Hank49:
OK, the question then is, how long does it take?
<snip example>
It is nearly impossible to say how long it takes for a species to form, as it is highly variable. Some speciation events are near-instantaneous. For example, plants have a tendency to alter their chromosomal numbers (i.e. they'll suddenly start carrying an extra copy of one or more of their chromosomes). This results in a new species of plants in literally one generation, as in the kids are a different species then the parent.
Animals tend to go a lot slower, but it can even be quite fast. For example, the australian rock wallaby is in a phase of rapid speciation right now, as it's genome has been invaded by a virus which is causing whole-sale restructuring of the wallaby's DNA. In the past decade we've seen the rise of ~11 new species, as well as the extinction of some of these species. When/if this will stabilize is an open question, but when it does there is no doubt that we'll see both radical changes within the wallabeys, as well as the formation of new species.
The conventional mode of speciation, as in slow change over time leading to a new species, takes much longer then the above examples, generally speaking thousands or tens of thousands of generations. I don't know what the generation time of a parrot fish is, but multiple that by 5,000-50,000 and you'll get a rough estimate.
However, it isn't that simple. In your example evolution should occur very fast, as you have a small population (so mutations spread quickly) and strong selective pressure (i.e. 0.1% survival). This could drive the formation of a new species much more quickly (several hundred, to thousands of generations). But with ~0.1% survival, I'd expect extinction to be a more likely outcome of your experiment.
But I would end with the same statement I made last time - evolution does not have to lead to new species. There are many possible outcomes, from speciation, to extinction, to remaining the same species for a very, very long time.
Bryan