Naturwissenschaften (2000) 87:476–486 Q Springer-Verlag 2000 REVIEW ARTICLE Stephen C. Stearns Life history evolution: successes, limitations, and prospects Abstract Life history theory tries to explain how evo- natural selection could explain many of the facts of bio- lution designs organisms to achieve reproductive suc- logy. However, the mechanism for inheritance that he cess. The design is a solution to an ecological problem presented was not plausible, and he came under vigor- posed by the environment and subject to constraints in- ous attack on this point in the late nineteenth century. trinsic to the organism. Work on life histories has ex- When Mendel’s theory of inheritance was rediscovered panded the role of phenotypes in evolutionary theory, in 1900 and was shown, by Fisher (1930), Wright extending the range of predictions from genetic pat- (1931), and Haldane (1932) to be consistent with natu- terns to whole-organism traits directly connected to fit- ral selection, evolutionary biologists were relieved that ness. Among the questions answered are the following: a major deficiency in evolutionary theory had been re- Why are organisms small or large? Why do they mature moved. They not only adopted genetics as the central early or late? Why do they have few or many offspring? pillar of the Modern or Neo-Darwinian Synthesis Why do they have a short or a long life? Why must they (Dobzhansky 1937; Mayr 1942); they also committed to grow old and die? The classical approach to life histo- genetics as the center of their explanatory paradigm. It ries was optimization; it has had some convincing em- became the reference point for rigor in the field. pirical success. Recently non-equilibrium approaches The concentration of evolutionary biologists on ge- involving frequency-dependence, density-dependence, netics has lasted almost a century and continues to evolutionary game theory, adaptive dynamics, and ex- yield important results. Despite some claims to the con- plicit population dynamics have supplanted optimiza- trary, however, evolutionary genetics established the tion as the preferred approach. They have not yet had consistency, not the sufficiency of genetics plus natural as much empirical success, but there are logical reasons selection to explain evolution – particularly the evolu- to prefer them, and they may soon extend the impact of tion of whole-organism traits (phenotypic evolution), life history theory into population dynamics and inter- evolutionary developmental biology, and paleontology. specific interactions in coevolving communities. The concentration on genetics therefore elicited a pre- dictable reaction: What is the role of phenotypes in evolution? This reaction started to gain momentum in Introduction the 1960s and 1970s. It has both a selectionist part, which is concerned with how phenotypes are designed Where life history theory fits in the history of ideas for reproductive success, and a developmental part, which is concerned with the restrictions placed on the When Darwin published the “Origin of Species” in expression of genetic variation by developmental mech- 1859, he showed that descent with modification and anisms. The selectionist part of the phenotypic reaction, S.C. Stearns which encompasses behavioral and evolutionary ecolo- University of Basel, Zoology Institute, Rheinsprung 9, gy, has developed as a theory-driven, predictive, experi- 4051 Basel, Switzerland mental enterprise. This contrasts it with those parts of e-mail: stephen.stearns6yale.edu evolutionary biology that are historical, retrospective, Tel.: c1-203-4328452 and descriptive. Both approaches have strengths and Present address: weaknesses. The kinds of information that they yield S.C. Stearns, Department of Ecology and Evolutionary Biology, are different and sometimes complementary. Osborn Memorial Laboratories, PO Box 208106, To reiterate, life history evolution is part of evolu- Yale University, New Haven, CT 06520–8106, USA tionary ecology, which is itself part of the more general 477 attempt to explain phenotypic evolution. The develop- logists want to discover the mechanisms common to all mental part of the explanation, not discussed here fur- living things; they focus on what is similar and general ther, includes evolutionary developmental genetics and in all cells. Physicists want to discover the general prop- morphology, often now referred to as “Evo-Devo”. erties of matter and energy, laws valid at all places and all times. Evolutionary biologists ask questions inspired The nature and status of the classical explanations by comparisons of differences, differences between spe- of life history variation cies, between populations, between individuals. They want to understand why things are different, not why Classical life history theory, summarized in Roff (1992) they are the same. They want to understand what and Stearns (1992), is based on optimization models. It causes diversity. Much of their thinking is colored by aims to explain variation in size at birth, growth rates, this concentration on the causes of variation. age and size at maturity, clutch size and reproductive investment, and mortality rates and lifespan. By the 1990s the field had achieved consensus on the general The assumptions of optimality theory features of a plausible explanation of the evolution of To treat life history evolution as an optimality problem, life history traits: (1) life histories are shaped by the in- one assumes a definition of fitness, defines a relation- teraction of extrinsic and intrinsic factors, (2) the ex- ship between traits and fitness, describes tradeoffs be- trinsic factors are ecological impacts on survival and re- tween traits, then finds the combination of traits that production; (3) the intrinsic factors are tradeoffs among maximizes fitness. Optimality theory usually assumes life history traits and lineage-specific constraints on the that mortality and fecundity rates are constant and that expression of genetic variation. the age structure of the population is therefore stable, Thus the classical approach makes a strong simplify- which allows tractable measures of fitness to be de- ing claim: you only need to understand two things to fined. It also assumes that the fitness of a phenotype is understand the evolution of life histories. One is exter- adequately measured by estimating its reproductive nal: how the environment affects the survival and re- success in a population composed exclusively of identi- production of organisms of different ages, stages, or cal phenotypes. Fitness is measured as the rate at which sizes. The other is internal: how traits are connected to a population of identical phenotypes would grow. Note each other and the constraints on how traits can vary. that the assumption of identical phenotypes implies As we will see below, there are limitations to the classi- asexual reproduction or perfect heritability. cal approach that theorists are currently exploring, but The success of the classical theory suggests that the all modern theory builds upon the focus achieved by things on which it chose to concentrate – age and size this simplifying claim. specific impacts on mortality and reproduction in a sta- Like many simplifying claims, it is easy to think of ble, asexual population – were more important for its reasons why this one may be false. For example, the purposes than the things it chose to ignore – such as manner in which limited resources are acquired and al- genetics, explicit population dynamics, and frequency located to survival and reproduction varies with type of dependence. organism. This means that there is an intrinsic phylo- genetic or developmental component to the way the ex- trinsic factors interact with the organism, and casts in The conceptual advantages of reaction norms doubt our ability cleanly to separate the extrinsic from A reaction norm is a property of a genotype. It de- the intrinsic. This comment can be translated into a re- scribes the set of phenotypes produced by that geno- search program, but it would not have been appro- type across a range of environmental conditions. Reac- priate to do so at the time that life history analysis was tion norms for life history traits have at least two im- being clarified. Then it was important to make the sim- portant characteristics: plifying assumption and to push on, undaunted by the 1. They clearly distinguish between the effects of “na- complications that were being ignored, precisely to see ture” and “nurture,” between the component of the how much could be explained without worrying about reaction to the environment that has evolved over them. many generations and the component that is due to Before describing how well that classical approach a developmental reaction of this particular organism has succeeded in explaining the major life history traits, to this particular environment in just this genera- I first discuss some important background: the focus of tion. evolutionary biologists on explaining variation, the as- 2. They can be predicted by theory. This extends the sumptions of optimality theory, the conceptual advan- range of things that the theory can predict and thus tages of reaction norms, and bet-hedging as a method makes the theory easier to test and to improve. of dealing with risk. The evolutionary emphasis on explaining variation Bet-hedging to deal with risk Evolutionary biologists differ from molecular biologists The classical theory did make one excursion away from and physicists in an important basic aim. Molecular bio- stable populations living in constant environments. It 478 dealt with the problem of the risk of reproductive fail- ductive event. One benefit of later maturation is a long- ure. Evolutionary risk is variance in fitness: increases in er period in which to grow, leading to a larger size at variance reduce fitness. Measuring fitness as the geo- maturity and greater fecundity, for fecundity often in- metric mean of per-generation reproductive success creases with body size. Another benefit of later matura- properly accounts for long-term risk, for this measure tion can be the production of higher-quality offspring, takes into account the effects of variance.