POPULATION GENETICS Brian Charlesworth University of Edinburgh I. Variation within Populations selection The differential survival or reproductive suc- II. Deterministic Population Genetics cess of individuals, associated with differences in III. Random Genetic Drift phenotype or genotype. IV. The Interaction of Drift with Deterministic Forces V. Conclusions DARWIN’S THEORY OF ‘‘DESCENT WITH MODIFI- CATION’’ implies that all of the stupendous diversity of life on Earth is ultimately traceable to genetic diversity GLOSSARY within populations. The study of the nature and causes of within-population variation, and of the mechanisms allele frequency The frequency of a variant form of a by which it is transformed into differences between genetic locus within a population. populations over space and time, is the province of genetic drift Evolutionary change caused by random population genetics. The subject involves both theoreti- sampling of genotype frequencies in a finite popu- cal modeling of evolutionary processes, based on lation. knowledge of the mechanisms of inheritance, and the genotype The state of an individual with respect to a testing of these models using data on variation and defined genetic locus or set of loci. evolution in natural and artificial populations. heritability The proportion of the variance in a trait that is due to additive genetic effects. inbreeding Matings between close relatives. I. VARIATION WITHIN POPULATIONS mutation rate The frequency with which new muta- tions arise per generation. A. Types of Phenotypic Variation neutral mutations Mutations whose effects on fit- Since evolutionary change depends on the existence of ness are either nonexistent or so small that genetic variation within populations, measurement of their fate is controlled by genetic drift rather than the extent of such variation is crucial (Lewontin, 1974). selection. Variation at the level of externally visible phenotypes phenotype The state of an individual with respect to can be divided into three categories. a trait of interest. polymorphism The existence at intermediate frequen- 1. Discrete Variation cies of two or more variants at a locus within a popu- This involves traits which can be divided into a small lation. number of discrete categories, such as eye color in hu- Encyclopedia of Biodiversity, Volume 4 Copyright 2001 by Academic Press. All rights of reproduction in any form reserved. 777 778 POPULATION GENETICS mans or shell color and pattern in the land snail Cepaea caused by genes of small effect (Crow, 1993). The net (Ford, 1975). It is often controlled by one or a few fitness of fully inbred Drosophila is only a few percent genes, and usually it involves relatively superficial traits of that of outbred flies. Even more extreme effects of such as color patterns. Only a relatively small propor- complete inbreeding are likely in vertebrates, which tion of the phenotypic variation of interest to evolution- have much larger genomes. The deleterious fitness ef- ists is of this kind. fects of inbreeding have probably played a major role in promoting the evolution of mechanisms of inbreeding 2. Quantitative Variation avoidance, such as the self-incompatibility loci of flow- Quantitative variation is all-pervasive. This can involve ering plants. either meristic traits, such as bristle number in Drosoph- ila, in which there is a large number of discrete catego- 4. Interpreting Phenotypic Variation ries, or continuously varying metrical traits such as body size. Variation in typical quantitative traits is The previously mentioned basic facts were established known to be under the joint control of environmental by the early 1950s and led to an active debate effects, accidents of development, and sets of genes concerning the causes of natural variation (Lewontin, whose individual effects are small relative to the total 1974). In one view, championed by H. J. Muller, range of variation in the traits (Falconer and Mackay, variation is mostly due to rare deleterious alleles 1996). Statistical methods that utilize the degree of maintained by mutation pressure at a large number of resemblance between close relatives enable the determi- loci; the coexistence of alleles at a locus at intermediate nation of the proportion of the total phenotypic varia- frequencies (polymorphism) is characteristic of only tion that is contributed by additive genetic causes—the a small number of loci. In the other view, advocated heritability, which controls the rate of response to selec- by Dobzhansky, polymorphism is the norm, and it tion on a trait (see Section II,D). Heritabilities for quan- reflects variation that is actively maintained by selec- titative traits typically are between 20 and 80%, corre- tion. In the absence of any means of identifying loci sponding to the fact that artificial selection is highly without the prior existence of genetic variability, no effective in changing the mean value of almost every unbiased survey of the extent of genetic variation at trait that has been examined (Falconer and Mackay, individual loci was possible with the methods of 1996). classical genetics, and so this question could not be answered. 3. Concealed Variability A more subtle form of phenotypic variation is con- cealed variability, i.e., variability that is only exposed B. Molecular Variation when homozygous genotypes are produced by close inbreeding. This is responsible for the increased varia- 1. Protein Electrophoresis tion among inbred lines when a set of such lines is The previously mentioned situation was transformed made from a random-bred base population and for by the development of molecular genetics. Gel electro- inbreeding depression, which is the decline in the phoresis of soluble enzymes and proteins provides a mean values of fitness-related traits such as viability rapid and simple method for surveying populations and fecundity with inbreeding (Falconer and Mackay, for variants affecting the structure of a large number 1996). Both of these phenomena reflect, at least in part, of different proteins and hence genes (Lewontin, 1974; the widespread occurrence of recessive or partially Hartl and Clark, 1997). The results of such surveys recessive rare alleles in random-mating populations reveal that a high fraction of loci coding for soluble (see Section II,F), whose phenotypic effects are only proteins are polymorphic in the sense of having at fully exposed when they are made homozygous and are least one rare variant whose frequency exceeds 5%; therefore not evident in randomly mating populations. the average individual from a randomly mating popula- In Drosophila, special breeding methods involving tion is typically heterozygous for a significant fraction the use of genetically marked chromosomes with inver- (several percent) of such loci. Despite some biases sions that suppress crossing over have shown that up in the methodology, particularly the inability of elec- to 50% of haploid genomes carry recessive lethal genes. trophoresis to detect many types of amino acid se- These contribute about half the inbreeding depression quence changes and the restriction of the method to manifested when fully homozygous genotypes are pro- soluble proteins, it is clear that protein polymorphism duced; a similar magnitude of inbreeding depression is is not an exceptional situation. POPULATION GENETICS 779 2. Measurement of Variation II. DETERMINISTIC POPULATION in DNA Sequences GENETICS The introduction of recombinant DNA technology has meant that population geneticists can now study varia- A. Allele and Genotype Frequencies tion at the level of the nucleotide sequence. Surveys of within-species DNA sequence variation of nuclear genes If we focus on a given nucleotide position, the basic have been most intensively carried out in Drosophila, descriptor of the state of a population is the set of but comparable results are emerging from other species frequencies of the four alternative states, A, T, G, and (Li, 1997). The basic conclusion is that variants due to C. If recombination within a gene is ignored, we can single nucleotide changes are the most abundant source consider the set of all nucleotide sequences observed of variation in natural populations. For silent substitu- at a locus as alternative alleles, whose frequencies char- tions in third coding positions, which do not change acterize the state of the population with respect to this the amino acid sequence, and for changes in introns and locus. Mendelian inheritance implies that this state is flanking sequences, the probability that two randomly not changed in the absence of evolutionary forces (the chosen alleles from a Drosophila population differ at a occurrence of intragenic crossing over and gene conver- given site (the nucleotide site diversity) is typically of sion at low frequencies means that in practice this is a the order of 1% or a few percent, depending on the good approximation rather than an exact description). species. The level of this type of variability is about 10 This is enshrined in the Hardy–Weinberg principle, times higher in the bacterium Escherichia coli and one- which states that the frequencies of diploid genotypes tenth as high in humans (Li, 1997). For most genes, in a random mating population with a set of n alleles иии diversity is much lower for replacement changes, which with frequencies p1, p2, , pn rapidly reach equilibrium alter the amino acid sequence. In addition to single values given by the multinomial expansion of ϩ ϩ иии n nucleotide polymorphisms, DNA variability is contrib- (p1 p2 , pn) . The importance of this result is uted by small insertions and deletions of sets of nucleo- that existing natural variation is preserved by Mendelian tides and by insertions of tranposable elements, mostly inheritance. This removes Darwin’s difficulties over the in noncoding regions. Other types of variability include rapid loss of variation under blending inheritance, variation in the sizes of tandem arrays of microsatellite which led him to adopt a theory of the inheritance of and minisatellite loci, which are often highly polymor- acquired characteristics for which there is no empirical phic and provide useful genetic markers (Bruford and foundation (Fisher, 1930).