The Costs of Reproduction and Their Consequences

This paper identifies the effects of different types of reproductive cost on the pattern of life histories. By reproductive cost I mean the generally deleterious effect of present reproduction on future survival or fecundity or both. My approach is first to discover the circumstances in which reproduction at a given age should be maximal, and then to investigate the conditions under which reproduction at given age should be zero; this leads to a discussion first of semelparity and later of the optimal age at maturity. When all entries in the life table are independent of one another there is no reproductive cost, and it can be shown that semelparity is unlikely to evolve if maturity is delayed or if prereproductive survival is poor. The existence of a number of conspicuous exceptions to these rules points to an inadequacy in the model. Once reproductive costs are introduced, reproduction will be optimized rather than being merely maximized. The "survival cost" is the decrease in the rate of adult survival which accompanies a given increase in fecundity. If the graph of fecundity on survival is linear, semelparity will be an optimal life history provided that the slope of the graph, and thus the magnitude of the survival cost, does not exceed a quantity which can be defined mathematically. More generally, the graph of fecundity on survival will be curvilinear. If it is convex upwards there will be a stable equilibrium at which a given combination of survival and fecundity maximizes fitness; if it is concave upward there will be a point of unstable equilibrium at which some combination of survival and fecundity minimizes fitness. Semelparity may evolve either through selection toward a stable equilibrium or through selection away from an unstable equilibrium. The latter route seems to be responsible for the evolution of semelparity in some North American salmonids. The same principles continue to hold in fluctuating environments, but the evolution of semelparity is further favored if variation in prereproductive survival is less than variation in adult survival. Delayed maturation, either in semelparous or in iteroparous species, may be favorably selected given that a survival cost exists, if fecundity increases with age (the "actual fecundity cost"). This idea is supported by the facts that poikilotherms usually mature later in life than homoiotherms, that females mature later than males among poikilotherms, and that males mature later than females among homoiotherms. A necessary but not sufficient condition for the existence of an optimal age at maturity is that present reproduction should cause a decrease in potential future fecundity (the "potential fecundity cost"); a sufficient condition is that the potential fecundity cost should decrease with age relative to the actual fecundity cost. Potential fecundity costs appear to be widespread amongst poikilotherms, and their general use in interpreting ages at maturity is explained. Methods of finding the optimal age at maturity are outlined, and the estimated optima are compared with the mean ages at maturity in two poikilotherms. Finally, I argue that experimental tests of life history theory are not yet feasible, and that we must instead rely on comparative techniques. After criticizing a number of recent comparative studies on methodological grounds, I suggest areas in which further research should be undertaken, and conclude by stressing the importance of life histories to general evolutionary theory.