Cost of selection puts a limit on what natural selection can do because it tells us that selection has a cost, you cannot select Ad Infinitum. If, in a given species in a given generation, 5000 individuals can be killed by selection and still maintain the population size stable, then that is the maximum ''cost'' you can pay in that generation to filter the deleterious mutations.
Ok, so what size population might be limited to 5000 deaths a generation? What kind of organism are we talking about here?
Let's take a typical mammal, which have
very low fecundity by most standards, let's say a cat. A cat has 3-5 kittens in the average litter. An adult female produces can produce 2 (or more) litters a year. So let's conservatively suppose that a typical female that breeds at all, breeds 4 times, for a total of 16 kittens on average.
The male is likely to vary, but each kitten has one father and one mother, so we know that
if the population stays steady only two of these kittens must survive to the next generation. That means just 1 in 8 or 12.5% of kittens survive. So your 5000 that can be killed per generation while maintaining a stable population suggests a population size of just 625.
That's a very small population, using conservative numbers on the number that can be killed. Consider a frog, a turtle or a bacterium and you can easily see that the number that can be killed per generation outnumber the actual population size by
many orders of magnitude.
Edited by Mr Jack, : subtitle