Developmental Plasticity and Evolution
- 246 Downloads
The environment plays a crucial role in the developing organism, first in defining the developmental trajectory from genotype to phenotype, then by modifying that trajectory by natural selection. Nearly all traits exhibit some degree of phenotypic plasticity: the capacity to change, or to develop in response to, the environment. The plasticity of a trait can itself evolve, and some of the most specialized adaptations include evolved responses to environmental variation. Plasticity has long been theorized to potentiate adaptive evolution, by environmental induction of phenotypes that boosts the potential for subsequent genetic evolution or by revealing cryptic alleles in new environments that in turn generate new adaptive phenotypes. A plastic trait may vary continuously, which can be described by norms of reaction, or it may produce discrete types as a polyphenism, a codified adaptive response to specific environmental signals. The concept of plasticity can also be applied to variation in phenotype associated with a single genotype in a single environment. Such microenvironmental plasticity defines in part the robustness of a trait. In the evolution of complex traits, tension between plasticity and its opposite, canalization, may be crucial for rapid evolution, adaptation, and the emergence of novelty.
KeywordsReaction norm Polyphenism GxE Genetic accommodation Genetic assimilation Canalization Evo-devo
In natural systems, the environment plays two roles: first, it mediates how genotypes are translated into phenotypes; second, it imposes selection (West-Eberhard 2003; see also the chapter on “Eco-Evo-Devo”). Together, these functions drive trait evolution, but the role of the environment in the first function can be overlooked even as its dominance is assumed in the second. This oversight may arise due to the explanatory power of genetics and its amenability to controlled laboratory experimentation. However, the ability of organisms to respond plastically to the environment is so ubiquitous that it is an essential component of the living world (Ehrenreich and Pfennig 2016). In many ways, we intuitively understand this: our expectation that the environment will impose influence is folded into many decisions we make, both scientifically and in our everyday lives. Nevertheless, in both mainstream culture and in scientific research, we often turn to genes first for biological explanations. In fact, genetic determinism – which does indeed play a profound role in human health, applied efforts in agriculture, and the evolution of natural populations – cannot be separated from environmental influence. The outcome of a genotype is undefined without specifying the environment. As genotype translates into phenotype, it travels along developmental trajectories that may be labile or robust and may be defined by generations of adaptive evolution or vulnerable to new influences (see the chapter on the “Genotype-Phenotype Map”). In turn, environmental selection pressures shape not only the phenotypes of an organism but the interactions between development and the environment that produces them.
Definitions and Related Concepts
Phenotypic plasticity is the ability of an individual to alter its phenotype in response to the environment, or the potential of an individual genotype to develop into alternative phenotypes in different environments (Fusco and Minelli 2010; Levis and Pfennig 2017). The first scenario can occur when a phenotype is labile over the course of an individual’s lifetime: behavior, for example, or body size or composition, gene expression, or aspects of physiology. Even sex can change, in the case of some trees, polychaetes, gastropods, and fish. The latter scenario occurs when a trait is expressed only once in a lifetime, like the age of reproductive maturity or the shape of a developmentally irreversible bony appendage. For such fixed traits, the critical aspect is that a single genotype holds the potential for different phenotypes, which are determined by the environment during development. One of the most well-studied examples is the formation of defensive “helmets” in Daphnia in the presence of predators (Agrawal et al. 1999).
Most traits are not expressed as discrete types, but instead vary continuously (Fig. 1b). When plotted against variation in the environment, a continuous plastic trait can be represented by a reaction norm (Fig. 1c) (Stearns 1989). (It is worth noting that the distinction between a polyphenism and a continuously variable plastic trait can nevertheless be vague; the concept of a reaction norm was first introduced by Richard Woltereck while working in the early 1900s on helmet length in Daphnia (Simpson et al. 2011).) Plasticity is extremely pervasive, as the environment influences the expression of most traits. Most traits also exhibit nonzero heritability, indicating that phenotypes are almost always affected by genotype as well. Within populations, these two determinants – environment and genotype – influence trait expression such that different genotypes are likely to exhibit different reaction norms (Fig. 1d). When these functions have different slopes and intersect, we observe a crossing of reaction norms. Such nonparallelism is evidence of gene-by-environment interaction (GxE), in which the environment influences trait determination nonadditively across genotypes. For example, a genotype conferring slow metabolism might grow only slightly faster in a high-glucose environment compared to a low-glucose environment, whereas the difference might be dramatic for a genotype conferring fast metabolism. GxE is pervasive in natural populations (Morgante et al. 2015), and this is to be expected: if different genotypes produced parallel reaction norms, one should be fittest in all environments and eventually fix in the population (Stearns 1989). Reaction norms themselves can evolve; this is discussed further in the section “Evolution of Plasticity.”
The opposite of plasticity is canalization (see the chapter on “Canalization”), the production of an invariant phenotype even in a noisy environment (Stearns 1989). The term was coined by Waddington (1957), to illustrate the entrenched grooves or “canals” that developmental trajectories occupy in the formation of morphological features. Waddington conceived of canalization with regard to genetic variation, but now the term is used both ways, often specified as either “environmental canalization” or “genetic canalization” (Wagner et al. 1997). Although opposite in definition, the relationship between plasticity and canalization is intimate: a plastic trait can evolve into a canalized one via changes in the regulation of the trait’s expression (Ehrenreich and Pfennig 2016), and plastic responses can themselves become canalized when responses to an environmental cue become stereotyped, as in a polyphenism. Trait lability and trait robustness may both evolve from the existence of conditionally functional variation, and a major motivation for studying plasticity and canalization is the hypothesis that dynamic tension between these phenomena might enable rapid evolution, adaptation, and the emergence of novel traits (Paaby and Gibson 2016).
The term robustness may be used as a synonym for canalization, though canalization is more often conceived in association with development, as in the ability of a developmental trajectory to withstand perturbations to produce an invariant phenotype (Masel and Siegal 2009; see also the chapter on “Robustness”). If the allelic or epigenetic state of a genic element is responsible for producing either an environmentally canalized phenotype or a plastic phenotype, then this element is considered a phenotypic stabilizer (Masel and Siegal 2009). A phenotypic stabilizer is analogous to a phenotypic capacitor, likewise a genic element with potential to store or release phenotypic variance, but in this case due to genetic, rather than environmental, variation. The switch of a phenotypic capacitor from one state to another, for example the disabling of the heat-shock protein HSP90, can lead to the release of cryptic genetic variation (Geiler-Samerotte et al. 2016; Rutherford and Lindquist 1998). Cryptic genetic variation is standing variation in a population that has little effect on phenotype until a perturbation induces expressivity, either through epistasis, like by the disruption of a phenotypic capacitor, or through GxE, if the perturbation is environmental (Paaby and Rockman 2014; see also the chapter on “Epistasis”).
A major question regarding plasticity in evolutionary biology is the relative importance of a “plasticity-first” mode of trait evolution, in which a plastic response to environmental change precedes and enables adaptive change, compared to a genes-first mode, in which selection acts on new genetic variants first and changes in plasticity are secondary. In other words, since the first step toward a new adaptive phenotype involves the production of a new developmental variant, the question is whether trait evolution more often starts with an environmentally induced variant or a variant induced by mutation (Levis and Pfennig 2016). Three factors support a plasticity-first mechanism: first, environmental change is likely to affect many or all individuals in a population, whereas de novo mutation, by definition, occurs in only one individual; second, since plastic responses are always linked to the environments that induce them, they may be primed to increase fitness; and third, environmental plasticity can promote the accumulation of cryptic alleles, a store of heritable variation that might potentiate rapid genetic adaptation (Levis and Pfennig 2016).
A crucial mechanism by which plasticity probably facilitates trait evolution relies on accumulation of cryptic genetic variation on which selection may eventually act, as exampled by Waddington’s (1953) evolution of the Drosophila crossveinless wing phenotype. Accumulation of cryptic variation is an emergent property of any genetic system with epistasis or GxE, though it may also be facilitated by the evolved stability of developmental trajectories, which can shelter cryptic alleles (Hermisson and Wagner 2004). Once a sufficiently destabilizing environment reveals cryptic variation, selection can target the focal trait but also the environmental dependence of its expression, such that plasticity itself evolves, in the process of genetic accommodation. Waddington’s genetic assimilation experiments demonstrate a complete loss of plasticity because the trait became constitutively expressed, but selection can also promote developmental sensitivity to environmental signals. The most extreme form of sensitization results in the evolution of polyphenism (Levis and Pfennig 2016). Evidence for genetic accommodation in natural systems is mostly indirect – since testing trait expression in a true ancestor is typically impossible – but observations of physiological and behavioral traits in house finches, behavior and morphology in stickleback fish, and melanin production in Daphnia provide compelling support (Moczek et al. 2011). Trait plasticity can theoretically promote evolvability, because the location of a decision point along the developmental path can evolve just as the terminal phenotype of the plastic trait is adaptively refined (see the chapters “Developmental Evolvability” and “Variational Approaches to Evolvability: Short- and Long-Term Perspectives”). This potentially dramatic lability of ontogenetic specification can increase the “evolutionary degrees of freedom” of the system (Moczek et al. 2011).
Despite the still-growing accumulation of observations elucidating how plasticity influences the evolution of natural systems (Levis and Pfennig 2017; West-Eberhard 2003), whether plasticity tends to jump-start evolutionary innovation remains an open question (Moczek et al. 2011). The best case study for this phenomenon involves investigation of spadefoot toads in the genus Spea, by Pfennig and colleagues. Spea tadpoles develop into either small-jawed omnivores or large-jawed carnivores, depending on diet. Examination of Scaphiopus couchii, an omnivorous species used as a proxy for the non-plastic Spea ancestor, revealed slow growth and a release of cryptic genetic variation for size, development, and gut length when exposed to carnivorous conditions. Two Spea species also show evidence of genetic assimilation. Both exhibit intermediate ecomorph frequencies when they live alone in a single-species “ancestral” condition, but they show near fixation of one or the other ecomorph, independent of dietary resources, when they co-occur (Levis and Pfennig 2016). A macroevolutionary analysis of nematode evolution also provides evidence for the plasticity-first hypothesis. Here, plasticity appears to have promoted the diversification of mouthpart feeding mechanisms, including the evolution of a predatory and sometimes polyphenic morph with moveable teeth. The comparative analysis of 90 species showed that the historical appearance of mouthpart plasticity is associated with faster evolution, increased diversification, and subsequent independent losses of plasticity (Susoy et al. 2015). On the other hand, one analysis of gene expression and molecular evolution in the spadefoot toad system suggests that plasticity could be a consequence, rather than a cause, of rapid evolution. Genes with differential expression between the omnivore and carnivore ecomorphs were compared to a set of unbiased genes, both in the plastic Spea species and in four species that diverged before the evolution of plasticity. The biased genes, those presumed to be associated with plasticity, showed higher variance in expression and faster evolution than the unbiased genes, but the elevated rates predate the evolution of plasticity. With these findings, the authors speculate that plasticity may emerge when fast-evolving, dispensable genes become available for environment-dependent adaptation (Leichty et al. 2012).
Evolution of Plasticity
Given its inherent physical and chemical properties, by default development should respond plastically to environmental influence (Nijhout 2003). Evolved modifications to these biophysical responses to the environment include both the dampening of plasticity, via canalization, and codification of plasticity into a polyphenism. Under this expectation, Nijhout (2003) classifies plasticity two ways. Type 1 emerges as a function of the physical and chemical interface between the developing organism and the environment and is unlikely to increase fitness. Type 2 is an evolved adaptation to a particular environment. For continuous traits, evolution of plasticity means the evolution of reaction norms. Greater environmental heterogeneity speeds reaction norm evolution, though the extent to which environments are novel or rare may matter; cryptic genetic variation released in new environments will expose new reaction norms (Schlichting and Pigliucci 1998). The shape of reaction norms can vary dramatically by trait. Thermal traits often exhibit convex norms, with the lowest values at the temperature extremes; threshold-mediated traits are likely to be logistic; and morphological traits show a diversity of reaction norm shapes, including in different environments (Schlichting and Pigliucci 1998). As for possible genetic mechanisms that permit plasticity, three have been proposed: pleiotropy, epistasis, and overdominance. The pleiotropic model states that one gene can pleiotropically affect fitness in different environments; the epistatic model expects the interaction of two or more genes, some affecting the height of the reaction norm and some determining its shape (see the chapters on “Pleiotropy and Its Evolution: Connecting Evo-Devo and Population Genetics” and “Epistasis”). The overdominance model predicts that plasticity should increase with homozygosity, as heterozygous loci buffer phenotypes from environmental perturbations (Pigliucci 2005).
Although polyphenisms are discrete, some may be the product of a discontinuous environment acting on the norm of reaction of what is otherwise a continuous trait (Nijhout 2003). Others are mediated by developmental switches, but even those may have evolved to direct the development of what was an ancestrally continuous trait. Hormone regulation, especially in insects, is a pervasive mechanism of polyphenism, wherein changes to hormone secretions or hormone sensitivity act on the process of metamorphosis. Notably, the environment that induces a polyphenic trait is often not the environment that imposes selection. For example, seasonal polymorphisms that promote fitness in either warm or cold environments often anticipate those conditions by photoperiod (Nijhout 2003).
In an unpredictable world, the ideal organism would have total plasticity: the ability to form or reform to every situation with maximum fitness. Obviously, total plasticity is not possible, and both costs and limits probably constrain its evolution (Murren et al. 2015; see also the chapter on “Developmental Constraints”). Costs to plasticity can be inferred from the evolution of specialist species; if a generalist could maximize fitness in multiple environments, then we would not expect specialists, with little ability to thrive outside their niches, to evolve in any of those environments (Kawecki 1994). And yet they do, across broad taxonomic and ecological ranges.
However, empirical studies often fail to detect any fitness costs to generalists. Identifying costs of plasticity can be tricky because they should not be confused with costs of the phenotypes themselves. If a generalist and a specialist both produce adaptive traits in a specific environment, but the traits are unequal and the specialist is fitter, this is not a cost of plasticity but a cost of phenotype. A cost of plasticity is the universal cost incurred to the generalist in all environments, such as the expense of larger genome to house additional genetic machinery. It may be that such universal costs are not so important in limiting the reach of plasticity and that instead insufficient selection is a more important constraint for the evolution of generalists (Murren et al. 2015). For example, generalist species may experience weaker selection against deleterious mutations that erode the fitness of environment-specific traits. Unlike specialists, in which fitness-compromising alleles will be exposed in all members of the population, in generalist populations only the subset of individuals within the specific environment will be subject to selection (Kawecki 1994). Variation in selection also affects the evolution of plasticity. Fluctuations in selection arising from environmental heterogeneity favor plasticity, over timescales occurring within individual lifespans as well as those spanning generations, especially when environmental cues are reliable. Limited genetic variation for plasticity has also been proposed as a constraint for its evolution, though recent analyses have estimated greater evolution of norms of reaction than of the traits themselves, suggesting that this may not be an important limitation (Murren et al. 2015).
Micro-plasticity and the Quantitative Genetics Perspective
This chapter, and the literature on plasticity in general, has been focused at the level of the organism. However, plasticity can be observed and quantified at the population level, and indeed this is an important consideration in the evolution of populations. From a quantitative genetics perspective, if the mean of a trait within a population changes when the environment changes, the trait is plastic (Pigliucci 2005). The plasticity may include GxE (and it usually does), but it need not, if the environment pushes trait expression in the same direction and with the same magnitude for every genotype. Technically, plasticity may occur even if the mean and variance of the population remain unchanged; consider crossing reaction norms, as in Fig. 1d, but in a perfectly symmetrical hourglass arrangement with a balanced number of individuals for each genotype. In this scenario, plasticity is only at the level of the organism and not at the level of the population, so the plastic response does not change the ability of the population to evolve if a new phenotype is favored. However, population-level plasticity can occur without a change in trait mean so long as the variance changes. Here, the plastic response could enable faster phenotypic evolution, as outliers become targets of selection. This is the premise behind the theory that cryptic genetic variation can potentiate adaptive evolution (Paaby and Rockman 2014). Or, if stabilizing selection disfavors phenotypic outliers, evolution of allele frequencies can occur without phenotypic evolution. Both of these scenarios require that the higher trait variance is associated with the new environment (see the chapter on “Quantitative Genetics and Evo-Devo”)
Usually, plasticity is discussed in terms of defined differences in the environment. However, even within controlled or static environments, individuals of the same genotype do not produce identical phenotypes. This variance in phenotype, due to unknown and uncontrolled variations in the developmental or external environment, is microenvironmental plasticity. This phenomenon has also been understood as developmental noise, stochastic or residual variation, and environmental sensitivity (Morgante et al. 2015).
Exploiting Plasticity in the Laboratory
Research into the evolution of development can take advantage of dramatic plastic responses to environmental perturbation. Related taxa often exhibit conservation of developmental processes at the morphological level but divergence in the genetic mechanisms that govern them, a phenomenon called developmental system drift (True and Haag 2001; see also the chapter on “Developmental System Drift”). Analyses of molecular evolution and comparative genomics can provide insight, but functional dissection of developmental processes is especially hard in non-model systems and systems that cannot produce viable interspecies crosses. However, environmental perturbations can decanalize development and induce morphological aberrations that may in turn provide clues to the cellular processes that connect the diverged genes to the conserved phenotype. This is analogous to using genetic perturbations to reveal hidden but functional differences in the developmental mechanisms that vary cryptically within populations (Paaby et al. 2015) or have diverged across species (Verster et al. 2014).
Unlike a gene-based perturbation, an environment-based perturbation might not provide a hypothesis-testing framework with mechanistic specificity regarding developmental variation across lineages. However, this may be compensated by the relative ease and consistency that an environment-based perturbation affords. For example, temperature stress, chemical exposure, and nutrition are easily controlled in the lab. There is no expectation that an environmental stress will necessarily reveal functional differences connected to the stress itself, for example, through a history of selection. Rather, the idea is that the environmental perturbation will destabilize developmental trajectories to reveal cellular or genetic differences between the tested lineages. For example, polymorphism in the candidate gene Ultrabithorax was shown to underlie variation of expression in, and eventual genetic assimilation of, the bithorax phenotype in D. melanogaster lineages exposed to ether (Gibson and Hogness 1996).
Like those that induced the bithorax (Gibson and Hogness 1996) and crossveinless (Waddington 1953) phenotypes, an informative perturbation is one that is sufficient to deform the developmental trajectory but not so effective that it kills the organism outright. Temperature stress has been used in multiple systems to induce heritable, intraspecific phenotypic variation, some of which recalls other naturally evolved phenotypes (Moczek et al. 2011). Across lineages, divergence in function but also the expression of recurrent phenotypes arises from the twin aspects of lability and robustness that characterize developmental processes (Paaby and Gibson 2016). For traits undergoing developmental system drift, natural selection will favor changes that stabilize the phenotype around the canonical type when de novo mutations or new environments cause deformations to the developmental trajectory. Consequently, a perturbation that unveils glimpses of these deformations will not be one that evokes type 2 plasticity, an evolved adaptation to a known environment, but one that evokes type 1, a physical and chemical response (Nijhout 2003) that can exaggerate differences in mechanism.
- Levis NA, Pfennig DW (2017) Phenotypic plasticity. In: Pfennig K (ed) Oxford bibliographies in evolutionary biology. Oxford University Press, New YorkGoogle Scholar
- Schlichting CD, Pigliucci M (1998) Phenotypic evolution: a reaction norm perspective. Sinauer Associates Incorporated, SunderlandGoogle Scholar
- Waddington CH (1957) The strategy of the genes. George Allen and Unwin, LondonGoogle Scholar
- West-Eberhard MJ (2003) Developmental plasticity and evolution. Oxford University Press, OxfordGoogle Scholar