There are traditional wisdoms in the study of the evolutionary origins of aging. For example that greater extrinsic mortality due to predation or an otherwise harsh environment selects for shorter life spans, producing species whose individual members are optimized to reproduce rapidly and age rapidly, as health assurance mechanisms that create longer reproductive and overall life spans are either lost or never evolve in the first place. This and other consensus theories initially emerged from simple models that have nonetheless largely continue to do fairly well over the years, but all simple models are eventually challenged. With the falling cost and vast increase in available computing power ever more sophisticated evolutionary modeling has taken place, and researchers are finding that there can be exceptions to almost every hypothesis in the field.
Here is an interesting paper in which researchers propose that under some circumstances high extrinsic mortality can result in species with longer lives, not shorter lives, and further that aging may have evolved because it actually increases lifespan in a species living in a high mortality environment. These are not the only researchers producing models in which this sort of thing happens. The argument here is that aging and improved youthful ability are linked: the evolution of capabilities that improve survival and reproductive success in the face of adversity in early life goes hand in hand with an inability to maintain tissues and metabolism over the long term. Thus individuals better survive their hostile environment and live longer on average, but age as a consequence.
Given an extrinsic challenge, an organism may die or not depending on how the threat interacts with the organism's physiological state. To date, such interaction mortality has been only a minor factor in theoretical modeling of senescence. In general, it holds that mortality does not affect evolution if it affects all organisms equally. The intuitive reason for this is that evolution favors a phenotype (strategy) if it is better at propagation than other strategies. If all strategies are affected equally, no strategy improves relative to others, and selection gradients remain unchanged.
Mortality that does not distinguish between individuals is often called 'extrinsic mortality' and modeled as an age-independent parameter in the mortality function of age-structured models. In these models, extrinsic mortality is a discounting factor in the survival function that cannot be molded in any way by the (fictitious) organism that is studied. However, whether environmental threats result in mortality depends on the interaction of those threats with an organism's physiological state. By adjusting its state, an organism can influence death from environmental causes.
To investigate mortality-environment interactions from a theoretical perspective, we model a trade-off between an age-independent and an age-dependent mortality term. As an example of a biological rationale for such a model, [it has been] suggested that it could be beneficial from an evolutionary standpoint to attain a state that is unmaintainable by its very nature, causing mortality to be low at young ages, but to increase over time [as senescence takes hold]. As a result, death can be postponed to later ages, depending on the magnitude of initial reduction relative to the ensuing increase in mortality with age.
We find that depending on the physiological constraints, any outcome is possible in any environment, be it 'no senescence' or 'high rate of senescence'; that the highest optimal rate of senescence emerges for an intermediate physiological constraint; and that the optimal rate of senescence as a function of the environment is driven by the way the environment changes the effect of the organism's state on mortality. We conclude that predicting the outcome requires knowledge about the interaction of the environment and the organismal physiology: separately, these have little predictive power.
We propose, perhaps paradoxically, that senescence may have evolved because it extends lifespan. Lifespan is equal to the inverse of average mortality. If mortality increases over age, but [for a senescent species] starts off from a much lower level than would otherwise be the case [for a non-senescent species], average mortality may go down, implying lifespan extension.