Historical and synthetic topics

11.1 Historical controversies in population genetics

• The classical and balance hypotheses.

• How to explain levels of allozyme polymorphism.

• The selectionist/neutralist debates.

Although past debates in population genetics might not seem relevant today, it is usually the case that history has a strong influence on the present. This is certainly true in population genetics. Often, contemporary approaches to problems or accepted interpretations of a pattern are the products of rich and sometimes contentious debates. New data types such as DNA sequences and the sheer volume of genetic data available today have put to rest some open questions of the past. However, it is often the case that older controversies take new forms with some modifications, often because old problems are not completely resolved even though the field moves on to new topics. An appreciation of some of the ideas that have occupied population genetics in the past provides invaluable perspective on the present and future of population genetics. The brief and selective history presented here starts in the 1940s and 1950s and is meant to provide an overview of the spirit of past disagreements rather than a rigorous review. Note that the technical foundations involved in these topics are presented in earlier chapters as indicated but are not repeated in detail here. Readers interested in a history of early population genetics from Darwin through the 1930s should refer to Provine (1971).

The classical and balance hypotheses

The theoretical work of Fisher, Haldane, and Wright established the core principles of population genetics. These included that natural selection was able to alter allele frequencies rapidly, mutation and recombination supply genetic variation, that mating patterns and gene flow shape the hierarchical organization of genetic variation, and that the effective population size regulates the process of genetic drift. Taken collectively, this body of theoretical expectations served to fuse Darwin's concept of natural selection with the principles of Mendelian particulate inheritance. These expectations form the foundation of population genetics and were labeled neo-Darwinism by Huxley (1942).

While the neo-Darwinian synthesis achieved by population genetics reached orthodoxy in the 1930s and 1940s, a long-running debate began to take shape. Under the logic of early neo-Darwinism, natural selection was the dominant evolutionary force in almost all aspects of evolutionary change. It was then a matter of debate as to what type of natural selection - directional or stabilizing - was most common in captive and natural populations. The answer to this question gradually turned into two broad points of view based on what one assumed and how one interpreted available data on genetic variation. Dobzhansky (1955) labeled these schools of thought the classical hypothesis and the balance hypothesis. Both hypotheses rely on natural selection as the principal process operating in populations, although they differ greatly in the predicted consequences of natural selection.

Classical hypothesis The point of view that directional natural selection is the dominant process in populations, predicting relatively little genetic variation except when selection pressures are heterogeneous in time or space or are frequency-dependent. Balance hypothesis The point of view that balancing natural selection is the dominant process in populations, predicting extensive genetic variation caused by overdominance for fitness.

The classical hypothesis was that directional selection was the predominant process in populations and from this two major predictions arose as a consequence. The first prediction was that under random mating populations contained individuals homozygous at most loci. The second prediction was that populations harbored relatively little genetic variation since the equilibrium points for any sort of directional selection on a diallelic locus are fixation and loss or near fixation and loss (see Chapter 6). The classical school recognized the existence of "wild-type" alleles, or alleles at high frequency in a population because such alleles were of higher fitness and were brought to high frequency by directional selection. Alternative "mutant" alleles that appeared in populations were most often deleterious but on very rare occasions would have a higher fitness than the current wild-type allele and would then become the new wild-type allele. The classical school predictions were supported by a range of empirical observations, especially from laboratory populations of organisms such as Drosophila. In such populations, phenotypes are of the wild type (within some range of variation) and mutations with visible phenotypic effects appear rarely but are almost universally deleterious and do not reach high frequencies.

The classical hypothesis predicted that genetic variation in populations was produced in four ways (Dobzhansky 1955). First, deleterious mutations continually occur and segregate for a short time before they are eliminated by directional natural selection. Most of these deleterious mutations are likely to be recessive and thus exist mostly in heterozygote genotypes where they are sheltered from natural selection. (Dobzhansky pointed out that these are the sorts of mutations that cause hereditary diseases when homozygous.) Second, some proportion of mutations are adaptively neutral because they have marginal fitness values very near the mean. A third possibility is that rare beneficial mutations are found in a population before they have become established as the new wild-type allele. The fourth possibility is that some mutations are slightly beneficial in one environment but slightly deleterious in another environment. Alleles will then persist in a population exposed to environments that are heterogeneous in time or space. This last category motivated numerous population genetics models where directional selection occurs but fitness varies in time or space, or individual fitness is frequency-dependent (see Chapter 7).

The balance hypothesis took the alternate perspective that overdominance for fitness was the general rule in most populations, so that balancing selection was the dominant process that regulated genetic variation. Under balancing selection, heterozygotes would have higher frequencies than in the absence of selection or under directional selection (see Chapter 6). The balance hypothesis, therefore, predicted that loci would maintain two or more alleles indefinitely. Owing to balancing selection, heterozygotes would be more frequent and homozygotes much less frequent than expected by Hardy-Weinberg or under directional selection. Of the new mutations that enter a population, only those that exhibited overdominance as heterozygotes were expected to be retained in the population.

As Dobzhansky (1955) explained, the balance hypothesis was also associated with predictions about the inter-relationship among loci. Natural selection on multiple loci can lead to the accumulation of gametic disequilibrium (see Chapter 2). High levels of gametic disequilibrium are expected under balancing selection in populations with all loci at intermediate allele frequencies since only that subset of gametes that produce multilocus heterozygote zygotes would have high fitness. (Note that under the classical hypothesis there is also strong natural selection, but relatively little gametic disequilibrium in absolute terms is expected because populations would be close to fixation for the wild-type allele.) Using this expectation for gametic disequilibrium and then assuming that most loci experience balancing selection leads to the prediction of coadapted gene complexes or supergenes within species (reviewed by Hedrick et al. 1978). A supergene is a haplotype or genotype at many loci that is held together and frequently inherited as an intact unit because gametic disequilibrium is very strong. A coadapted gene complex is a supergene where natural selection acts or has acted so that the alleles or genotypes at each locus have high fitness in the context of the alleles or genotypes at all other loci. Stated another way, selection will increase the frequency of new mutations that interact well with the alleles and heterozygous genotypes at other loci. In contrast, any mutations that have reduced relative fitness caused by interactions among loci will be reduced in frequency by natural selection. Thus, the notion of a coadapted gene complex assumes that epistasis for fitness is common.

Supergenes and coadapted gene complexes presented both a research agenda and a conceptual challenge to biologists of the classical/balance hypothesis era. A great deal of research was devoted to the study of multilocus genetic variation in laboratory and natural populations. At the same time, numerous models of multilocus natural selection and recombination were developed and studied. Very high levels of gametic disequilibrium caused by balancing selection and epistasis for fitness served to negate the Mendelian process of independent assortment. What then was the source of genetic variation required for evolutionary change? The answer was often sought in population genetic mechanisms that had the potential to recombine or break up supergenes.

Although the notion that large supergenes or coadapted gene complexes are widespread is not popular now, contemporary population genetics has inherited an appreciation of the processes that cause gametic disequilibrium and evidence that multiple loci are not necessarily independent. We now have well-characterized examples of genome regions with high levels of gametic disequilibrium. One of the best examples is the major histocompatibility complex (Mhc) loci in mammals. These loci experience balancing selection because of their functional role in recognizing non-self peptide fragments and compose a large chromosomal region that has relatively high levels of gametic disequilibrium. The amorphous supergene prediction has now been refined into a series of much more specific hypotheses tailored to diverse areas of population genetics. Non-independence of loci is central to models of molecular evolution that seek to explain polymorphism within populations, as embodied by concepts such as hitchhiking, background selection, and genetic draft (see Chapter 8). Quantitative genetics recognizes non-independence of traits caused by phenotypic and genetic correlations. The idea that selection favors alleles that interact well across multiple loci is now called the Dobzhansky-Muller model and it serves as an explanation of how isolated populations might develop reproductive isolation that leads to speciation (reviewed by Coyne and Orr 2004).

The study of ecological genetics can be traced to efforts to test the classical and balance hypotheses with empirical data. Today, ecological genetics is defined as the study of genetic variation within species in the context of environmental and organismal interactions. Ecological genetics seeks to identify the causes of patterns of genetic variation, often with reference to the assumed or demonstrated pressures of natural selection imposed by ecological context. Early ecological genetics was focused on testing the classical and balance hypotheses for genetic variation. On the one hand was the classical school prediction that relative fitness of alleles varied in time and space.

On the other hand, the balance school predicted that overdominance for fitness was very common. Both of these possibilities were testable to some extent in natural populations, by measuring the relative fitness of phenotypes with a known genetic basis or observing the frequency of genetic polymorphisms. Dobzhansky was among the first to study "laboratory" organisms in the wild. He pioneered field research in Drosophila and established a tradition of empirical research that is now the norm in population genetics. Edmund B. Ford was also instrumental in the establishment of the field of ecological genetics. Ford studied wild butterflies and moths and wrote the enormously influential book Ecological Genetics, which was published in 1964.

Many of the widely known empirical studies in ecological genetics take on new meaning when viewed through the lens of the classical hypothesis/ balance hypothesis debate. For example, industrial melanism in populations of spotted moth (also known as the peppered moth) in England was evidence for the classical-school position since it shows that directional selection pressures vary among populations based on proximity to industrial centers whose soot stained tree trunks black (reviewed by Majerus 1998). (It is no coincidence that Bernard Kettlewell, who performed much of the original spotted moth work as a research scientist at the University of Oxford, was supervised by E.B. Ford.) Another universally known example in ecological genetics, evidence that heterozygous blood-group protein genotypes have higher fitness in malarial areas of Africa, supports the balance hypothesis of overdominance for fitness as the force that maintains genetic variation.

How to explain levels of allozyme polymorphism

Another long-running controversy in population genetics grew out of the classical hypothesis/balance hypothesis debate. The new controversy revolved around how to explain genetic polymorphism within natural populations observed with a then radically new technique. The technique was gel electro-phoresis of enzyme polymorphisms, or allozymes (see Box 2.2). Two papers published in 1966 ushered in the new controversy. Hubby and Lewontin (1966) presented allozyme estimates of heterozygosity for 21 loci estimated from multiple populations of 15-20 Drosophila pseudoobscura individuals. Nine of these 21 loci exhibited between two and six alleles segregating within populations. The Hubby and Lewontin paper showed a technique that could be used to determine both the proportion of loci that possessed more than one allele and the level of heterozygosity for each polymorphic locus.

The controversy over the causes of allozyme polymorphism changed the focus of much of population genetics within the span of only a few years starting in the mid-1960s. Initially, the classical and balance hypotheses were considered as primary explanations. In fact, in a paper published along with the allozyme data themselves, Lewontin and Hubby (1966) argued that the level of heterozygosity observed (averaged over populations 30% of loci were polymorphic) were inconsistent with the balance hypothesis because of the segregation load that would have been required (see the Genetic load section, below). The remaining explanation for the allozyme polymorphism within the context of the time was directional natural selection consistent with the classical hypothesis.

The balance hypotheses experienced some setbacks from empirical data around the same time. Apparent overdominance in maize was shown to decline over multiple generations (Moll et al. 1964; reviewed by Crow 1993b). True overdominance should persist indefinitely as a function only of heterozygosity. These maize results, however, were consistent with the prediction that overdominance was actually caused by gametic disequilibrium between loci bearing beneficial dominant alleles and other loci bearing deleterious recessive alleles. When two individuals homozygous for different alleles at two such loci are crossed, the progeny will experience a great increase in fitness because the recessive deleterious phenotype will be masked by dominance. The maize results demonstrated that apparent overdominance phenomena were caused by the combination of simple dominance and linkage rather than by true overdominance.

The classical hypothesis/balance hypothesis debate soon receded. Selective neutrality was an explanation under the classical hypothesis that predicted a low level of genetic variation in populations. This idea of selectively neutral alleles, which was developed and mathematically formalized starting in the 1950s and 1960s, emerged as a primary hypothesis for genetic polymorphism. The neutral theory hypothesized that many loci have selectively neutral alleles and that polymorphism was a product of the non-equilibrium random walk that new neutral mutations experience because of genetic drift.

The waning of the balance hypothesis and the ascension of the two components of the classical hypothesis produced what was labeled the neo-classical theory of population genetics by Lewontin (1974).

This label came about because both elements of the classical hypothesis explanation for polymorphism - selectively neutral mutations and mutations under directional or purifying selection - are drawn from the early classical hypothesis. Under the neo-classical hypothesis, the debate became one about the relative contributions of neutral mutations or mutations acted on by natural directional selection to levels of genetic polymorphism. There was also continuing work on the selection element of the classical hypothesis, which was updated and bolstered with empirical support from more elaborate theoretical models and ecological genetic studies.

In the years after 1966, until DNA-based molecular techniques became available in the late 1980s, allozyme electrophesis was perhaps the most widely used empirical technique in population genetics. The allozyme era of population genetics is sometime derisively referred to as the period of "find 'em and grind 'em", in reference to collecting and then homogenizing samples in preparation for allozyme electrophoresis. It is certainly true that during this time a great deal of empirical research focused on gathering single-population estimates of hetero-zygosity via allozyme electrophoresis. However, these data were part of a very substantial shift in the conceptual focus of population genetics. The addition of allozyme data contributed to the shift in emphasis away from the entrenched positions of those who favored the classical or balance hypotheses to new models, data, and hypothesis tests.

Genetic load

For natural selection to change allele frequencies in a population, individuals of some genotypes must experience higher rates of death (either actual for viability selection or reproductive for fecundity selection) than other genotypes. Natural selection works by culling individuals of some genotypes in favor of individuals of other genotypes, increasing the mean fitness in the process. The amount of death or failed reproduction associated with natural selection was first called the load by Muller (1950). Genetic load comes in two forms. Substitutional load refers to the reduction in mean fitness caused during fixation of beneficial mutations or purging of deleterious mutations. Distinctly, the production of individuals with lower-fitness genotypes by Mendelian segregation during reproduction is called the segregational load. Sexual reproduction leads to segregational load because both recombination and independent assortment produce novel progeny genotypes that have a range of fitnesses. Those progeny genotypes with a lower fitness perish (or do not reproduce) under viability selection. In principle, the genetic load places an upper limit on the ability of natural selection to change genotype frequencies in a population. Deaths due to viability selection cannot greatly exceed the total demographic excess of a population (the number of individuals produced each generation beyond those needed for demographic replacement) for long or the population will go extinct.

The genetic load has been a tool used to try to estimate the upper limits to the process of natural selection. The goal has been to determine how strong natural selection can be before an unrealistic genetic load occurs. The genetic load has been used for two main purposes. The substitutional load has been used to estimate mutation parameters in populations, such as the rate of fixation of beneficial mutations, the rate of deleterious mutations, or the decline in fitness associated with deleterious mutations. Substitu-tional load also played a prominent role in attempts to set acceptable thresholds for human radiation exposure during the 1950s. The segregational load was used as a counter-argument against the balance hypothesis during the 1960s in the early days of the neutral theory of molecular evolution. In these roles genetic load has been controversial and engendered passionate arguments over the span of many decades (see reviews by Wallace 1991; Crow 1993b). The genetic load has been employed widely in population genetics and evolutionary biology during the last three decades. For example, the concept has been invoked in the accumulation of deleterious alleles that might cause mutational "meltdowns" in small and endangered populations (e.g. Lynch and Gabriel 1990), and in arguments for the fitness advantage of sexual reproduction.

The genetic load concept (although not the term) was originated by Haldane (193 7). Haldane's result can be seen with the general dominance model of selection on a single diallelic locus (see Table 6.4). Assume a population at Hardy-Weinberg equilibrium, with complete dominance (h = 0), and a maximum fitness that equals one. In such a population the fitness-weighted genotype frequencies are p2, 2pq, and q2 - sq2 where s is the selection coefficient. The mean fitness in the population is then the sum of the frequency-weigh ted fitness values or w = p2 + 2pq + q2 - sq2. The mean fitness is therefore w = 1 - sq2 because p2 + 2pq + q2 = 1. The process of forward mutation (A to a) will make new recessive alleles in the population each generation, transforming some number of Aa genotypes into aa genotypes. Assuming that natural selection and mutation are at equilibrium means that selection removes aa genotypes as fast as they are made by mutation. The rate at which new aa genotypes are made from Aa genotypes is the forward mutation rate or p. The mean fitness of the population is then w = 1 - p (also assuming that reverse mutation can be ignored). With incomplete dominance (h ^ 0), w = 1 - 2pqhs - sq2 and the mean fitness can be shown to be approximately w = 1 - 2 p.

The genetic load is defined as w - w

and expresses the difference between the maximum fitness (wmax), which corresponds to the most fit genotype, and the average fitness (w) in a population at a given point in time (Crow 1958). If wmaxis defined as one, then the genetic load is given more simply by r 1 - w i -

0 0

Post a comment