so that the genetic load is 0.03. Allozyme surveys of the time also suggested that about one-third of all loci in Drosophila had more than one allele segregating. Extrapolated to the entire genome, thought to be composed of around 8000 to 10,000 loci, it was believed that perhaps 2000-3000 loci were variable. If each locus is completely independent, then the segrega-tional load for the entire genome is

L = (1 - (0.3)(0.1))3000 = (0.97)3000 = 2.07 x 10-40 (11.16)

This produces a conclusion that the genetic load would be enormous. Interpreted in biological terms, this genetic load means that an individual heterozygous at 3000 loci would have to produce 1040 progeny (the inverse of the load) for each progeny produced by an individual homozygous at all loci. Even using lower average heterozygosity per locus and selection coefficients leads to impossibly high genetic loads.

Segregational and substitutional loads played an important role in attempts to explain the proportion of segregating loci and levels of heterozygosity in the 1960s. Balancing selection was considered and rejected by Lewontin and Hubby (1966) as a hypothesis to explain the first allozyme polymorphism data in Drosophila. Expectations for the substitutional and segregational loads were explored and developed in a series of papers authored and coauthored by Kimura (Kimura 1960, 1967; Kimura et al. 1963; Kimura and Maruyama 1966). The genetic load ultimately played a key part of the argument given by Kimura in his proposition of the neutral theory of molecular evolution (Kimura 1968). Kimura argued that there was too much genetic polymorphism or too fast a rate of divergence for all genetic changes to be caused by natural selection. This is because natural selection imparts a genetic load. Kimura's alternative was that many polymorphisms are selectively neutral. The neutral explanation greatly reduces the genetic load since only a small portion of polymorphisms would be acted on by selection and accrue a genetic load.

There are a series of counter-arguments to the idea that natural selection is limited by the segregation load (reviewed in Wallace 1991; Crow 1993b; section 2.11 in Ewens 2004). One criticism revolves around the point of reference used for the maximum fitness in a population (wmax above). Haldane and Kimura defined load relative to the most fit genotype in the population. In the case of the balance hypothesis, the most fit genotype would be one that is heterozygous at all loci. However, such a genotype that is heterozygous at all loci would be very, very infrequent in an actual population if we assume that fitness is based on numerous loci. For example, imagine that all allele frequencies are equal to 0.5 and fitness is determined by 100 loci of equal effect. The expected frequency of a 100 locus heterozygote is (0.5)100 = 7.89 x 10-31 in a randomly mating population. Ewens (reviewed in 2004) showed that if genotypes with fitness values four standard deviations greater than the population mean of one are used as a reference point, then wmax equals 1.98. This implies that the most fit individuals need to produce about two progeny for every one progeny produced by average fitness individuals. This cost of selection seems tolerable for many populations and species.

Another counter-argument focuses on the form that natural selection takes while it works to cull individuals with less-fit genotypes from a population. Estimates of genetic load commonly assume that every locus is independent, so that natural selection must act against homozygous genotypes at each and every locus independently. This is equivalent to assuming multiplicative fitness across multiple loci, seen as an exponent in equation 11.16. This assumption has the consequence of maximizing the perceived genetic load. Another possibility is that natural selection can cull less-fit genotypes by acting on several loci simultaneously. For example, if selection results in the death of an individual because it bears a homozygous genotype at one locus, it also has the effect of culling any other homozygous loci in that individual from the population at the same time. Thus, counting each homozygous locus toward the genetic load, without regard to the fact that some homozygous loci occur in the same individual and can be selected against by only one selective death, overstates the total genetic load.

A final category of arguments against the cost of selection limits involves the way in which natural selection works. A viability selection model assumes that individuals selected against will die before reproduction. An obvious alternative is that selection instead takes the form of fecundity selection among reproducing adults. Some degree of fecundity selection (in lieu of some amount of viability selection) would reduce the number of selective deaths required for substitution or segregation. Instead, genetic load would take the form of differences in fecundity among individuals of different genotypes. This is often described as the distinction between hard (viability) and soft (fecundity) forms of selection. Further, truncation selection as an alternative form of natural selection (see Fig. 9.10) was modeled in detail as an alternative to assumption of independent loci in estimates of the genetic load. The genetic load is considerably less under truncation selection, making the cost of selection low enough to be approximately consistent with the balance hypothesis. However, it is not clear whether truncation selection ever actually occurs in nature.

The selectionist/neutralist debates

The neutral theory of molecular evolution was proposed by Kimura in 1968 (reviewed in Kimura 1983a), coupling his sophisticated knowledge of models of genetic drift with the then novel (and scarce) data on rates of amino acid divergence. Kimura used the amino acid divergence data in mammals to estimate that the rate of nucleotide substitution genome-wide was about one site every other year. As discussed above, this implied a very large genetic load if natural selection was the principal process that governed the eventual substitution of new mutations. Kimura showed, instead, that if new mutations are neutral (meeting the condition that 12Nes |<< 1) then the genetic load was low enough to seem reasonable. (In the process of making the load calculation he also showed that the rate of substitution of a neutral mutation was approximately the mutation rate.) In the same paper, Kimura considered the level of polymorphism in the data of Hubby and Lewontin (1966), estimating that an effective population size of between 2300 and 9000 would produce the levels of heterozygosity observed in Drosophila if alleles were selectively neutral.

The controversy caused by the neutral theory was both intense at times and long-standing. Defending and extending the neutral theory would occupy Kimura for the rest of his life. The proposal of the neutral theory ushered in a new era in population genetics that saw the development of numerous models constructed with the goal of explaining levels of polymorphism or rates of divergence (see Chapter 8). At the same time, there was an increasing volume of genetic data available to test population genetic models. New data often revealed patterns of polymorphism or divergence that were not strictly compatible with neutral theory, motivating continual extensions to the neutral theory. At the same time, numerous advances in the theory of natural selection at the molecular level (e.g. hitchhiking, codon bias, background selection) were developed as alternative hypotheses to the neutral theory. After Kimura's (1968) initial proposal of neutral theory, genetic load faded in importance while levels of polymorphism and rates of divergence became the primary issues.

Pan-neutralism and pan-selectionism are caricatures that illustrate some of the exaggerated stances taken in selectionist/neutralist debates. We can redraw the mutation fitness spectrum presented in Chapter 5 in a way that schematically represents the extreme positions of pan-neutralism and pan-selectionism (Fig. 11.1; compare with Fig. 5.1). Both of these points of view are extreme because they rely on a picture of the fitness of mutations that does not match most observations. The neutral theory was often misunderstood and taken by some as a proposal that all nucleotide or amino changes were selectively neutral (Fig. 11.1a). Pan-neutralism is

0 0

Post a comment