“Ex ovo omnia” wrote Harvey (1651), and more than three centuries later his dictum still stands. Many examples of vegetative reproduction in animals have been described since Trembley (1744) astonished the world by describing the asexual budding of Hydra, but all lineages, so far as we know, pass through a single-cell stage sooner or later. The only more certain observation is that the individual that develops from the egg will eventually senesce and die. The journey between the two fixed points of egg and corpse has no prescribed route, however, and biologists have struggled to discover and interpret the lives of animals, plants, fungi, and seaweeds since the beginning of scientific biology in Harvey's time.
In very broad terms, two main approaches have been followed. The first is concerned largely with juvenile forms, and specifically with discrete developmental stages such as larvae. This is the older school, and its subject is usually called “life cycles.” The second is concerned largely with adults, and specifically with the quantitative schedule of reproduction. This has developed over the last 50 years, and its subject is usually called “life histories.” The two have been developed to a large extent independently of one another, and the linkages between them remain few and weak.
The study of life cycles is the older tradition, and dates back to the time when it was first demonstrated by Steenstrup (1845) that very dissimilar individuals could be produced, one from another, as a succession of forms belonging to the same lineage. The only example that is familiar to everyone (now that classical zoology has withered from the curriculum) is the succession of polyp and medusa in some cnidarians, but much more complex life cycles have evolved in groups such as digenean trematodes. It is often difficult to work out how and when one stage gives rise to another, especially when this involves reproduction by eggs. In other cases, however, the developmental sequence is clear even though it passes through larval stages that are very different from the adult, by virtue of the physical continuity of macroscopic individuals. There is no profound difference between the two, however. The juvenile starfish, for example, develops as a miniature version of the adult from a small patch of tissue within the body of a small ciliated-band animal living in the plankton. As the starfish animal enlarges, the ciliated-band animal shrinks, and its remnants are eventually discarded. This process is called “development” because the physical continuity of larva and adult is clear, but it is manifestly the same kind of process as the transition between polyp and medusa, a succession of morphologically distinct phases within a single lineage.
The study of life cycles has been predominantly descriptive, a tradition that has continued, in the form of discovering the regulatory genes responsible for the evolution of body plans, down to the present day. The possibility of a theoretical account of the life cycle was raised by Garstang (1928) nearly a century ago: given that an animal developing from an egg must grow steadily larger, it must be functional first as a small individual and subsequently as a large individual. A ciliary-band animal living at low Reynold's number in the plankton is one possible route to a large hydraulically powered animal living on the sea floor, whereas the reverse route is impracticable. Nevertheless, a systematic theoretical framework capable of interpreting the succession of phases in development has yet to be constructed.
(p.vii) The parallel effort to understand the sexual phases of life cycles is a similar but more complicated story, largely because of the great difficulty of establishing the site and timing of the crucial events of fusion and reduction, even after the advances of microscope design in the later nineteenth century. It was not until the closing decades of the century that the essential distinction between spore and gamete was unequivocally established, and by the end of the first decade of the twentieth century a correct account of the alternation of generations in the life cycle of seaweeds and land plants had been successfully accomplished (see the review by Farley 1982). A theoretical basis for understanding the alternation of generations in terms of the fundamentally different requirements of spores and gametes, and thus the different structure and behavior of spore-producing and gamete-producing individuals, was provided at the same time (Bower 1908). Before it could be extensively developed, however, the field was largely abandoned, swept away by the flood of research into transmission genetics, which had just been provided with the firm theoretical framework of Mendelism. The endlessly varied sexual cycles of eukaryotes have continued to provide rich material for extending our knowledge of natural history, but we cannot yet interpret them within a consistent theoretical framework. In recent years there has been a modest revival of interest in the theory of phenomena such as gamete dimorphism, but this has not yet become firmly coupled to field studies and experimentation.
The field of life histories has developed more recently and in a very different fashion, being highly theoretical from the outset. Simplifying history (as one must in a foreword) it was galvanized by Cole's paradox: a lineage in which females live forever and produce an infinite number of litters of whatever size has the same rate of increase as one in which all females die immediately after producing their first litter, but produce one more offspring (Cole 1954).
Working out the reasons for this counter-intuitive result led to a general interpretation of suicidal versus repeated reproduction that was subsequently elaborated into an account of the schedule of reproduction over the whole of the adult stage. Juvenile stages such as larvae are ignored; alternative adult stages such as sporophytes and gametophytes are not distinguished. Freed from zoology, botany, and genetics a highly abstract and general theory of the life history could be developed.
Much of this theory was based on a principle of optimality: quantities such as the rate of reproduction and the probability of survival are negatively correlated, such that intermediate values of both maximize the overall rate of increase of a lineage. This enabled the schedule of reproduction to be predicted from the costs of reproduction, in terms of reduced growth or survival. This approach has had some brilliant successes, beginning with David Lack's classical interpretation of clutch size in birds (see Lack 1966). It could also be extended to the puzzling phenomenon of senescence, which could now be interpreted as a non-adaptive side effect of selection for early reproductive maturity (Williams 1966a). Moreover, the generality of the predictive framework made it possible to contrive experimental tests in laboratory model systems, which had never been possible for life cycles.
The very generality that gave the theory such power was also a source of weakness, insofar as the sources of the costs of reproduction on which the theory was based did not need to be specified and therefore could not be investigated within the confines of the theory itself. For this reason, attention began to shift towards the nature of the costs themselves. This is not as straightforward as it might appear. It seems obvious that producing larger offspring will mean producing fewer, and almost equally obvious that allocating a greater share of resources to reproduction must deplete the stock available to support maintenance and defense. One function necessarily interferes with the other. It was soon found, however, that in practice the correlations between fitness components are usually positive rather than negative. This was quickly attributed to environmental variance of productivity among sites when comparisons are made in the field, or to genetic variance of overall fitness among strains when comparisons are made in the laboratory. The expected costs are then expressed only at evolutionary equilibrium, when genetic variance for overall fitness has been reduced to a low level by selection. This evolutionary argument, however, implies that costs of reproduction measured as (p.viii) negative genetic correlations need not have any mechanistic basis in terms of functional interference. A simple illustration of this conclusion is to generate n random numbers and label them x1, x2 etc; then repeat the process to obtain a second set labeled y1, y2 etc. Plotting yi on xi produces a cloud of points with zero correlation. Now choose the small percentage of pairs with the highest values of (xi + yi); plotting yi on xi now produces a graph with a slope of –1 (if the xi and yi have equal variance) and r2 ≈ 0.5. This striking pattern has been carved out of the original unstructured data by the act of choosing an unrepresentative set of cases, and natural selection will likewise generate negative genetic correlation among components of fitness from random life histories through the propagation of an unrepresentative set of genotypes. No causal connection between the components, for example through pleiotropic gene expression, is necessary for such correlations to arise.
Understanding the mechanistic basis for costs of reproduction is thus an important aspect of understanding life histories as a whole. Moreover, it may contribute directly to human well-being. The evolutionary reasons for senescence can be framed in terms of pleiotropy or delayed gene expression for example, whereas the physiological reasons must be framed in terms of factors such as the irreversible damage caused by reactive oxygen species, the accumulation of somatic mutations, the intrinsically limited metabolic capacity of tissues, and so forth. Much of the recent research into senescence in the Drosophila and Caenorhabditis model systems has been driven by the physiological agenda, in part because of its potential for identifying palliative therapies. The value of this research is indisputable. It would be a pity, however, if the evolutionary agenda were to be obscured or lost as a result. Part of the raison d’être of this volume is to emphasize that it is important to understand the physiological basis of the costs of reproduction in order to understand how life histories evolve—but conversely that it is equally important to understand how life histories evolve in order to predict how physiological processes are likely to operate. The clk genes of Caenorhabditis, which slow down vital processes and extend lifespan, are an excellent example of recent research in molecular developmental genetics at the interface between evolution and physiology (Hekimi et al. 2001).
A more fundamental task for the future is to build a synthetic evolutionary theory of development that would bring together all the phenomena of life cycles within the same framework. It is not even clear whether this is possible. The classical theory of life histories invokes natural selection, leading to optimal phenotypes. The most successful theories of certain aspects of the life cycle, such as gamete dimorphism and the sex ratio, invoke sexual selection, leading to evolutionary stable states. For some phenomena, such as the alternation of generations, both approaches have been tried without any decisive outcome so far; for others, such as the succession of phases, hardly any formal theory has yet been developed. But the possibility is worth contemplating, and if this volume is not the last word on the subject, it will at any rate be the next word.
23 May 2010