The Evolution of Life History Traits: A Critique of the Theory and a Review of the Data

Abstract

In 1976, I reviewed (84) two models which give alternative explanations for the adaptation of life history traits to stable and fluctuating environments. Deterministic models (rand Kselection) predict that organisms exposed to high levels of densityindependent mortality, wide fluctuations in population density, or repeated episodes of colonization will evolve towards a combination of earlier maturity, larger broods, higher reproductive effort, and shorter lifespans than will organisms exposed to density-dependent mortality or constant population density (48, 65). Stochastic models (58, 77) predict the evolution of the same combinations of life history traits, but for different reasons: when fluctuations in the environment result in highly variable juvenile mortality, then a syndrome of delayed maturity, smaller reproductive effort, and greater longevity should evolve. Several years ago I set out to test these predictions by measuring the reproductive traits of two species of small fish that had been introduced to Hawaiian reservoirs in 1907 and 1922. Ambiguities appeared in the interpretation of the results, some of them inherent in the theory, others in the observations. I could not decide which of several possible causal systems had produced the pattern I observed. To determine what my results meant, I first tried to understand what life history data could mean in general, given the present state of our knowledge. In brief, the theory is not yet refined enough to be tested by crucial experiments that can pinpoint flaws. Under these circumstances, observation and experiment cannot falsify predictions definitively, but they can profitably arbitrate among the various simplifying assumptions that theorists may want to try out in their pursuit of unambiguous predictions that cleanly touch reality.