But as already noted, how about when new characters are needed?
Also, it really isn't about the number of species, it is about their place in the world and thematic coherence. I don't think that choosing four species effectively random and building the world around them is necessarily a super good way to produce a thematically coherent world, especially as this seems to assume that said world building probably happens very quickly. Granted, I think it could be an interesting challenge, but definitely not how I prefer to work. I intentionally left my current world very sketchy and vague so that there is room to improvise and add things. I still started making it at least year before the characters were created. I'd imagine that world that was effectively created only after character creation (so presumably a week or so before you start playing) would by necessity ne incredibly shallow.
Furthermore, why should the setting creator even limit themselves to the things WotC decides to print? Race mechanics are very light, so mechanically creating new ones is super easy (unlike with classes.) Personally I alter the species to better suit the theme of my world, and create completely new ones.
I made the eldri, a small elf-like species to fulfil the niche of elves, halflings and gnomes in my setting. It would be thematically incoherent to have normal D&D elves, halflings and gnomes alongside with them. (More pics and sketches
here.)
Personally it bugs me how many D&D species are very similar to each other, and there is a lot of thematic overlap, or if you want give them a definite niche it makes them weirdly narrow. Having so many species makes more sense if you see them as a toolkit to choose from or as an inspiration to your own creations.