The Princeton Guide to Evolution

V.12

Genetics of Phenotypic Evolution

Catherine L. Peichel

OUTLINE

1. Genetic architecture of phenotypic evolution

2. Molecular basis of phenotypic evolution

3. Using genotypes to test whether phenotypes are adaptive

4. Genetic basis of repeated phenotypic evolution

5. Prospects for future research

The incredible diversity of life on earth is most easily evidenced at the level of the phenotype, which is any characteristic of an organism that can be observed or measured. Thus, the term phenotype encompasses morphological, behavioral, and physiological traits. For evolution of a phenotype to occur, it must have a genetic basis (i.e., be heritable). It is therefore necessary to understand the genetic underpinnings of phenotypic traits to understand the process of phenotypic evolution. This chapter will focus on the genetic and molecular basis of phenotypes that are adaptive; that is, phenotypes that contribute to fitness in a given environment. Although the genetics of adaptation has a long history of study, experimental progress was somewhat limited for much of the last century; however, recent technological advances in genetics and genomics have enabled the identification of genes and mutations that underlie phenotypic evolution in both plants and animals. These initial studies have begun to address long-standing questions about the number and effect sizes of the genetic changes that underlie adaptation, the types of genetic changes involved, the evolutionary history of the genetic changes, and whether the same genetic changes are used when similar phenotypes evolve in independent populations; however, more work needs to be done across a number of systems to gain a complete picture of the genetic basis for phenotypic evolution and adaptation. Fortunately, rapid progress in genome sequencing technologies and the development of new experimental systems are providing an unprecedented opportunity to understand the link between the environmental agents of selection and the phenotypic and genotypic targets of selection.

GLOSSARY

Adaptation. The process by which a population evolves to have higher fitness in a given environment.

Enhancer. A DNA sequence that controls the expression of a gene; enhancers are also referred to as cis-regulatory elements.

Inversion. A chromosomal rearrangement in which the sequence of DNA is reversed. When present in a heterozygous state, recombination is suppressed within the inversion.

Phenotypic Effect Size. The amount of variation in a phenotype that can be explained by a particular genetic change.

Pleiotropy. The same gene or mutation affects multiple phenotypes.

Quantitative Trait Locus (QTL) Mapping. A genetic linkage mapping approach that seeks to identify associations between genotype and phenotype on a genome-wide scale. This approach provides information about the genomic location, number, and effect sizes of the genetic loci that underlie a given phenotype. It requires the ability to cross individuals with different phenotypes.

Repeated Evolution. The appearance of similar phenotypes in independent evolutionary lineages that experience similar environments; also referred to as parallel or convergent evolution.

1. GENETIC ARCHITECTURE OF PHENOTYPIC EVOLUTION

A fundamental question in evolutionary biology is whether phenotypic changes occur via an infinite number of genetic changes, each with an extremely small effect on phenotype, or via relatively few genetic changes, each with a large effect on phenotype. This question has been hotly debated over the last 150 years, starting with Darwin’s view that evolution must occur through many small changes with slight phenotypic effects (“micromutationism”) because of the observation that gradual and continuous phenotypic changes are found in nature. The rediscovery of Mendel’s laws of inheritance challenged this view because early geneticists began to find mutations that caused large and discontinuous phenotypic effects, at least in the laboratory. This led to a vigorous debate between the Darwinian gradualists and the Mendelian geneticists around the turn of the twentieth century. The successful fusion of micromutationism with Mendelism by the founders of population genetics, particularly Ronald A. Fisher, during the “evolutionary synthesis,” seemed to resolve the debate (see chapter I.2). Using mathematical arguments, Fisher demonstrated that an infinite number of small genetic changes could underlie continuous phenotypic variation. Fisher then compared the genetic process of phenotypic adaptation to the process of focusing a microscope. When focusing a microscope, a large adjustment has a much smaller probability of improving the focus than a small adjustment. In the same way, Fisher considered it very unlikely that mutations of large effect would be beneficial and thus concluded that only mutations with extremely small effects would be beneficial and contribute to phenotypic adaptation (the “infinitesimal model”). While satisfactory, for much of the twentieth century this theoretical argument prevented any further empirical work on the question because it seemed pointless to look for infinitesimally small genetic changes.

Fisher’s initial theory was later revisited by Motoo Kimura, who realized that while mutations with large phenotypic effects were more likely to be deleterious, those that were beneficial were less likely to be lost due to drift, particularly in small populations. Extending this work, H. Allen Orr showed that mutations of large effect that are beneficial are more likely to be fixed, particularly early in the process of adaptation. Taken together, this theoretical framework predicts that the genetic architecture of phenotypic evolution will involve a few genetic changes with large effects and many genetic changes of smaller effects (the “geometric model”).

Although empirical work on this question was limited by the almost-universal acceptance of the infinitesimal model for much of the last century, recent advances in technology have enabled experimental approaches to identify the genetic architecture of phenotypic evolution. In particular, quantitative trait locus (QTL) mapping (see chapter V.13) has been used to identify the genomic location, number, and effect sizes of genetic changes that underlie phenotypic differences among natural populations of plants and animals. Many studies have now conclusively demonstrated that genetic changes of both large and small effect contribute to phenotypic evolution; however, very few studies have had the experimental power to address the relative contributions of these changes during adaptation. More data are needed to determine whether the geometric model will generally hold true, but the recent emergence of new genome sequencing technologies will make it feasible to investigate the number and effect sizes of the genetic changes underlying the evolution of a wide variety of phenotypes across numerous taxonomic groups.

Genetic mapping studies such as QTL mapping also enable researchers to address a second question about the genetic architecture of phenotypic evolution: Are the genetic changes that underlie phenotypic evolution and adaptation found in particular regions of the genome or are they distributed across the genome? This question is of particular interest when an organism adapts to a new environment, because multiple phenotypic changes are usually required for adaptation. For example, a plant living in a cool, wet environment and a plant living in a hot, dry environment will differ in many respects, including morphology, physiology, and life history. Thus, adaptation to a new environment might be facilitated if the same genetic changes give rise to multiple phenotypic differences (pleiotropy), or if independent genetic changes are each responsible for a single phenotype, but multiple phenotypes are inherited together (linkage).

The evidence for pleiotropy is certainly good after decades of genetic studies in model laboratory organisms; mutations in a single gene often affect multiple phenotypes. Although there is also evidence that multiple phenotypes map to the same locus in genetic studies of natural populations, most of these studies have insufficient resolution to determine whether multiple phenotypes are controlled by the same gene or by tightly linked genes. Classic examples of such supergene complexes are found in Müllerian mimicry in Heliconius butterflies. Multiple aspects of wing color patterns (e.g., the type and distribution of pigments) map to a single locus, but it is still not known whether these are due to mutations in different enhancer elements of the same gene, or to mutations in multiple, closely linked genes. To distinguish these possibilities, it will be necessary to identify the actual sequence changes that give rise to the color pattern phenotypes (see next section).

In either case, recombination within the supergene complex would create individuals that are not perfect mimics and thus would be less likely to survive. Interestingly, one of the supergene complexes in Heliconius is found within a chromosomal rearrangement that suppresses recombination. It has long been thought that such chromosomal rearrangements, particularly inversions, might be hot spots for genes involved in phenotypic evolution and adaptation, as well as speciation (see chapter VI.9). Many closely related species differ by one or more chromosomal rearrangements, and differences in the frequency of chromosomal inversions within species have been correlated with environmental clines. These data suggest that inversions might be important for adaptation, but only recently has this hypothesis been directly tested using yellow monkeyflowers (Mimulus guttatus). Two forms of this species are found along the west coast of the United States: a perennial form adapted to the cool and wet coastal region, and an annual form adapted to the hot and dry inland region. Many of the phenotypic differences between the forms, including flowering time and a number of morphological traits, map to an inversion. By a clever crossing scheme, the perennial inversion was placed into annual plants and vice versa. Then, plants were placed in the two habitats; remarkably, the inversion altered the phenotypes and conferred increased survival and fitness in the appropriate environment. Although this study provides good evidence that genes involved in phenotypic evolution and adaptation can be clustered within chromosomal rearrangements and thus coinherited, there are still many cases where the genes that underlie multiple phenotypes required for adaptation to a particular environment map to distinct genomic locations. Comprehensive genetic mapping studies of morphological, behavioral, and physiological traits across a number of systems are required to determine the extent to which pleiotropy and linkage contribute to the genetic architecture of phenotypic evolution and adaptation.

2. MOLECULAR BASIS OF PHENOTYPIC EVOLUTION

An ultimate goal of modern evolutionary genetics is to identify the actual genes and specific mutations that underlie phenotypic evolution. Using classical genetic mapping and candidate gene approaches in combination with modern molecular tools such as transgenic technologies (see chapters V.11 and V.13), it has been possible to pinpoint the specific genes and in some cases the mutations that underlie a wide variety of phenotypic differences among natural populations of both plants and animals. Interestingly, most of these cases involve morphological or physiological traits, while only a handful of genes have been identified underlying the evolution of behavioral phenotypes. This is a fruitful area for future research (see below).

There has been some vigorous debate about whether the genetic changes responsible for phenotypic evolution would more likely occur in coding sequences of genes (i.e., the protein itself) or in the cis-regulatory regions of genes (i.e., the regions that control the time and place of gene expression). Thus far, ample evidence has been presented that both coding and regulatory changes can and do contribute to phenotypic evolution; however, the data collected thus far do not represent a completely unbiased set, so it is not yet possible to determine the relative roles of these different types of mutations.

Although finding the actual mutations responsible for phenotypic changes can be challenging, particularly when the mutations are in regulatory regions, this information enables a fine-grained view of the process of phenotypic evolution. As discussed in the previous section, there has been a long debate over whether a few mutations of large effect or many mutations of small effect contribute to phenotypic evolution and adaptation. To date, many studies have found that a single genetic locus can have a relatively large effect on a given phenotype. An interesting question then follows of whether a single genetic locus comprises many mutations, each with small effect, or a single mutation of large phenotypic effect. In the case of coding mutations, there is abundant evidence that a single nucleotide mutation leading to a single amino acid change can be sufficient to cause a large phenotypic effect, as is exemplified by mutations in the Mc1r gene that lead to dramatic differences in pigmentation (see chapter I.4). Although there is good evidence that regulatory mutations can have large effects on phenotype, only a few studies to date have identified the actual enhancers and mutations responsible. In two different cases in fruit flies, multiple single base-pair mutations within a single enhancer are required to create a large phenotype effect. By contrast, in stickleback fish, a single mutational event in a specific enhancer is sufficient to create a large phenotypic effect. With such a limited data set, it is too early to speculate on the reasons these differences may exist. Although conducting such studies to identify mutations in regulatory elements is difficult and long term, they can provide unprecedented insights into the process of adaptation, as discussed in the next section.

3. USING GENOTYPES TO TEST WHETHER PHENOTYPES ARE ADAPTIVE

It has been known for a long time that phenotypes do evolve and thus have a genetic basis; thus, some evolutionary biologists have wondered what new insights can be gained by identifying the actual genes and mutations that underlie phenotypic evolution. In fact, there are many long-standing questions about the genetic basis of phenotypic evolution and adaptation that can be addressed with such information.

First, once a gene that underlies a particular phenotype is known, it is possible to investigate whether the trait evolved in response to selection or as a result of neutral processes like genetic drift. Such an investigation can be conducted by looking for molecular signatures of selection, as has been widely done across genomes or at specific genes (see chapter V.14); however, most of these studies have been genotype focused and suggest only that a locus has experienced selection, but they do not reveal the phenotypes associated with those genotypes or the selective agent involved. Thus, selection is only inferred from the molecular data. By first identifying genes that underlie phenotypes known to be under selection, it is possible to determine whether molecular tests for selection do in fact identify loci associated with phenotypes under selection and thereby inform genotype-based approaches. Furthermore, in studies in which it is not known whether a phenotype is under selection, identifying the genes responsible enables tests for selection, as demonstrated by a pair of recent studies in stickleback fish and fruit flies. In both cases, the actual mutations responsible for phenotypic changes were identified: deletion of an enhancer for Pitx1 gene contributes to loss of pelvic spines in sticklebacks, and mutations in an enhancer for the ebony gene contribute to changes in body coloration in fruit flies. Although these phenotypes were predicted to be adaptive, evidence for selection on these phenotypes was obtained by identifying the molecular changes involved and then looking for molecular signatures of selection; thus, these two studies provide compelling examples of the ability to make a connection between selection, phenotype, and genotype.

Even once such connections are made, it is still difficult to discern the selective agents responsible for the evolution of the phenotype. A second advantage of identifying the genes responsible for a phenotype is that it can enable studies to identify specific agents of selection. For example, the fitness effects of alternative alleles at a particular locus can be assessed in the field or in controlled, seminatural habitats. To date, very few studies have done this, but the ones that have are classics. For example, a single locus of major effect, called YUP, is responsible for the difference in color between a pink, bumble-bee-pollinated species of monkeyflower (Mimulus lewisii) and its red, hummingbird-pollinated sister species (M. cardinalis). By swapping the YUP locus between these two species and planting them in the field, a reversal in pollinator preference can be observed. These data demonstrate that the change in flower color is adaptive and that pollinator preference is the selective agent at work.

A third advantage of identifying the genetic basis of a phenotype is the ability to learn about the evolutionary history of an adaptive allele. For example, it is possible to determine whether adaptation to a new environment involves new mutations or selection on existing genetic variation. There is now clear evidence that both contribute to phenotypic evolution, sometimes at the same genetic locus. In the case of fruit fly body coloration described above, both existing variation and new mutation at the ebony gene played a role in the evolution of darker pigmentation at higher altitudes. It is further possible to use sequence data to determine when adaptive alleles arose within populations, as demonstrated by studies of an allele of the agouti pigmentation gene that arose by new mutation in deer mice with light coloration adapted to living on lighter-colored soil in the Sand Hills of Nebraska. Such studies are important because they inform us about the speed of adaptation to new environments, which is particularly relevant when thinking about how quickly organisms might adapt to global changes in climate in the future.

4. GENETIC BASIS OF REPEATED PHENOTYPIC EVOLUTION

The repeated evolution of similar phenotypes in response to similar environmental conditions is generally taken as strong evidence of a role for natural selection in the evolution of that phenotype. Such repeated evolution is referred to as parallel evolution when similar phenotypes evolve in closely related lineages and as convergent evolution when similar phenotypes evolve in distantly related lineages; however, the distinction between parallel and convergent evolution can be contentious and is further complicated by identification of the molecular basis of these traits. For example, the same genes, and sometimes even the same mutations, underlie the repeated evolution of the same phenotype in independent lineages, as exemplified by the finding that mutations in the Mc1r gene underlie differences in coloration among multiple species of fish, birds, snakes, lizards, and mammals. Strikingly, the same mutation in Mc1r can be found in some melanic birds and mammals; however, not all pigmentation differences are controlled by Mc1r, and even closely related subspecies of mice do not share the same genetic basis for similar pigmentation patterns. Thus, phenotypic parallelism or convergence may not be mirrored at the genetic level, creating an issue of whether these terms should be defined at the level of phenotype or genotype. Despite this semantic debate, an important and interesting question remains: When selection favors the evolution of similar traits, are the same genes and mutations used, or can different genes and mutations create similar phenotypic changes?

Although the available data are far from comprehensive or unbiased, the answer so far appears to be yes to both. Repeated evolution of phenotypes as diverse as flowering time in plants, insecticide resistance in insects, and pigmentation in vertebrates and invertebrates does occur via changes in the same genes, but not always. However, when particular phenotypes appear to evolve using a limited number of genes, it begs the question of why the same genes might be used repeatedly. For some traits, it might be that only a few genes can alter a given phenotype. For example, changes in color vision almost necessarily occur through changes in opsin genes. However, it is likely that for most phenotypes, mutations in any of hundreds of genes might give rise to a particular phenotype. For example, more than 80 genes can regulate flowering time, but only a few of these genes appear to be used during the evolution of differences in flowering time. Though the question of why particular genes are reused during evolution is far from resolved, there are several possible reasons, each supported by empirical evidence.

First, there may be a difference in mutational bias between genes; that is, some genes are larger targets for mutation or found in regions of the genome with higher mutation rates. For example, an enhancer in the Pitx1 gene has been deleted at least nine independent times in stickleback populations that have lost the pelvic spines. This enhancer is located in a region of the stickleback genome that has features associated with DNA fragility, which may predispose it to deletion. However, not all repeated use of the same genes can be explained by differences in mutation rate and might reflect either selective constraints or historical contingency; these possibilities are discussed next.

A second reason for the repeated use of particular genes is related to pleiotropy. As discussed above, many genes are known to affect more than one phenotype. If a mutation in a particular gene causes one phenotypic change that is beneficial, but also causes additional phenotypic changes that are detrimental and outweigh the benefits, there could be selection against that particular mutation. Thus, genes that are expressed only in a specific tissue might be more likely to be used repeatedly. Consistent with this hypothesis, the Mc1r gene has been repeatedly implicated in the evolution of pigmentation across vertebrates and seems to have relatively few pleiotropic effects. Importantly, even though a gene might have an effect on multiple phenotypes, it is possible to identify mutations in the gene that have no pleiotropic effects. For example, mutations in an enhancer that drives expression in a particular anatomical location might be less likely to confer deleterious pleiotropic effects than a mutation in the coding region or in a general promoter region of a protein expressed in multiple tissues or with multiple functions. The Pitx1 gene mentioned above provides a nice example; this gene is expressed in a number of tissues, and laboratory mice with a deletion of Pitx1 die shortly after birth. In sticklebacks, the deletion of an enhancer that drives expression of the Pitx1 gene only in the developing pelvic region is therefore a mutation with no pleiotropic effects, despite the fact that the gene itself exhibits pleiotropy.

Third, the use of a particular genetic change for phenotypic evolution might be dependent on the historical background of genetic variation already present within a population. If a large ancestral population repeatedly colonizes similar environments, there may be standing genetic variation that can be repeatedly selected in the new environment, leading to reuse of the same genetic change to evolve the same phenotype. A now classic example occurs in stickleback fish; marine ancestors have colonized freshwater environments and consequently evolved reduction of body armor in stickleback populations across the Northern Hemisphere. Repeated evolution of this phenotype is due to repeated selection of a particular allele of the Ectodysplasin gene present at low frequency in the ancestral marine population. In other cases, it is possible that the ancestral genetic background has no phenotypic effect on its own, but rather influences that new mutations have a phenotypic effect and thus are selected. Studies of the shavenbaby gene in fruit flies demonstrated that the phenotypic effect of any single mutation in an enhancer was dependent on which other mutations were also present in the enhancer. Not only are interactions within a locus important, but interactions between two or more loci might also influence the types of mutations that are selected. Reduced pigmentation in beach mice that live on light sand dunes in Florida is due to mutations in two previously discussed pigmentation genes, Mc1r and agouti. In this case, the phenotypic effects of the Mc1r mutation can be observed only when the agouti mutation is present. Thus, the selective advantage for the Mc1r mutation occurs only on a specific genetic background.

As more and more genes are identified that contribute to the repeated evolution of similar phenotypes across a number of different plant and animal groups, a more complete picture will begin to emerge of the relative roles of mutational bias, selective constraints such as pleiotropy, and historical contingencies such as genetic background. Such data will allow us to determine whether there are any “rules” for phenotypic evolution and adaptation.

5. PROSPECTS FOR FUTURE RESEARCH

This is an exciting time to be an evolutionary biologist. Along with a renaissance of interest in understanding the genetics of phenotypic evolution and adaptation, genome sequencing technologies are becoming widely available. Thus, it is now feasible to develop genetic tools and genomic resources for many systems with interesting phenotypic differences, rather than focusing on just a few model systems or a limited number of phenotypes. In particular, very little is known about the genetic changes that contribute to the origin of novel phenotypes or the evolution of behavior. The recent successes in identifying genetic changes that underlie morphological and physiological evolution suggest that by using appropriate systems, progress will be made on these types of traits as well. And continued efforts to identify the genetic underpinnings of many phenotypes in diverse plant and animal systems will surely provide a comprehensive picture of the relative contributions of small-versus large-effect mutations, coding versus regulatory mutations, and new mutations versus standing variation to phenotypic evolution and adaptation.

With these genetic changes in hand, it will also be possible to begin to connect the genotypic and phenotypic targets of selection with the environmental agents of selection. In particular, once a genetic change has been identified that contributes to a phenotype, that genetic change can be made on controlled genetic background. The phenotypic and fitness effects of the genetic change can then be measured in controlled environmental conditions or in seminatural habitats. These types of experiments will ultimately enable identification of the particular environmental conditions that select for a specific genetic and phenotype change. Such studies will provide unprecedented insights into the process by which phenotypes evolve and organisms adapt to their environments.