>Darwinian evolution optimizes a lot of system's degrees of freedom, among
>other things (localized) mutation rates and generation durations. (It also
>is a fount of contraintuitive nonlinearities, but that's another story).
>The trivial argument is that immortal individuals are almost instantly
>outperformed, especially in a Red Queen environment (you have to run like
>mad just to stay in the same place, and twice as fast if you want to get
>to somewhere else).
Trivial; but also not supported. The primary distinction of lifespan
arises from reproductive costs, population growth rates, and the probability
of adult survival. (Futuyma, "Evolutionary Biology" p. 273-273) Predation,
which is hard to evolve for, generates the same results as disease and
parasitism, which are generally easier to evolve for.
Where the hazards of reproduction and the growth rates
are low and the probability of adult survival high (e.g. trees) populations
reproduce continuously over their adult lifespan. Such populations evolve
to live as long as possible - there is virtually no selectable variation
for lifespan. (M.R. Rose "Laboratory Evolution of Postponed Aging" Evolution
38: 1004-1010) What selectable lifespan variation remains is strongly tied
to negative things, primarily late reproduction.
Where the hazards of reproduction and the growth rates
are high and the probability of adult survival low (e.g. salmon) individuals
reproduce only once. Here evolution does indeed tune lifespans for optimal
reproduction (sorry, no references in the house). Such individuals, however
don't age; they die in the prime of life, usually from hormonal death clocks.
If this premature death is prevented, the creatures usually live greatly
extended spans, displaying typical aging features.