologies currently being available to address various experimental needs and applications. These can broadly be categorized into short-read technologies (~ 50-900 bp), which tend to be cheaper with higher per-base accuracy and throughput than the counterpart long-read technologies (~ 1-500 kbp) [Li 2018]. The advantage of obtaining long reads is that they are less prone to assembly and mapping-related errors, making them suitable for resolving true genome arrangements even in repetitive regions and regions of low complexity, e.g., found in heterochromatin regions. Due to the issues mentioned above with generating long fragments from sodium bisulfite-treated DNA samples, however, these advantages are frequently lost. Short-read sequencing technologies, therefore, are dominantly selected for methylation analyses.

Among these short-read sequencing technologies, Ion Torrent [Rothberg 2011], Illumina [Bentley 2008], and SOLiD [Shendure 2006, McKernan 2006] have all been used successfully for bisulfite sequencing [Cokus 2008, Lister 2008, Lang 2017, Mirouze 2009, Bormann Chung 2010, Pabinger 2016, Venney 2016]. Each may be appropriate depending on the desired read size and number of reads per run, which is influenced by the size of the genome and the nature of the study in question (i.e., sample size). The most extensively used of these is Illumina, wherein the HiSeq 2500, 3000, and 4000 or for large projects, HiSeqX and NovoSeq are all appropriate high-throughput applications. A good comparison between these and some alternative systems is given by Grehl et al. [Grehl 2018]. A problem arises for bisulfite sequencing from the requirement of many sequencing machines that nucleotides should be present in the DNA fragments in roughly even proportions to perform efficient base calling. Following the bisulfite treatment, cytosines are underrepresented. To still enable high throughput sequencing, additional DNA with balanced base proportions and known sequence (e.g., Phi X) is added or "spiked in" during library preparation. This DNA standard is sequenced together with the target DNA and later filtered out. The disadvantage is that it reduces the potential sequencing coverage considerably because the machines may require 20-40% spiked in standard DNA. Grehl et al. [Grehl 2018] mention, however, that the HiSeq 2500 benefits specifically from an optimization of the cluster calling algorithm, which allows for the handling of bisulfite-treated libraries without the need for additional base proportion balancing (i.e., spiking). Thus, the HiSeq 2500 is a good baseline to act as a starting point for designing your sequencing experiments. However, your final choice should be influenced by whether or not the specific limitations of this method need addressing within the scope of your study.

As methylation calling is calculated from the number of overlapping reads aligning to each position, it is clear that the statistical power increases with greater sequencing depth. In an ideal scenario, the sequencing depth would be uniform and consistent across all positions of interest to the study, but this is rarely the case. Sequence-related biases in random hexamer priming [Hansen 2010], random DNA fragmentation [Poptsova 2014], and the PCR amplification [Kozarewa 2009, Aird 2011] stages of the library preparation impede uniformity and lead to coverage underrepresentation in regions of extreme GC-content [Benjamini 2012]. A straightforward approach in mitigating these issues is to select a level of coverage that ensures a minimum lower bound in the majority of regions where the distributed coverage is lower than the mean.

To give a point of reference, Ziller et al. [Ziller 2015] found that coverage between 5-15x was optimal in terms of statistical power for detecting differentially methylated regions between a range of human tissue and cell types in the CG context. Beyond that, resources would be better allocated towards expanding the number of biological replicates, starting at a minimum of two to capture within-group variance. In plants, it would be wise to consider this point of reference as a bare minimum for a homozygous diploid. Both the heterozygosity and the ploidy level invariably influence the minimum level of coverage due to the increased variation on single positions, which may lead to greater within-group heterogeneity necessitating a larger number of replicates. The highly-repetitive regions, as well as low-complexity regions often found in plant genomes, are also notoriously difficult to map to, leading to multi-mapped reads and alignment ambiguities [Treangen 2011]. This problem can be mitigated by increasing the coverage and using paired-end (PE) sequencing in the absence of long-read data. Furthermore, the magnitude of methylation differences is usually less pronounced in the CHG and CHH contexts than in the CG context requiring more power for detection. If the study seeks to capture differences in these contexts, then an increase in the number of replicates should be considered relative to CG alone.

Once the optimal level of coverage and number of replicates have been decided, it may be the case that the total genome size for the species of interest pushes the cost outside the range of affordability. In these instances, it is sometimes possible during library preparation to subset the material and exclude regions of minor or no relevance to the scope of the study. Whether this is appropriate or not depends on the research questions. An overview of each method is given in the following section.

When considering the epigenome, researchers may refer to changes in chromatin structure due to post-translational modification of histone proteins, populations of non-coding RNA (ncRNA), several chemical modifications to DNA sequences, or a combined effect of these factors. Most often, however, they are referring specifically to the methylome. That is to say: the distinct arrangement of methylcytosines present in the genome and the fluctuations between different organisms, tissues, or cell types within species. This generalization is not a reflection of a greater epigenetic significance but instead an overrepresentation of our current understanding of this type of methylation relative to other epigenetic mechanisms. However, recently more studies realize that epigenetic factors often interact.

In plants, DNA methylation can affect both cytosine [zhang 2018] and adenine [ratel 2006] nucleotides and has been associated with changes in gene expression [lang 2017, zhang 2006, lei 2015], chromosome interactions [grob 2014, feng 2014], and genome stability through the repression of transposable elements [mirouze 2009, tsukahara 2009, la 2011]. The modification can be further characterized as either 5-methylcytosine (5mC) or 5-hydroxymethylcytosine (5hmC) within the context of cytosine methylation. These subgroups may well have contrasting epigenetic functions, though, in Arabidopsis thaliana, at least no appreciable level of 5hmC has been observed in genomic DNA [erdmann 2015]. Methylation of cytosine residues is not a base modification that is limited to genomic DNA. However, the fraction of 5hmC present in RNA may be much higher [huber 2015]. Though, the topic of 5hmC in plants is debated among researchers heavily (Mahmood et al. 2019).

The underlying basis for the perceived emphasis on DNA cytosine methylation is due to the development of bisulfite sequencing as a means for studying epigenetics. Since its conception and initial application in 1992 by Frommer et al. [frommer 1992], the method has received much attention for its capacity to resolve DNA methylation patterns at the nucleotide level. This allows researchers to study the effect of differential methylation between organisms, tissues, or cell types on specific genomic elements such as gene bodies, promoter regions, or other regulatory motifs. This, in turn, provides a roadmap for linking epigenetics to gene expression, heritability, or the activation of particular genes or transposons.

The basis for the method is the usage of sodium bisulfite. This chemical compound catalyzes the hydrolytic deamination of cytosine to uracil via an intermediary sulfonation step from cytosine to unstable cytosinesulfonate [hayatsu 1970, shapiro 1970]. The loss of the amine group from cytosinesulfonate yields uracilsulfonate, which in turn is desulfonated under basic pH conditions to form uracil (Figure 1). Though principally also methylated cytosines can react with sodium bisulfite, the presence of a methyl group on position five of the aromatic ring inhibits the process by an order of two magnitudes [hayatsu 1979]. This inhibition is sufficient to confer selectivity for the conversion of unmethylated cytosines. Unfortunately, this selectivity does not extend to a measurable differentiation between 5mC and 5hmC, which are therefore indistinguishable during the regular application of this method. During the sequencing processes or preceding PCR reactions, uracil positions convert on the newly synthesized DNA fragment to thymine so that the sequencing reads contain the four DNA bases.

Figure 1 Bisulfite conversion of unmethylated Cytosin to Uracil.

Once a given DNA sequence has undergone treatment with sodium bisulfite, any remaining cytosines present in the sequence can be inferred as methylated positions. The problem then progresses to the question of mapping these sequences back to their original location on the genome. As is often the case with next-generation sequencing, it is very difficult to achieve a singular, continuous strand of DNA representing an entire chromosome. DNA stability is such that the molecules are easily fragmented during library preparation, and the sequencing approach itself is often restrictive in terms of the length of DNA that can be sequenced in a single iteration. The fragmentation is further confounded by the harsh bisulfite treatment, which undermines long-read sequencing technologies such as PacBio [Eid 2009, Uemura 2010] or Nanopore [Cherf 2012, Mikheyev 2014] and reduces the cost-benefit ratio for using them.

The dominating circumstance is one where the sequencing data consists of many, short bisulfite-converted DNA sequences that subsequently need to be aligned to a reference genome to determine the relationship of methylated positions with nearby genomic elements. In this regard, it is imperative that a high-quality reference genome already exists for the species of interest. Any improvements made to the existing genome assembly contribute towards mapping precision and reducing any false correlations between methylation patterns and nearby loci.

However, it is not a simple task to map bisulfite-treated reads to the reference genome as it would be performed with DNA-Seq data. The challenge arises from the many unmethylated cytosines, which have since been converted to thymines. The fraction of methylated cytosines varies much between different plant species, with as few as ~5% of total cytosines methylated in A. thaliana [Leutwiler 1984] to as many as ~33%, for example, in cultivated tobacco, Nicotiana tabacum [Wagner 1981]. Taking A. thaliana as an example, with a reported GC content of 41.4% [Leutwiler 1984], this could mean that almost ~20% of the total sequence content on one strand is artificially replaced with thymines where before, they were cytosines. In a standard read alignment procedure, this would result in many differences in the alignment of reads to the reference sequence due to the high fraction of C-T mismatches occurring in place of C-C matches. It would not be enough to specify in the algorithm to treat all cytosines and thymines unanimously because the bisulfite treatment dictates that only thymines present in the sequencing reads could potentially match cytosines in the reference genome. Vice versa, it does not follow conceptually that thymines in the reference genome might be able to match cytosines in the reads. Therefore treating the two bases equally would still generate many differences during read alignment.

As standard alignment tools are often not designed with this base-matching asymmetry in mind, the solution is to make further adaptations to these tools or operate specifically with algorithms designed to handle bisulfite data. As mentioned earlier, several specific alignment algorithms for bisulfite data exist. Prominent among them are Bismark [Krueger 2011], BWA-meth [Pedersen 2014], relying on popular alignment algorithms, and Segemehl [Otto 2012] or ERNE [Prezza 2012], with new specific algorithms. Despite the heuristics involved in adapting conventional software (see Read Alignment), recent benchmarks have shown that software such as BWA-meth compete favorably both in terms of precision-sensitivity and overall computational performance [Nunn et al 2021]. Most tools perform capably in non-repetitive regions, but some indeed struggle more-so than others in repetitive, hard-to-align loci. In some cases it may even be the case that the increased base complexity conveyed by methylated nucleotides under bisulfite sequencing may lead to bias in alignment, which can have downstream consequences for methylation calling. Further consideration should always be given to unique aspects of each tool in the context of the planned study. For example, BWA-meth defines mapping quality values differently from Bismark and this too can have implications for downstream analyses such as variant calling.

Another problem that arises during bisulfite sequencing is the loss of variant information following the sodium bisulfite treatment. A standard sequencing experiment identifies polymorphisms where most reads overlapping a single nucleotide position on the reference genome indicate a deviation from the original base in that position. Single nucleotide polymorphisms (SNPs) are used in genotyping to identify which samples belong to different variants or strains. During epigenetic studies, it is often of crucial importance that genetic variation is kept to a minimum to reduce confounding genetic effects. However, any true SNPs in the C-T context are obscured by the artificially converted bases. It is possible to retrieve this variant information by comparing the converted bases to their complementary bases on the opposite strand. Only true SNPs should have the correct sequence complement. But this is computationally intensive and yet to be implemented efficiently. Analyzing the strands separately in this way would theoretically imply that twice as much sequencing coverage would be necessary compared to normal genotyping to achieve a similarly credible result.

Following a successful read alignment, methylation calling can be performed similarly to variant calling but without the need for a complex model. Instead, each cytosine position is extracted from the alignment and the methylation level determined by majority voting. The ratio of reads with either cytosine or thymine in that position is used to calculate an overall methylation percentage or rate. The total collection of positions can be further subset into different genomic sequence contexts such as CG, CHG, or CHH, where H can be either A, T, or G (see Chapter DNA Methylation). This subsetting later allows for downstream analyses of methylation patterns.

Once sequencing has taken place, the question of identifying DNA methylation is effectively reframed to a computational concern. Like standard sequencing, the basic workflow involves initial quality control (QC) of the generated raw reads, followed by mapping these reads to a reference genome to produce alignment files that are the basis of downstream analyses. Methylated positions can then be extracted from these files in a much similar manner to variant calling. Bisulfite sequencing, however, presents its own challenges, particularly during the read alignment step. This section explores common issues and significant divergences from standard practices in bioinformatics.

Quality Control

Like all reads generated via sequencing by synthesis, bisulfite reads are subject to a drop in quality towards the 3’-end due to the propensity of base-calling errors to accumulate following failures in the synthesis process. During a single cycle of Illumina sequencing, the next base in the read is incorporated into the template strand together with a reversible terminator molecule containing a fluorescent tag. The terminator prevents the following base from being incorporated, so the sequencer can read the color of the fluorescent tag and identify the current base. The molecule is then cleaved to facilitate the next cycle that repeats the process for the next base in the sequence. In these synthesis cycles, it can happen that the terminator molecule is either not cleaved or cleaved too early so that two bases are incorporated during a single cycle. If such an error occurs, the strand is out of sync compared to the other strands of the cluster for all remaining cycles and makes it more difficult for the imager to assess the correct base color within the sequence cluster. The consequence is a quality drop for the complete cluster. This phenomenon is known as phase-shifting and is usually corrected by trimming some bases that fall below a quality threshold (eg. phred score < 20) from the 3’-end of a read, limiting the negative effect on the previous correctly sequenced bases.

Another commonly-encountered issue is the tendency to sequence into the adaptor sequence on DNA fragments smaller than the total number of Illumina cycles [Illumina bulletin]. These sequence subsets are not part of the original DNA, making it much more difficult to map such reads to their true location on the reference genome. Fortunately, the adaptors are synthetically designed, distinct sequences which are therefore known and can thus be readily identified [Illumina technical document]. By considering the overlap of the known adaptor sequence at the 3’-end on each read, the reads can again be trimmed to remove the DNA that is not part of the original sequence.

These common problems can be identified with a standard QC tool such as FastQC [Andrews 2010], and frequently occur in standard sequencing and bisulfite sequencing data. As such tools (i.e., FastQC) are usually not designed for bisulfite sequencing, they may also flag errors such as unbalanced base proportions with as high as ~50% thymine content in read one (or adenine content in read two). If dealing with RRBS data, which has been digested with MspI, there is also the possibility that non-random sequence content is flagged at the 5’-ends of reads, as digested fragments always start with a C base. So long as there is confidence that the standard precautions were taken during library preparation, these warnings can be safely ignored at this stage.

The other facet of using standard QC tools that are not designed for bisulfite sequencing is the tendency to miss bisulfite-related sequencing problems. One such problem can occur during the initial DNA fragmentation step of the library preparation procedure, which often leaves protruding 5' and 3’-ends that must be restored to double-stranded DNA by a process known as overhang end-repair [Poptsova 2014]. The incorporation of unmethylated cytosines during this step can introduce artificially low methylation rates at each end of the DNA fragment, which cannot be detected in standard QC [Lin 2013]. Another such issue is thought to occur due to the re-annealing of single-strand sequences adjacent to the methylated sequencing adaptors during MethylC-Seq, which partially restores double-strandedness thereby affording a measure of protection from the bisulfite treatment [Lin 2013]. Therefore, there is a tendency for bisulfite conversion failure to be enriched towards the 5’-end of reads, leading to artificially high methylation rates. BS-Seq and PBAT libraries should theoretically avoid this bias due to the adaptor-ligation occurring after the bisulfite treatment.

As neither of these issues causes changes to the DNA sequence, they can only be detected once methylation calling has occurred (after read alignment). The standard procedure is to look at the total distribution of methylation levels across the average length of the reads in an approach known as M-bias analysis [Hansen 2012, Lin 2013]. A uniform distribution is expected across the read length, but spikes in methylation level can be observed at each end of the distribution. As the sequence information at each end remains unaffected, they should be used for the alignment and not be clipped similarly to quality trimming for repeated read alignment. Instead, the start and end positions of reads are "masked" from a follow-up repeat of the methylation calling procedure, depending on the deviation of their methylation status from the uniform distribution. It should be noted that a significant variation of read lengths reduces the accuracy of this step.

Read Alignment

When all quality concerns are sorted out, the next step in the workflow is to align these reads to an appropriate high-quality reference genome. If a reference genome is not available, de novo assembly is first required before any methylation information can be retrieved. The availability of a good reference genome is fundamental to DNA methylation analysis, when the project aims to relate methylated positions to nearby annotations such as gene bodies, promoter regions, or transposable elements. The higher the genome quality and relatedness of the genome to the test sample, the more confidence you can have in the study's findings [Mardis 2002]. Unfortunately, this degree of confidence is not something that can be measured statistically. Therefore every effort should be made to ensure the validity of your reference genome prior to analysis.

With standard sequencing data, mapping typically involves the use of dynamic programming to determine the best alignment for a given read according to a scoring matrix. Positive scores are given for base matches, or certain types of mismatches, whereas penalties are given for other mismatches and positions where insertions or deletions (indels) are present. The cumulative score is then compared to other potential alignments above a set threshold, and in most cases, only the best one is selected as the most likely point of origin for the read.

However, mapping bisulfite-treated DNA presents a challenge. The majority of cytosine positions on the reference genome are likely unmethylated in the test sample [Wagner 1981, Leutwiler 1984] and therefore represented as thymines in the reads following bisulfite conversion. Aligning these reads results in many C-T mismatches, which negatively influence the scoring matrix and significantly inhibit successful read mapping (Figure 3). It is not enough to simply allow for a higher number of errors, as this would only obscure the correct alignments through an increased number of false positives.

Figure 3 Alignment missmatches with the reference genome resulting from bisulfite conversion.

To further complicate the issue, bisulfite conversion of DNA fragments results in two strands that are no longer complementary to each other. In paired-end sequencing, this means that four distinct sequences are now present, each one varying to some degree from the original DNA (Figure 4). From the original, untreated Watson (+) strand, the first mate pair is the direct bisulfite-converted variant (+FW), and the second is the reverse complement of this (+RC). From the original, untreated Crick (-) strand, the first mate pair is the direct bisulfite-converted variant (-FW), and the second is the reverse complement of this (-RC). In standard DNA sequencing, it is simply the case that the second mate-pair obtained from one strand aligns to the other strand, but in bisulfite sequencing, this no longer holds. In BS-Seq libraries, this is compounded further because the method already encompasses all four-strand variants, even in single-end sequencing. In this case, the directionality relative to the strand (indicated by the box arrows in Figure 4) is thus lost, which is why it is sometimes referred to as an unstranded bisulfite sequencing protocol.

Figure 4 Strand specific point mutations in the new synthesized strand resulting from bisulfite conversion of unmethylated cytosins.

One potential solution for this alignment problem is to adjust the scoring matrix so that a mismatch of thymine to cytosine is instead treated as a match. This can be implemented in standard sequence aligners by "collapsing" the genetic alphabet in both the read and the reference genome so that all cytosines are rewritten as thymines (Figure 5a). Mapping is then performed normally, and the methylated positions are retrieved through post-processing based on the composition of the pre-collapsed sequences. However, this procedure results in two undesirable scenarios that make no sense conceptually: 1) true thymines in the read match with cytosines in the reference genome, and 2) cytosines from the read (indicating methylated positions) match thymines in the reference genome. Such a solution undoubtedly produces many false positives and obscures the correct read alignments. Bisulfite read aligners such as Bismark [Kreuger 2011], BSmooth [Hansen 2012] (in bowtie2 mode), BS-Seeker [Chen 2010, Guo 2013], and BWA-meth [Pedersen 2014] follow this strategy.

A better strategy would be to allow matches between thymines and cytosines, but only between read-based thymines and reference-based cytosines (not vice versa). In this case, methylated cytosines are correctly considered mismatched with thymines in the reference genome, thus reducing false positives (Figure 5b). However, it is still possible for true read-based thymines to match incorrectly with reference-based cytosines (see scenario 1 in the last paragraph). This asymmetric base scoring is not easy to implement in most index structures (e.g., Burrows-Wheeler transform, suffix arrays) used in standard sequence aligners. Therefore specialized read alignment software is required that is explicitly designed for bisulfite sequencing. Such tools include BSMAP [Xi 2009], BSmooth [Hansen 2012] (in merman mode), and ERNE-BS5 [Prezza 2012]. The specialized read aligner segemehl [Otto 2012] uses collapsing (Figure 5a) during a starting step and then changes to asymmetric matching (Figure 5b) in the following. A common drawback of these methods is the increased memory consumption and processing time relative to tools that rely on a collapsed alphabet.

Figure 5 Collapsed alphabet versus assymmetric matchning.

Whichever approach is followed, the entire process likely has to be repeated to account for both C-T conversions and G-A conversions due to the aforementioned loss of complementarity between a given sequence complement and the opposite strand (figure). The possibility of four distinct sequences in bisulfite sequencing, as opposed to two in standard sequencing, dictates that two distinct variants of the reference genome are required to resolve the best alignments. These two alignment procedures may either be run in parallel, as is the case in Bismark [Kreuger 2011], or consecutively as is the case in segemehl [Otto 2012]. Still, the point is that they are interlinked with each other and treated as one.

One problem that might arise during sequencing is the potential for genomic rearrangement [Saxena 2014] that may have occurred in the test sample relative to the reference genome due to evolution. Particularly with reads that originate from a locus that has translocated to the opposite strand. In this case, the strand-specificity is inverted, and the aligner may attempt to map the read to the wrong strand. These false-stranded reads can be detected by the high proportion of G-A mismatches on the Watson (+) strand or C-T mismatches on the Crick (-) strand. In practical terms, a threshold of 3-5% regarding the length of the read is usually enough to identify false-stranded reads. As there is a high probability that these reads originate elsewhere in the genome, they are usually excluded from the alignment because it is not a trivial task to infer where the locus may have translocated to [Onishi-Seebacher 2011].

Filtering can also be applied based on several other factors, in a study-dependant manner, prior to methylation calling. Any spike-in of Lambda or Phi X DNA can be filtered easily in the case of multiplexing with alternative sequencing adaptors, based on the different sequencing tags or it is not indexed at all and therefore automatically filtered out during de-multiplexing. In the case of Lambda, the alignment indexes must be generated to contain both the sample genome and the Lambda phage genome. Any read alignments to the Lambda genome can therefore be filtered out. When splitting the alignment files in this manner, it is essential to remove any reference to the Lambda genome from the file header if there is any intention to view the alignment file with a genome browser (e.g., IGV, JBrowse).

Finally, filtering based on multi-mapped reads or PCR duplicates may also be considered, like in standard sequencing experiments. Any given read should conceptually originate from a single position on the genome. Still, equally-scoring alignments may be possible, particularly in highly repetitive regions or regions of low complexity [Treangen 2011]. If each of these alignments is considered separately during methylation calling, then the statistical power is skewed, especially if the methylation is not the same between these regions. The options are to exclude these alignments entirely, as is performed intrinsically by some sequence aligners [Krueger 2011, Guo 2013], to select one such alignment at random, or to accept the reduction in statistical power to retain the information from that read. Regarding PCR duplicates, several tools exist to identify such reads that arise from a single DNA fragment, such as Picard MarkDuplicates [Broad Institute] and samtools rmdup [Li 2009]. These tools identify PCR duplicates based on the proportionally higher likelihood that identical reads arise from PCR, then that they are separate fragments. In this case, such reads are counted only once during methylation calling. However, deduplication works only in WGBS sequencing projects. If PCR is absent or negligible during library preparation, however, this step should be avoided.

Methylation Calling

To detect the level of DNA methylation at any given cytosine position within the test sample, the reads overlapping that position are evaluated to give the proportion of the coverage that is methylated bases (cytosine) versus the sum of methylated bases (cytosine) and unmethylated bases (thymine). In practice, there is only one major divergence in the implementation of this method by different tools: what to do with C-T polymorphisms that arise before bisulfite treatment has taken place. Most tools simply ignore SNPs. If the consensus indicates anything other than a cytosine or a thymine in a cytosine position on the reference genome, then the entire position is often excluded from analysis. It is not necessarily accurate to consider such an SNP as evidence against DNA methylation, particularly in the case of adenine which may also be methylated [Ratel 2006].

A more robust analysis will recognise also that not all read-based thymines overlapping reference-based cytosine positions are representative of the bisulfite treatment [Liu 2012]. A mutation in this context is deceiving, as we might interpret it as demethylation resulting from an epigenetic mechanism rather than a genetic one. The only way to identify such a mutation from the data itself, rather than from independent genotyping of the test sample, is to compare the overlapping reads in that position to those on the opposite strand. The DNA strands after treatment with sodium bisulfite are no longer complementary to each other since the methylation information is strand-specific. Artificially converted bases should therefore always lack a complementary base on the opposite strand, whereas a C-T mutation will have support on the opposing strand. This difference can be used to exclude such base positions from the analysis; the independent testing of both strands in this manner however would suggest that twice the coverage be required to achieve a similar degree of confidence to standard genotyping.

Consideration should also be made when using PE sequencing to regions that overlap between read pairs. Generally each read pair will start from opposing ends of a single DNA fragment, and it is possible in cases where the fragment size is less than double the number of cycles during Illumina sequencing that a part of the fragment will be sequenced twice for the same read pair [Magoč 2011]. In this case, methylation information in this overlap region is redundant and does not constitute an independent observation of the methylation status as if they arose from separate DNA fragments. It is therefore wise to identify overlapping regions prior to methylation calling and mask those bases from either one of the two reads of a pair.

Although the use of whole genome bisulfite sequencing to analyse DNA methylation in CG, CHG, and CHH context is highly relevant to the study of epigenetics in plant ecology, it is not all-encompassing. There are indeed other techniques based on similar principles that can be used to capture methylation information, such as methylated DNA immunoprecipitation (meDIP) which can be used in combination with high-resolution DNA microarrays or high-throughput next-generation sequencing.

Recently there has also been some investigation into treatments that can facilitate nucleotide-level base conversion without the harsh side effects of sodium bisulfite [Yibin et al. 2019]. The advent of long-read sequencing technologies such as single molecule real-time (SMRT) analysis with PacBio or Nanopore has also provided an alternative to bisulfite sequencing. The base calling in these approaches can provide wavelength profiles for each nucleotide that differ between bases with and without base modifications [Flusberg et al. 2010]. This has the advantage of detecting DNA methylation without the need for harsh bisulfite treatment, while also allowing for detection of other forms of base modification. Unfortunately the profiles can be difficult to differentiate, but development of machine learning techniques may be a promising avenue of advancement in this regard.

To start using the EpiDiverse analysis pipelines, follow the steps below:

1) Install Nextflow

Nextflow runs on most POSIX systems (Linux, Mac OSX etc). It can be installed by running the following commands:

# Make sure that Java v8+ is installed:
java -version

# Install Nextflow v19.09+
curl -fsSL get.nextflow.io | bash

# Add Nextflow binary to your $PATH:
mv nextflow ~/bin
# OR system-wide installation:
# sudo mv nextflow /usr/local/bin

2) Install the pipeline

2.1) Automatic

The pipelines themselves need no installation - Nextflow will automatically fetch them from GitHub if eg. epidiverse/wgbs is specified as the pipeline name.

2.2) Offline

The above method requires an internet connection so that Nextflow can download the pipeline files. If you're running on a system that has no internet connection, you'll need to download and transfer the pipeline files manually using the following (pseudo)code:

# Download the latest release of the pipeline
# eg. (see https://github.com/epidiverse/wgbs/releases)
curl -L https://github.com/epidiverse/[PIPELINE]/archive/[VERSION].zip -o epidiverse-[PIPELINE]-[VERSION].zip
unzip epidiverse-[PIPELINE]-[VERSION].zip
cd /path/to/my/data
nextflow run /path/to/pipelines/epidiverse-[PIPELINE]-[VERSION] [PARAMETERS]

2.3) Development

If you would like to make changes to the pipeline, it's best to make a fork on GitHub and then clone the files. Once cloned you can run the pipeline directly as above.

3) Pipeline configuration

By default, the pipelines run with the -profile standard configuration profile. This uses a number of sensible defaults for process requirements and is suitable for running on a simple (if powerful!) basic server. You can see this configuration in conf/base.config from the base directory of each pipeline repository.

Be warned of two important points about the default configuration:

1. The default profile uses the local executor

   • All jobs are run in the login session. If you're using a simple server, this may be fine. If you're using a compute cluster, this is bad as all jobs will run on the head node.
2. Nextflow will expect all software to be installed and available on the $PATH

3.1) Configuration profiles

Nextflow can be configured to run on a wide range of different computational infrastructures. In addition to pipeline-specific parameters it is likely that you will need to define system-specific options.

Whilst most parameters can be specified on the command line, it is usually sensible to create a configuration file for your environment. A template for such a config can be found in assets/custom.config from the base directory of each pipeline repository.

If you are the only person to be running this pipeline, you can create your config file as ~/.nextflow/config and it will be applied every time you run Nextflow. Alternatively, save the file anywhere and reference it when running the pipeline with -config /path/to/config.

If you think that there are other people using the pipeline who would benefit from your configuration (eg. other common cluster setups), please let us know. We can add a new preset configuration profile which can used by specifying -profile <name> when running the pipeline.

The pipelines already come with several such config profiles - see the installation appendices and usage documentation for more information.

3.2) Software dependencies: bioconda

If you're unable to use either Docker or Singularity but you have conda installed, you can use the bioconda environment that comes with the pipeline. Using the predefined -profile conda configuration when running the pipeline will take care of this automatically.

If you prefer to build your own environment, running this command will create a new conda environment with all of the required software installed:

conda env create -f environment.yml     
conda clean -a                          # Recommended, not essential
conda activate wgbs                     # Name depends on version

The env/environment.yml file can be found from the base directory of the pipeline repository. Note that you may need to download this file from the GitHub project page if Nextflow is automatically fetching the pipeline files. Ensure that the bioconda environment file version matches the pipeline version that you run.

3.3) Software dependencies: Docker and Singularity

If you prefer to use your own container, running the pipeline with the option -with-singularity <container> or -with-docker <container> and pointing towards a specific image will allow it to be automatically fetched and used.

If running offline with Singularity, you'll need to download and transfer the Singularity image first:

singularity pull --name epidiverse-[PIPELINE]-[VERSION].simg docker://epidiverse/[PIPELINE]:[VERSION]

Once transferred, use -with-singularity but specify the path to the image file:

nextflow run /path/to/epidiverse/[PIPELINE] -with-singularity /path/to/epidiverse-[PIPELINE]-[VERSION].simg

Appendices

Running on EPIDIVERSE infrastructure

There are also three shortcuts available for EpiDiverse species which can be used in place of --reference in pipelines that require a reference genome.

• --thlaspi

• --fragaria

• --populus

Adapter trimming: Why are adapter sequences trimmed from only the 3' ends of reads? (2018). Retrieved January 24, 2019, from https://support.illumina.com/bulletins/2016/04/adapter-trimming-why-are-adapter-sequences-trimmed-from-only-the--ends-of-reads.html

Aird, D., Ross, M. G., Chen, W. S., Danielsson, M., Fennell, T., Russ, C., … Gnirke, A. (2011). Analyzing and minimizing PCR amplification bias in Illumina sequencing libraries. Genome Biology, 12(2), R18. https://doi.org/10.1186/gb-2011-12-2-r18

Benjamini, Y., & Speed, T. P. (2012). Summarizing and correcting the GC content bias in high-throughput sequencing. Nucleic Acids Research, 40(10), 1–14. https://doi.org/10.1093/nar/gks001

Bentley, D. R., Balasubramanian, S., Swerdlow, H. P., Smith, G. P., Milton, J., Brown, C. G., … Smith, A. J. (2008). Accurate whole human genome sequencing using reversible terminator chemistry. Nature, 456, 53–49. https://doi.org/10.1038/nature07517

Bormann Chung, C. A., Boyd, V. L., McKernan, K. J., Fu, Y., Monighetti, C., Peckham, H. E., & Barker, M. (2010). Whole methylome analysis by ultra-deep sequencing using two-base encoding. PLoS ONE, 5(2). https://doi.org/10.1371/journal.pone.0009320

Chen, P.-Y., Cokus, S. J., & Pellegrini, M. (2010). BS Seeker: precise mapping for bisulfite sequencing. BMC Bioinformatics, 11(203), 2–7. https://doi.org/10.1186/1471-2105-11-203

Cherf, G. M., Lieberman, K. R., Rashid, H., Lam, C. E., Karplus, K., & Akeson, M. (2012). Automated Forward and Reverse Ratcheting of DNA in a Nanopore at Five Angstrom Precision. Nature Biotechnology, 30(4), 344–348. https://doi.org/10.1038/nbt.2147.Automated

Cokus, S. J., Feng, S., Zhang, X., Chen, Z., Merriman, B., Haudenschild, C. D., … Jacobsen, S. E. (2008). Shotgun bisulfite sequencing of the Arabidopsis genome reveals DNA methylation patterning. Nature, 452(7184), 215–219. https://doi.org/10.3390/ijms151222874

Eid, J., Fehr, A., Gray, J., Luong, K., Lyle, J., Otto, G., … Dewinter, A. (2009). Real-Time DNA sequencing from Single Polymerase Molecules. Science, 323(January), 133–138. https://doi.org/10.1126/science.1162986

Erdmann, R. M., Souza, A. L., Clish, C. B., & Gehring, M. (2015). 5-Hydroxymethylcytosine Is Not Present in Appreciable Quantities in Arabidopsis DNA. G3: Genes, Genomes, Genetics, 5(1), 1–8. https://doi.org/10.1534/g3.114.014670

Feng, S., Cokus, S. J., Schubert, V., Zhai, J., Pellegrini, M., & Jacobsen, S. E. (2014). Genome-wide Hi-C Analyses in Wild-Type and Mutants Reveal High-Resolution Chromatin Interactions in Arabidopsis. Molecular Cell, 55(5), 694–707. https://doi.org/10.1016/j.molcel.2014.07.008

Flusberg, B. A., Webster, D. R., Lee, J. H., Travers, K. J., Olivares, E. C., Clark, T. A., ... & Turner, S. W. (2010). Direct detection of DNA methylation during single-molecule, real-time sequencing. Nature methods, 7(6), 461-465.

Fojtová, M., Kovařík, A., & Matyášek, R. (2001). Cytosine methylation of plastid genome in higher plants. Fact or artefact? Plant Science, 160(4), 585–593. https://doi.org/10.1016/S0168-9452(00)00411-8

Frommer, M., McDonald, L. E., Millar, D. S., Collist, C. M., Watt, F., Griggt, G. W., … Paul, C. L. (1992). A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proceedings of the National Academy of Sciences, 89(March), 1827–1831. https://doi.org/10.1073/pnas.89.5.1827

Greagg, M., Fogg, M. J., Panayotou, G., Evans, S. J., Connolly, B. A., & Pearl, L. H. (1999). A read-ahead function in archaeal DNA polymerases detects promutagenic template-strand uracil. Proceedings of the National Academy of Sciences, 96(16), 9045–9050. https://doi.org/10.1073/pnas.96.16.9045

Grehl, C., Kuhlmann, M., Becker, C., Glaser, B., & Grosse, I. (2018). How to Design a Whole-Genome Bisulfite Sequencing Experiment. Epigenomes, 2(21), 1–11. https://doi.org/10.3390/epigenomes2040021

Grob, S., Schmid, M. W., & Grossniklaus, U. (2014). Hi-C Analysis in Arabidopsis Identifies the KNOT, a Structure with Similarities to the flamenco Locus of Drosophila. Molecular Cell, 55(5), 678–693. https://doi.org/10.1016/j.molcel.2014.07.009

Guo, W., Fiziev, P., Yan, W., Cokus, S. J., Sun, X., Zhang, M. Q., … Pellegrini, M. (2013). BS-Seeker2: A versatile aligning pipeline for bisulfite sequencing data. BMC Genomics, 14(1). https://doi.org/10.1186/1471-2164-14-774

Hansen, K. D., Brenner, S. E., & Dudoit, S. (2010). Biases in Illumina transcriptome sequencing caused by random hexamer priming. Nucleic Acids Research, 38(12), 1–7. https://doi.org/10.1093/nar/gkq224

Hansen, K. D., Langmead, B., & Irizarry, R. A. (2012). BSmooth: from whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biology, 13(10), 1–10. https://doi.org/10.1186/gb-2012-13-10-r83

Hattman, S., Schlagman, S., & Cousens, L. (1973). Isolation of a mutant of Escherichia coli defective in cytosine specific deoxyribonucleic acid methylase activity and in partial protection of bacteriophage ?? against restriction by cells containing the N 3 drug resistance factor. Journal of Bacteriology, 115(3), 1103–1107. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC246359/pdf/jbacter00349-0397.pdf

Hayatsu, H., & Shiragami, M. (1979). Reaction of Bisulfite with the 5-Hydroxymethyl Group in Pyrimidines and in Phage DNAs. Biochemistry, 18(4), 632–637. https://doi.org/10.1021/bi00571a013

Hayatsu, H., Wataya, Y., Kai, K., & Iida, S. (1970). Reaction of sodium bisulfite with uracil, cytosine, and their derivatives. Biochemistry, 9(14), 2858–2865. https://doi.org/10.1021/ja00706a062

Hodges, E., Smith, A., Kendall, J., Xuan, Z., Ravi, K., Rooks, M., … Hicks, J. (2009). High definition profiling of mammalian DNA methylation by array capture and single molecule bisulfite sequencing. Genome Research, 19(9), 1593. Retrieved from https://genome.cshlp.org/content/early/2009/07/06/gr.095190.109.full.pdf

Huber, S. M., Van Delft, P., Mendil, L., Bachman, M., Smollett, K., Werner, F., … Balasubramanian, S. (2015). Formation and abundance of 5-hydroxymethylcytosine in RNA. ChemBioChem, 16(5), 752–755. https://doi.org/10.1002/cbic.201500013

Illumina Adapter Sequences. (2018). San Diego, California. Retrieved from https://support.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/experiment-design/illumina-adapter-sequences-1000000002694-09.pdf

Kint, S., De Spiegelaere, W., De Kesel, J., Vandekerckhove, L., & Van Criekinge, W. (2018). Evaluation of bisulfite kits for DNA methylation profiling in terms of DNA fragmentation and DNA recovery using digital PCR. PLoS ONE, 13(6), e0199091. https://doi.org/10.1371/journal.pone.0199091

Kozarewa, I., Ning, Z., Quail, M. a, Sanders, M. J., & Turner, D. J. (2009). Amplification-free Illumina sequencing-library preparation facilitates improved mapping and assembly of GC-biased genomes. Nature Methods, 6(4), 291–295. https://doi.org/10.1038/nmeth.1311.Amplification-free

Krueger, F., & Andrews, S. R. (2011). Bismark: A flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics, 27(11), 1571–1572. https://doi.org/10.1093/bioinformatics/btr167

La, H., Ding, B., Mishra, G. P., Zhou, B., Yang, H., Bellizzi, M. del R., … Wang, G.-L. (2011). A 5-methylcytosine DNA glycosylase/lyase demethylates the retrotransposon Tos17 and promotes its transposition in rice. Proceedings of the National Academy of Sciences, 108(37), 15498–15503. https://doi.org/10.1073/pnas.1112704108

Laehnemann, D., Borkhardt, A., & McHardy, A. C. (2016). Denoising DNA deep sequencing data-high-throughput sequencing errors and their correction. Briefings in Bioinformatics, 17(1), 154–179. https://doi.org/10.1093/bib/bbv029

Lang, Z., Wang, Y., Tang, K., Tang, D., Datsenka, T., Cheng, J., … Zhu, J.-K. (2017). Critical roles of DNA demethylation in the activation of ripening-induced genes and inhibition of ripening-repressed genes in tomato fruit. Proceedings of the National Academy of Sciences, 114(22), E4511–E4519. https://doi.org/10.1073/pnas.1705233114

Langmead, B., & Salzberg, S. L. (2013). Fast gapped-read alignment with Bowtie 2. Nature Methods, 9(4), 357–359. https://doi.org/10.1038/nmeth.1923.Fast

Lei, M., Zhang, H., Julian, R., Tang, K., Xie, S., & Zhu, J.-K. (2015). Regulatory link between DNA methylation and active demethylation in Arabidopsis. Proceedings of the National Academy of Sciences, 112(11), 3553–3557. https://doi.org/10.1073/pnas.1502279112

Leutwiler, L. S., Hough-Evans, B. R., & Meyerowitz, E. M. (1984). The DNA of Arabidopsis thaliana. MGG Molecular & General Genetics, 194(1–2), 15–23. https://doi.org/10.1007/BF00383491

Li, F. W., & Harkess, A. (2018). A guide to sequence your favorite plant genomes. Applications in Plant Sciences. https://doi.org/10.1002/aps3.1030

Li, H., & Durbin, R. (2010). Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics, 26(5), 589–595. https://doi.org/10.1093/bioinformatics/btp698

Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., … Durbin, R. (2009). The Sequence Alignment/Map format and SAMtools. Bioinformatics, 25(16), 2078–2079. https://doi.org/10.1093/bioinformatics/btp352

Lin, X., Sun, D., Rodriguez, B., Zhao, Q., Sun, H., Zhang, Y., & Li, W. (2013). BSeQC: Quality control of bisulfite sequencing experiments. Bioinformatics, 29(24), 3227–3229. https://doi.org/10.1093/bioinformatics/btt548

Lindahl, T., & Wood, R. D. (1999). Quality control by DNA repair. Science, 286(5446), 1897–1905. https://doi.org/10.1126/science.286.5446.1897

Lister, R., O'Malley, R. C., Tonti-Filippini, J., Gregory, B. D., Berry, C. C., Millar, A. H., & Ecker, J. R. (2008). Highly Integrated Single-Base Resolution Maps of the Epigenome in Arabidopsis. Cell, 133(3), 523–536. https://doi.org/10.1016/j.cell.2008.03.029

Liu, Y., Siegmund, K. D., Laird, P. W., & Berman, B. P. (2012). Bis-SNP: Combined DNA methylation and SNP calling for Bisulfite-seq data. Genome Biology, 13(7). https://doi.org/10.1186/gb-2012-13-7-r61

Liu, Y., Siejka-Zielińska, P., Velikova, G., Bi, Y., Yuan, F., Tomkova, M., ... & Song, C. X. (2019). Bisulfite-free direct detection of 5-methylcytosine and 5-hydroxymethylcytosine at base resolution. Nature biotechnology, 37(4), 424-429.

Magoč, T., & Salzberg, S. L. (2011). FLASH: Fast length adjustment of short reads to improve genome assemblies. Bioinformatics, 27(21), 2957–2963. https://doi.org/10.1093/bioinformatics/btr507

Mahmood, A. M., & Dunwell, J. M. (2019). Evidence for novel epigenetic marks within plants. AIMS genetics, 6(4), 70.

Marano, M. R., & Carrillo, N. (1991). Chromoplast formation during tomato fruit ripening. No evidence for plastid DNA methylation. Plant Molecular Biology, 16(1), 11–19. https://doi.org/10.1007/BF00017913

Mardis, E., McPherson, J., Martienssen, R., Wilson, R. K., & McCombie, W. R. (2002). What is Finished, and Why Does it Matter. Genome Research, 12, 669–671. https://doi.org/10.1101/gr.032102.O

McKernan, K. J., Blanchard, A., Kotler, L., & Costa, G. (2006). 2008/0003571. United States.

Meissner, Alexander, Andreas Gnirke, George W. Bell, Bernard Ramsahoye, Eric S. Lander, and Rudolf Jaenisch. "Reduced representation bisulfite sequencing for comparative high-resolution DNA methylation analysis." Nucleic acids research 33, no. 18 (2005): 5868-5877.

Mikheyev, A. S., & Tin, M. M. Y. (2014). A first look at the Oxford Nanopore MinION sequencer. Molecular Ecology Resources, 14(6), 1097–1102. https://doi.org/10.1111/1755-0998.12324

Mirouze, M., Reinders, J., Bucher, E., Nishimura, T., Schneeberger, K., Ossowski, S., … Mathieu, O. (2009). Selective epigenetic control of retrotransposition in Arabidopsis. Nature, 461(7262), 427–430. https://doi.org/10.1038/nature08328

Miura, F., Enomoto, Y., Dairiki, R., & Ito, T. (2012). Amplification-free whole-genome bisulfite sequencing by post-bisulfite adaptor tagging. Nucleic Acids Research, 40(17). https://doi.org/10.1093/nar/gks454

Nunn, A., Otto, C., Stadler P. F., Langenberger D. (2021). Comprehensive benchmarking of software for mapping whole genome bisulfite data: from read alignment to DNA methylation analysis, Briefings in Bioinformatics, bbab021, https://doi.org/10.1093/bib/bbab021

Onishi-Seebacher, M., & Korbel, J. O. (2011). Challenges in studying genomic structural variant formation mechanisms: The short-read dilemma and beyond. BioEssays, 33(11), 840–850. https://doi.org/10.1002/bies.201100075

Otto, C., Stadler, P. F., & Hoffmann, S. (2012). Fast and sensitive mapping of bisulfite-treated sequencing data. Bioinformatics, 28(13), 1698–1704. https://doi.org/10.1093/bioinformatics/bts254

Pabinger, S., Ernst, K., Pulverer, W., Kallmeyer, R., Valdes, A. M., Metrustry, S., … Weinhaeusel, A. (2016). Analysis and visualization tool for targeted amplicon bisulfite sequencing on ion torrent sequencers. PLoS ONE, 11(7), 1–16. https://doi.org/10.1371/journal.pone.0160227

Pedersen, B., Eyring, K., De, S., Yang, I. V, & Schwartz, D. A. (2014). Fast and accurate alignment of long bisulfite-seq reads (arXiv q-bio). Denver, Colorado. Retrieved from https://arxiv.org/pdf/1401.1129.pdf

Picard Toolkit. (2019). Broad Institute. Retrieved from https://github.com/broadinstitute/picard

Poptsova, M. S., Il’Icheva, I. A., Nechipurenko, D. Y., Panchenko, L. A., Khodikov, M. V, Oparina, N. Y., … Grokhovsky, S. L. (2014). Non-random DNA fragmentation in next-generation sequencing. Scientific Reports, 4, 1–6. https://doi.org/10.1038/srep04532

Prezza, N., Del Fabbro, C., Vezzi, F., De Paoli, E., & Policriti, A. (2012). ERNE-BS5: Aligning BS-treated sequences by multiple hits on a 5-letters alphabet. 2012 ACM Conference on Bioinformatics, Computational Biology and Biomedicine, BCB 2012, (May), 12–19. https://doi.org/10.1145/2382936.2382938

Raine, A., Liljedahl, U., & Nordlund, J. (2018). Data quality of whole genome bisulfite sequencing on Illumina platforms. PLoS ONE, 13(4), 1–15. https://doi.org/10.1371/journal.pone.0195972

Ratel, D., Ravanat, J. L., Berger, F., & Wion, D. (2006, March). N6-methyladenine: The other methylated base of DNA. BioEssays. https://doi.org/10.1002/bies.20342

Rothberg, J. M., Hinz, W., Rearick, T. M., Schultz, J., Mileski, W., Davey, M., … Bustillo, J. (2011). An integrated semiconductor device enabling non-optical genome sequencing. Nature, 475, 348–352. https://doi.org/10.1038/nature10242

Sanger, F., Air, G. M., Barrell, B. G., Brown, N. L., Coulson, A. R., Fiddes, J. C., … Smith, M. (1977). Nucleotide sequence of bacteriophage φX174 DNA. Nature, 265, 687–695. https://doi.org/10.1038/267585a0

Saxena, R. K., Edwards, D., & Varshney, R. K. (2014). Structural variations in plant genomes. Briefings in Functional Genomics, 13(4). https://doi.org/10.1093/bfgp/elu016

Shapiro, R., Servis, R. E., & Welcher, M. (1970). Reactions of Uracil and Cytosine Derivatives with Sodium Bisulfite. A Specific Deamination Method. Journal of the American Chemical Society, 92(2), 422–424. https://doi.org/10.1021/ja00705a626

Shendure, J., Porreca, G. J., Reppas, N. B., Lin, X., McCutcheon, J. P., Rosenbaum, A. M., … Church, G. M. (2005). Accurate multiplex polony sequencing of an evolved bacterial genome. Science, 309, 1728–1732. https://doi.org/10.1126/science.1117389

Šimková, H. (1998). Methylation of mitochondrial DNA in carrot (Daucus carota L.). Plant Cell Reports, 17(3), 220–224. https://doi.org/10.1007/s002990050382

Treangen, T. J., & Salzberg, S. L. (2011). Repetitive DNA and next-generation sequencing: computational challenges and solutions. Nature Reviews Genetics, 13(1), 36–46. https://doi.org/10.1038/nrg3117.Repetitive

Tsukahara, S., Kobayashi, A., Kawabe, A., Mathieu, O., Miura, A., & Kakutani, T. (2009). Bursts of retrotransposition reproduced in Arabidopsis. Nature, 461(17), 423–427. https://doi.org/10.1038/nature08351

Uemura, S., Aitken, C. E., Korlach, J., Flusberg, B. A., Turner, S. W., & Puglisi, J. D. (2010). Real-time tRNA transit on single translating ribosomes at codon resolution. Nature, 464(7291), 1012–1017. https://doi.org/10.1038/nature08925

Vanyushin, B. F., & Kirnos, M. D. (1988). DNA methylation in plants. Gene, 74(1), 117–121. Retrieved from http://zero.sci-hub.se/2223/39b0cdc1e6d65e6ac3b157a88ec31cc7/vanyushin1988.pdf

Venney, C. J., Johansson, M. L., & Heath, D. D. (2016). Inbreeding effects on gene-specific DNA methylation among tissues of Chinook salmon. Molecular Ecology, 25(18), 4521–4533. https://doi.org/10.1111/mec.13777

What is nucleotide diversity and why is it important? (2017). Retrieved January 24, 2018, from https://support.illumina.com/bulletins/2016/07/what-is-nucleotide-diversity-and-why-is-it-important.html

Wreczycka, K., Gosdschan, A., Yusuf, D., Grüning, B., Assenov, Y., & Akalin, A. (2017). Strategies for analyzing bisulfite sequencing data. Journal of biotechnology, 261, 105-115.

Zhang, H., Lang, Z., & Zhu, J.-K. (2018). Dynamics and function of DNA methylation in plants. Nature Reviews Molecular Cell Biology, 19(8), 489–506. https://doi.org/10.1038/s41580-018-0016-z

Zhang, X., Yazaki, J., Sundaresan, A., Cokus, S. J., Chan, S. W. L., Chen, H., … Ecker, J. R. (2006). Genome-wide High-Resolution Mapping and Functional Analysis of DNA Methylation in Arabidopsis. Cell, 126(6), 1189–1201. https://doi.org/10.1016/j.cell.2006.08.003

Ziller, M. J., Hansen, K. D., Meissner, A., & Aryee, M. J. (2015). Coverage recommendations for methylation analysis by whole genome bisulfite sequencing. Nature Methods, 12(3), 230–232. https://doi.org/10.1097/NCN.0b013e3181a91b58.Exploring

Best practice pipeline for processing and aligning whole genome bisulfite sequencing (WGBS) data from non-model plant species, ready for downstream analysis with the other following EpiDiverse pipelines.

Best practice pipeline for calling single nucleotide polymorphism (SNP) variants from bisulfite sequencing data and/or for clustering of eg. environmental samples according to methylation profiles while masking genomic variation.

Best practice pipeline for calling differentially methylated regions (DMRs) or positions (DMPs) in pairwise comparisons between groups of samples.

Best practice pipeline for epigenome-wide association studies (EWAS) from individual sample methylation calls which are optionally filtered according to DMRs or DMPs.

Recommended External Pipelines

A snakemake pipeline for genotyping-by-sequencing using a specialised reduced representation sequencing protocol.

Identification of differentially methylated regions between multiple epigenomes from BS-treated read mappings via methylated region calling.

Singularity issues

Sometimes Singularity runs into problems when pulling multiple images at the same time for a pipeline run. In these instances it is sometimes better to pull the images manually into the directory that the pipeline will be run from, using for instance the following code:

singularity pull --name epidiverse-[PIPELINE]-[VERSION].simg \
docker://epidiverse/[PIPELINE]:[VERSION]

Extra resources and getting help

If you still have an issue with running the pipeline then feel free to contact us at info@epidiverse.eu or by opening an issue in the respective pipeline repository on GitHub asking for help.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 764965