The ability to produce viable offspring is generally considered the primary differentiating characteristic between species and subspecies. That's one of a number of arguments in the debate over whether Neanderthals should be named H. neanderthalensis or H. sapiens neanderthalensis.

That criterion has had to be abandoned, replaced with "forms reproductive groups with practical boundaries".

Sometimes a river is that boundary, or a preferred prey species, a mating strategy, or odor preference.

Otherwise, we cannot distinguish bear, dog, or great-cat species.

Isn't there already a separate word—subspecies—for "isolated reproductive groups, with different phenotypes, which could still interbreed if the opportunity arose"? My understanding was that every phenotypically-distinct isolated reproductive group was considered a subspecies until its genetics diverged enough to have speciated, at which point it was now a species.

It seems to me that if e.g. American black bears and Asian black bears can interbreed, then we could call them all one species—black bears—and put all their subspecies together in into that taxonomic category. Maybe with some optional taxonomic level between "species" and "subspecies" for describing their phenotypic groupings.

But I see, looking at various sources, that those two types of bears are indeed considered separate species. Why do we do that? What's better/more useful about drawing the species boundary there?

"Species" is an organizational convenience for biologists. Nature doesn't have such a boundary. It just has varying degrees of reproductive compatibility, inclination, and opportunity.

"Subspecies" is a concession to what lumpers call splitters.

There is certainly a conservative definition for speciation, though: the point where something has zero reproductive compatibility—where there is no known example of viable offspring. At that point, inclination and opportunity cease to matter.

Why not just define “species” by that clear formal boundary, and then call everything that doesn’t manage to reach that line “subspecies”?

Because the line is very hard to discern, where it exists as a line at all, and it is nowhere sharp. Lions can be bred with tigers, in captivity. Are their offspring fertile? Well, sorta. Does it make sense to call lions and tigers subspecies? Hell, no. Say lions and tigers are one species and biologists will call you a lumper. You don't want that.

Sometimes the product of mating between species becomes, instantly, another species, if they prefer mating with one another over either progenitor. That just happened, with some birds, in the Galapagos.

Legally, there are no endangered subspecies, only endangered species. So, claiming some variety is "just a subspecies" may mean they get no legal protection against extermination. To me that's more than enough reason for a species.

I was addressing the willing rather than the able.