I can understand why some prey can't outrun a recently evolved species. However, since cheetahs have existed for so long, why haven't its prey evolved to always outrun it, driving cheetahs to extinction. Is it because the cheetah population size is so much smaller than the population size of most of its prey species that its prey species are under very weak natural selection to outrun cheetahs? Is it that those that can run faster have more other biological costly traits?
There are (at least) three important factors to consider here; evolution under selection requires genetic variation upon which to act, selection can act on covarying traits causing trade-offs, and adaptation also occurs in the predator. A lot of this is covered elsewhere on this site (including the effects of the other mechanisms of evolution), but little specific reference is made to predator-prey co-evolution.
Adaptation and genetic variance
One of the mechanisms of evolution is selection, and adaptation occurs when species evolve as a response to selection. For a response to selection to occur there must be genetic variation within the trait, such that the genes an individual carries affect the individuals fitness. For example, there may some genes which give the carrier better muscle structure for fast running, and this increases survival which, in turn, increases reproductive output. The importance of genetic variation is often overlooked, but is highlighted by the breeders equation. Genetic variance can arise by novel mutations, or can exist as standing genetic variance.
Selection on covarying traits
Selection is seldom, if ever, a univariate process. That means fitness is not determined by a single characteristic, so fitness in a prey species may be determined by the speed at which it can run, but also its metabolic rate, stamina, ability to acquire nutrients, the way it provisions nutrients for growth and repair etc.. Genetic covariance between traits (induced by linkage or pleiotropy) can impact the response to selection, because any selection acting on a covarying trait can cause affect the adaptation occurring in the focal trait (strengthening, weakening, neutralising, or even reversing the direction of response). Therefore, if running speed covaries with other traits it may be difficult for selection to increase running speed. Think about athletes, I doubt Usain Bolt could run 10,000 metres as fast as Mo Farah and vice versa, because there is a trade-off between speed and stamina.
Adaptation in other species
If selection did cause speed to increase in the prey species then that would strengthen selection for increased speed in the predator. This is called an evolutionary arms race, where the adaptive the evolution of two species interacts causing adaptation and counter-adaptation. For example, genes which allow faster running might spread through a population of gazelles (prey) because carriers are more likely to outrun the lion (or the other gazelles), but this will increase selection on lions (predator) to increase speed (or adopt other strategies) which will spread genes to make the lions faster (use other strategies); the result being adaptation and counter-adaptation between the gazelle and lion populations.
There is a little series here about Butch Brodie and his studies of coevolutionary arms races between garter snakes and toxic newts. It would be good to read on the Red Queen Hypothesis, which describes how species, under interaction with other species, must continue to evolve just to prevent extinction. Also worth noting that prey species often have more than one predator species (and predator species often have more than one prey); interaction and coevolutionary networks are complex. What may be adaptive in some interactions may also be maladaptive for other interactions (see specialist and generalist, and selection on covarying traits).
Evolutionary trade-off describes situations where one trait cannot increase without a decrease in one or more others. Some hypothetical examples:
- longer legs may help run faster, but past a certain point, it will increase the risk of injury, decreasing survival;
- lower body weight may increase top speed, but past a certain point, it will decrease starvation tolerance;
- more muscle may help acceleration but will increase energy requirements.
All changes have costs and benefits. In situations where the outcome of an event is simply "success/failure", there is therefore an evolutionary incentive to evolve to be just good enough. Once you are good enough getting better only increases the cost (I am oversimplifying this a bit).
Change in an extrinsic driver will change the balance of cost and benefit, shifting evolutionary pressure. For example, when predators are absent (island populations) birds sometimes become flightless, because one benefit of flight (escape from predators) no longer applies.
The interesting bit happens when the "extrinsic driver" is another living thing that is also capable of evolving. In this case you suddenly get an evolutionary arms race where each side is constantly subject to a selective pressure to be slightly better than the other, which is a moving target. You therefore get an arms race situation (or the extinction of one or the other side). The Red Queen hypothesis is named after a quote by the Red Queen in "Alice through the looking glass" (Carroll, 1871):
Now, here, you see, it takes all the running you can do, to keep in the same place.
Using the example of a lion and a gazelle: lions run fast enough to catch and eat the slowest gazelles. The remaining gazelles are on average faster, for whatever reason(s). Some of those reasons will be heritable and the next generation of gazelles will be faster. The slowest lions will starve, and some of the reasons for the remaining lions being faster will be heritable so the next generation will be a bit faster, and then you're back where you started. Rinse and repeat.
Predation obviously operates on a much faster timescale than selection, so this doesn't always occur (putting a fox in a chicken coop won't evolve fast chickens).
Coevolutionary "arms races" can be seen in predator-prey interactions, mimicry and much more (including for example the evolution of sex, but that's off-topic for this answer).
Predators always have to be much better hunters than the prey - they must eat every few days after all. But they can only get so good.
Predator/prey population balance will tend to look like a competition where if the predators are too efficient they will kill off the prey. If that happens they start to starve to death.
If the prey outrun the predators (or at least escape all the time) then the predators will starve to death. Then they breed until there are so many that they eat all the grass/vegetation and then they die off.
While both of these have certainly happened in natural history what is more stable for predators and prey to evolve in competition with each other such that their populations look like a stable equilibrium. If not, one of the would just disappear. Then later via migration another animal would come in to replace them.