The Princeton Guide to Evolution

Natural Selection and Adaptation

Douglas J. Futuyma

Natural selection is the centerpiece of Darwin’s great book, and is prominent in its title: On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. In fact, this book was a hastily written “abstract” of a book he had started, and intended to be much larger, titled simply Natural Selection. To be sure, Alfred Russel Wallace independently conceived the idea, but it was Darwin who deduced its many implications and followed its ramifications in detail, and with whom the concept is usually, and rightly, associated. Natural selection is the most important of Darwin’s many original ideas. It is one of the most important ideas in the history of the world.

Why? Because for the first time, there existed a purely scientific explanation of the most powerfully impressive examples of design and purpose in nature: the features of organisms that equip them most exquisitely for survival and reproduction. The adaptations of organisms—including humans—had long been attributed to the Creator’s beneficent design, indeed were among the most important arguments for the existence of such a supernatural Creator. As an alternative explanation lacking the slightest supernatural tinge, natural selection made biology a science. Philosopher Daniel Dennett (1995) has nominated natural selection as the best idea in history—and has also termed it “Darwin’s dangerous idea,” for it threatened the theological substructure of much of Western philosophy and has many ramifications outside biology.

Natural selection occurs whenever there is a consistent, average difference in fitness (reproductive success) among sets of “individuals” that differ in some respect that we may refer to as phenotype (see chapter III.1). Most (but not all) evolutionary biologists would add that the phenotypic difference is at least partly inherited. If this is the case, the difference in reproductive success may result in one phenotype increasing in frequency while another decreases. The eventual outcome may be complete replacement of one phenotype by the other. Throughout On the Origin of Species and most subsequent evolutionary discourse, the “individuals” are individual organisms, such as dark versus pale moths or people with different blood types. However, the theory also applies, mutatis mutandis, to other kinds of biological “individuals” that produce more such “individuals”: genes, genotypes, cell types within an organism, different populations of the same species, or different species. Thus, selection can act at different levels of biological organization (see chapter III.2).

In order to establish that there exists an average, nonrandom difference in fitness, we must estimate reproductive success of a number of individuals in each class, for we cannot tell whether a difference between two different individuals is caused by, or even correlated with, their phenotypes; one may have just been luckier. Luck—the random, unpredictable element in frequency changes due to sampling error—is termed genetic drift (see chapter IV.1). Natural selection, in contrast, is the nonrandom component of variation in reproductive success. A higher proportion of yellow than of brown land snails in a pasture might be trampled by cattle, just by chance, but if the grass is mostly yellow rather than brown, we can predict with considerable confidence that brown snails will suffer higher predation from thrushes, which use color vision to find prey. The thrushes, but not the cattle, act as an agent of natural selection, which is the antithesis of chance.

It is extremely important to recognize that natural selection is not an agent, and certainly not remotely similar to a rational agent; casual talk of “Mother Nature” or simply of “nature” as a personified entity (as in “nature tends to…,” or “nature selects” or “nature has found a solution”) is misleading. It is even misleading to say, “selection acts at different levels of biological organization” (as in the previous paragraph!), because selection does not “act.” Natural selection really is no more than a statistical difference in reproductive rate, often owing to some mechanistic relationship between a property of the phenotype and some feature of the environment. A common result of the difference in reproductive rate is that one type replaces others. That’s all there is to it. When we personify natural selection (as almost all biologists do at times, including Darwin, for the sake of comfortable, less stilted discourse), we can easily slip into describing selection as if it could plan, as if it had the species’ welfare in mind, as if it were beneficent—or cruel. Such modulation from one concept to another has resulted in frequent misunderstanding and misrepresentation of natural selection and evolution. For example, many writers have referred to natural selection as if it had a goal, such as perpetuation of the species or “evolutionary progress”; but selection cannot have a goal of any kind. Natural selection results in the evolution of characteristics that enhance the survival of the individual organism (or of its genes), characteristics that we might metaphorically call “selfish”; indeed, characteristics that promote “cooperation” among organisms call for special explanations, such as kin selection, in which cooperation with or aid to a related individual increases the frequency of their shared genes, including the genes that underlie the propensity to cooperate or aid (see chapter III.4, also chapters VII.8–VII.10). But natural selection itself is not selfish—or cruel, or kind—except in a purely metaphorical sense. Above all, natural selection is not a normative “law of nature,” prescribing right or ethical conduct. The behavior that may have evolved by natural selection holds no prescription for moral or ethical human conduct.

Among the levels at which natural selection can occur is the level of the individual gene or DNA sequence (see chapter IV.7). This occurs whenever different sequences make different numbers of copies that are transmitted to the next generation. For example, transposable elements proliferate within the genome at higher rates than “normal” genes. Natural selection can also occur at the level of species, for certain characteristics enhance the rate of origin of new species or diminish the likelihood of species extinction (see chapter VI.14). For instance, the number of species in lineages of herbivorous insects has generally increased faster than in closely related lineages that have other feeding habits. Neither gene selection nor species selection has molded the advantageous characteristics of individual organisms; rather, they have affected properties at the gene level or at the species level. But individual selection, selection among individual organisms within populations, is at the center of evolutionary theory. It is at this level that selection explains most of the adaptive features of organisms. An alteration of a bird’s beak caused by a gene mutation is a feature of an individual bird, not a feature of the gene or of the entire species, except insofar as more or fewer individuals of the species possess it. And it is the process of individuals’ birth and death rates that alters the frequency of the mutated gene, on the one hand, and the character of the entire species, on the other.

Understanding the dynamics and consequences of natural selection requires familiarity with theory, as framed in terms of population genetics. Most population genetic models of selection consist of (1) a postulated relationship (mapping) between genotype and phenotype, (2) a frequency distribution of genotypes, and (3) a postulated relationship between phenotype and fitness. Together, these determine the frequency distribution of genotypes and phenotypes in subsequent generations. For example, if body size differs among the three genotypes (A₁A₁, A₁A₂, and A₂A₂) at a locus with two alleles in a sexually reproducing population, the outcome of selection on body size depends, first, on whether or not the heterozygote (A₁A₂) is intermediate in size between the two homozygotes (i.e., on the mapping between genotype and phenotype). The rapidity of change depends on the frequency distribution of the phenotypes; all else being equal, evolution under selection is faster if all the genotypes are common than if one homozygote makes up most of the population. And the final outcome depends on the “mode” of selection, the relationship between fitness and phenotype. If largest individuals are most fit, the outcome is fixation of the A₁A₁ genotype if the heterozygote is intermediate in size, but stable maintenance of variation (polymorphism) if the heterozygote is largest. Some relationships among genotype, phenotype, and fitness are more complex, resulting, for example, in diverse, historically contingent outcomes that depend on initial conditions.

Much of this theory is cast in terms of frequencies of genotypes and alleles (see chapter III.3), and is readily applied to genetic data (e.g., DNA sequence variation) and to phenotypic data when the phenotypes are distinct categories (e.g., black and white); however, the variation in most phenotypic traits (e.g., human stature) is continuous, or gradual, owing to variation at many gene loci, as well as influences of environmental variables on growth and gene expression. In this case, the effects of individual genes are difficult to discern, so the frequencies of alleles and genotypes are very difficult to measure. Evolution of such traits is most often analyzed by the statistical tools of evolutionary quantitative genetics, which are founded on population genetics (see chapter III.3). These statistical tools are mostly variances, covariances, and correlations, which are used to partition variation in a character into its genetic and nongenetic components, and to characterize the degree to which different characteristics are inherited together. The genetic variance in a character plays a major role in its “response” to natural selection (see chapters III.5 and III.6), and the correlations among characters affect the extent to which characters can evolve independently. Consequently, the theory and analytical methods of population genetics and quantitative genetics are often indispensable both for predicting and studying evolutionary changes caused by selection, and for understanding the limits of natural selection—the ways in which species may fail to adapt (see chapter III.8).

The genetic theory of natural selection is often useful—although not always indispensable or easily applied—for analyzing and understanding adaptations, which many biologists would define as features of organisms that have evolved by natural selection. It is the adaptations of organisms, often so exquisitely suited to particular tasks or ways of life, that “so justly excite our admiration,” as Darwin wrote, and that constituted the argument for supernatural design before Darwin demolished it. The study of adaptations pervades much of biology and is at the core of many chapters throughout this book. For this section, we have assembled a set of chapters treating the adaptive evolution of several classes of characteristics that have been the focus of extensive evolutionary research. The evolution of biological systems (especially modes of reproduction and inheritance; see chapter III.9) concerns such questions as why some organisms reproduce sexually and others asexually. A reaction norm (see chapter III.10) is the variety of phenotypes that a genotype may have, depending on environmental conditions (e.g., slender or obese as a function of diet). Often the reaction norm describes phenotypic plasticity, the capacity for nongenetic, advantageous modifications of the phenotype. Why characteristics are or are not phenotypically plastic is a major focus of this chapter. Chapter III.11 treats the evolution of the great array of life histories of organisms, focusing on such features as reproduction (why do some species have so few, and others so many, offspring?), generation time (why do humans, periodical cicadas, and century plants take so long to reach reproductive age, compared with most other species?), and life span (why does maximum life span differ among species? Why do we grow senescent and die?).

Much of biology is concerned with analyzing the function of phenotypic traits such as anatomical or cellular structures, a study closely related to the physiological and biochemical processes inherent in living systems. The raison d’être of morphological and physiological traits, their adaptive advantages, is a traditional field of evolutionary biology, although the questions and research methods continue to change (see chapters III.12 and III.13). In particular, traditional analyses of the function of an anatomical or physiological characteristic may be joined with studies of the ways in which variation in the trait affects fitness in natural populations, or with genetic analyses of variation in the trait, or with paleontological and phylogenetic studies of its history of evolutionary change.

Many phenotypic traits are adaptations to environmental factors, including other organisms. Evolutionary biology has been intimately associated with ecology ever since Darwin, who may as justly be called the first great ecologist as the first great evolutionary biologist. This section’s chapter on the evolution of ecological niches (see chapter III.14) concerns the conditions under which species might evolve broad or narrow (specialized) tolerance of abiotic environmental conditions (e.g., temperature), habitat use, or diet. The species’ ecological niche, in turn, influences its interactions with other species—its biotic environment (see chapter III.15). Adaptation to other species, such as competitors, prey, predators, parasites, and mutualists, has greatly shaped many of the features of species and been of paramount importance in the evolution of biological diversity. These chapters provide a foundation for several other chapters in the guide, especially in Section VI: Speciation and Macroevolution.