I will try to be less facetious in the future.
Hang on! That's not a niche! Consequently, the following is a pointless discussion, as it is based on this misunderstanding of what a niche is:
You are certainly correct in that there is no environmental reason for them not to have done so, but you are forgetting the main theme of my previous post, which is that there is also no reason -- environmental or evolutionary -- for them to actually do so either.
You might simplistically see a niche as a particular set of obstacles or problems occurring naturally, which separates an organism seeking nutrients and energy from sources of one or both. These obstacles can typically be overcome in a variety of ways, and many of these ways are distinctly different.
Two already very different organisms that seek to overcome these obstacles are likely to utilize different methods to do so. This, as I outlined above, is at least partly for long-term historical reasons, that is, for reasons associated with the organisms' present morphology and biochemistry (for instance) prior to the utilization of this particular niche. You may see this as every organism having a particular set of tools -- some blunt, some already quite refined -- with which they will attempt to overcome these obstacles.
Given that the organisms initially have a very different -- perhaps non-overlapping -- set of tools, there is no general reason to assume that the solution both organisms have to overcoming the obstacles of this particular niche at a given future time will be identical, either on a macroscopic level or on a biochemical level. The typical prediction, from the viewpoint of evolutionary theory, would be that if two organisms seek to utilize the same niche, they will at a given time use a different set of tools, and the selection of tools they use will be influenced by the set of tools they had in a previous stage before this particular niche was explored by them.
In the present case, this means that a fish and a bacterium, when exploring the same niche at any given time when both lifeforms existed, the fish approached the obstacles with the tools inherent in a fish, whereas the bacterium approached it with the tools inherent in a bacterium. Therefore, unless the niche itself demands that only the tools of the fish can overcome the obstacle, there is no reason for the bacterium to adopt the tools of the fish, and become fishlike.
The reverse is not as clear-cut, of course, as there are so many cases where eukaryotes have formed a mutualistic relationship with a bacterium in order to overcome a certain obstacle. However, even in these cases, the bacterium never becomes fishlike, and the fish never becomes bacterium-like, and there is no reason, in evolutionary theory, for us to think that this will ever realistically be the case.
Meanwhile, in Davison-land, both the bacterium and the fish have the same set of tools, and thus could overcome these obstacles in the same way. Therefore, under Davison's theory, we would expect there to be cases where a bacterium had evolved into something resembling a fish, or even vice-versa.
The lack of cases where a bacterium has evolved into a fish meets the expectations of evolutionary theory -- that such cases should be exceedingly rare, if not nonexistent -- whereas it is hard evidence against the expectations of Davison's theory -- that such cases should be commonplace, or at least not rare.
I hope that putting the whole matter in layman's terms will not dissuade you from reading it. While such a strategy is prone to misunderstandings due to the impreciseness of analogies, it is sometimes necessary to fall back on them; this is certainly the case before the first cup of tea in the morning.
Like I have said over and over again, Grassé's views are irrelevant, as they are from before genetics were properly understood, and thus are based on only a (small) subset of the available data.
I am not an expert in fossil mammals, but I believe they were beach-living organisms that gradually abandoned the beach for a more marine lifestyle, for which they were eventually so well adapted that people who know little of the fossil evidence now find it hard to believe they were ever anything else.
The rest of your post is essentially a repetition of the points cited above.
Having only read a small part of the available literature, I must say that I am not aware of a test of this particular claim. However, that does not mean that we cannot perform such a test here and now.
My claim is:
"[T]he probability of two taxa to evolve into a sufficiently similar morphology decreases with the phylogenetic distance between these taxa."
A possible protocol for testing this would be to first select a number of pairs of taxa and, after having agreed on a (necessarily arbitrary) scheme of determining how well the specific pairs fit the label "sufficiently similar morphology", which I agree is vague, look for morphological and genetic data to build a matrix. We can then analyze it through normal phylogenetic methods, and thereby get an estimate of relatedness.
Given a large data set, I predict that the probability that two organisms will be "sufficiently similar" is proportional to their distance in the tree. I suggest using either only animals or only plants (or only a subtaxon within eithe), as that will give us an easy way to root the tree. I suggest further that we settle on a given gene, preferably nuclear, and discard all taxa for which we cannot find this gene. The most common gene used, the barcoding COI gene, is unfortunately mitochondrial, and will likely not give us any kind of useful resolution, so this may mean that the original data set has to be quite large.
After you have commented on this protocol, and removed as many of my biases as you can find and amended it as you see fit, we can proceed. I volunteer to do the phylogenetic analyses, and can do them according to parsimony, maximum likelihood, bayesian inference, or whatever scheme you prefer. This is likely to take some time and some effort, why I will rely on you to provide me with a list of taxon pairs for consideration. Preferably, we'd need about 2-300 of them, as many will be discarded due to lack of DNA data. The data set should ideally contain pairs which you feel are "sufficiently similar" and pairs you feel are not. I will amend this list after having received it, until we have a list of pairs we can both agree are representative. How you chose your pairs is entirely up to you; I will assume no bias from your end.
Today I will be somewhat busy at work, but feel free to present your taxon pairs whenever is suitable for you, either in this thread or through a PM.
None of these examples are niches, nor, for that matter, taxa which are very similar, which would disqualify them even if "occupying intestinal tracts" was a niche.
Yes, of course. But having whales in the oceans would severely limit the probability that a second taxon would succeed in filling what might clumsily be termed the "whale niche".
I see no reason not to accept that, as whales are undoubtedly mammals. Their closest living relative is the Hippo, and the closest living relative of these two together are the various even-toed ungulates, and then the odd-toed ungulates. All of these taxa are mammals; I accept this without qualifications.
"Aquatic life" is not a niche. Certainly the whales evolved because there was food to be had in the oceans, but if there had already been organisms occupying the "whale niche" in the oceans, whales would likely not have evolved to fill it. They didn't, for instance, while there were still plesiosaurs, pliosaurs, and other giant marine reptiles.
I certainly accept my part of the responsibility for prompting you to use "niche" imprecisely -- I go even further in this post! -- but I do not accept that I have knowingly implied that "aquatic life" is a niche which can only be occupied by one organism at a time. If I have, I apologise, and hope that you will now be able to return to using "niche" in a more proper sense.
