For designing a world that actually makes sense there should not be more sapient species than biomes. Based on the only model we have for sapience (and the one every customer will be comparing to when judging whether or not behavior is actually intelligent) sophonts do not tolerate competition. Look at what we did to the megafauna and they weren't even competing for the same ecological niche, just dangerous. Eventually civilization reaches the point at which trade is more interesting than extermination, but by then it's too late.
Underdark cannot conventionally support sapients. Big brains are expensive and there just isn't enough energy in the system. Magic could be used as an excuse, but cthonians would be too alien to use as anything but monsters due to different senses and different means of sustenance leading to wildly different psychology. If they evolved the latest common ancestor with surface life would be extremophile bacteria. Think abominations not humanoids with different skin color.
The oceans are not a useful biome. Intelligences of crustacean or cephalopod derivation are possible. Maybe even bottom feeding vertebrates could develop arms. The whale path may also be possible, though a homonid gone the whale route would be nothing like traditional merfolk. Truly aquatic sophonts cannot operate in other biomes at all, making them useless as adventurers and difficult to use as NPCs. Seal-like sophonts could possibly be adventurers, but need to reproduce on land and therefore cannot coexist before cosmopolitanism takes hold, by which point it is too late.
If humans exist they dominate all terrestrial biomes as on Earth. It might be possible to do race variants.
If humans don't exist I'm not sure what a plains sophont would look like. If we go for six limbed vertebrates the centaur is an option. Actually, let's do that. Having six limbed vertebrates allows the traditional western dragon and several of the chimeric monsters to exist. Centaurs need more sustenance than...
Eh, I would debate the 'humans would kill everything first' bit depending on how isolated land masses are. If there wasn't a land bridge between the America and Eurasia, there is a certain possibility that humans would not have spread to any large degree in that area. Admittedly, there are plenty of isolated islands with populations in the Pacific.
Hmmmm... but humans are unusually floaty. Lets put 'can't swim' as a relatively small parameter change for our 'humans' (hey, most gorrillas can't swim, as far as I know, so it is not too much of a stretch), and then ocean travel would not be anywhere near as big an influence. I mean, would you experiment much with a technology where falling off meant almost assured death?
Also, I could imagine that another sentient species could survive in the same biome if they are simply to hard to catch reliably given primitive means (more advanced means might mean civilization could progress to the 'trade' stage before major extermination).
Crows could serve as a good model there. They are social, tool using animals that coexist within human made environments, and given how much trouble Japan has trying to exterminate them even NOW (Japan wants their nests off the power lines, and the crows have taken to building a ton of fake nest to make this near impossible, which actually exasperates the powerline problem), they could make it. Lets say a small, avian race. Basically halflings with the advantages of wings.