That's the thing. Most of the evolutionary theories about how something happened 2 or 200 million years ago are just on the level of speculation. That's one of my gripes about evolution ... the field attracts a lot of pseudoscience (such as evolutionary psychology) in the form of "explanations" for various phenomena we observe today, or how a transition happened.
The truth is no one actually knows, or can test most of the theories that seem like they MIGHT be true but could also be fanciful speculations by researchers. That something was found is a fact. How it got there is largely speculation, but I'm surprised how often these theories are taken seriously, not just by the public but by other scientists.
Real tests include statistical significance, like exploring the fossil record across continents, simultaneously corroborating the theories of continental drift and common descent. Allopatric speciation can be discovered this way. You need a bunch of data points or functions of them, that should be independent but are highly correlated. That is what supports a theory, in the Fischer sense.
Here is an example: horse fossils are found physically above t-rex fossils 99.9% of the time, and in the remaining cases, the rocks show evidence of upheaval in that area. Or, radiometric dating vs the superposition principle. If 99% of the data samples agree, that is extremely significant. That is how you prove theories.