Here's a question for you: See the picture on the left? That's the famous marine ground-sloth, Thalassocnus. As you can see, the genus is split into five distinct species, T. antiquus, T. natans, T. littoralis, T. carolomartini, and T. yaucensis. If they were dinosaurs, most of these species would probably be given distinct generic names. In paleomammology, it seems that authors are far more reluctant to separate animals at the genus level, even when the various species within that genus are irritatingly disparate. This may be because of another hallmark of mammalian paleontology that I find infuriating: In one of the three or four papers I have on aquatic sloths, the author(s) toy with the idea of separating some of the sloths at the generic level, but decide against it, concluding that the five creatures form a linear progression from semi-aquatic sloth to more fully-aquatic sloth. As in one population simply replaces another.
That thinking is carried over dog evolution, too. I'm reading a great book right now called Dogs: Their Fossils Relatives and Evolutionary History. My only knock against it is that the author continually posits that a known form of ancient dog (like Hesperocyon) simply "gave rise" to another dog, which in turn evolved into yet another form. In the discussion on borophagines in particular, the idea of succession is hammered into the reader's head. Paleoanthropologists used to have this idea about hominid evolution. Remember when there were only two hominid genera, Australopithecus and Homo? Now things have gotten a little better, with Paranthropus, and Ardipithecus added to the mix. Each distinct genus, however, is packed with several species. Homo consists of at least five, Australopithecus includes four or five, and even Ardipithecus, which is known from scrappy remains, consists of two species.
In mammal taxonomy, it seems like generic distinctions are tied either to chronology or some debatable "breakthrough" feature within a particular lineage. For example, very little differentiates H. habilus from the later australopithecines except for circumstantial evidence of tool usage. And even tool usage is shakey ground for a new genus name--chimps, gorillas, and orangs all use tools to some degree, and I'm sure australopithecines did, too. I read back in college about an anthropologist who wanted humans to share the same genus as chimps. We'd just be called Pan sapiens. Seriously? There are enormous morphological differences, both skeletal and otherwise, between me and Cheetah. Luckily, this thinking has not flown with the scientific community.
Things get even more bizarre when you think about modern mammals. The Indian and African elephants are generically separate (Elephas and Loxodonta, respectively). In a few million years, when future paleontologists are digging up their bones, will they be considered similar enough to warrant unification under one genus? Better yet, rather than dividing up the numerous populations of Indian and African elephants into distinct species, they are merely given subspecies variations. The savanna African elephant is L. africana africana while the forest elephant is E. africana cyclotis. Even better? One of the closest extinct relatives of the modern elephant, the mammoth, comprises of one genus and up to eleven species, based mainly on where they lived.
None of this would be tolerated in dinosaur taxonomy. Virtually every bone that comes out of the ground is given, at the very least, a new species designation. More often, you get a new genus, but this is often based on how complete the material is. While mammal taxonomy might be a case of under-splitting, dinosaur taxonomy seems over-split. Arguments about as to whether the European species of Allosaurus is a new species or an old species. If Allosaurus was a mammal, it would probably be considered a different subspecies. The mind boggles. I was blown away when Microraptor gui, Microraptor zhaoianus, and Cryptovolans pauli were theorized as being synonymous (Senter, et al. 2004), an idea that has continued to hold sway. "Dave," a juvenile sinornithosaurine dromaeosaur from China, has not yet been officially given a taxonomic name, even though it's pretty clear that the little bugger is a juvenile Sinornithosaurus. But what species is it? Liu, et al., in 2004, named a new species (S. haoiana) based on differences in the skull and pelvis. YARG!
How would mammalian taxonomy deal with a group like tyrannosaurs, where each animal seems to represent its own distinct place in the family? There isn't really a lot of "progression" with tyrannosaurs. You've got the basal forms, like Dilong and Guanlong, sure, but after that things tend to go to hell. Asian alectrosaurs are wholly different than, say, Alioramus or Appalachiosaurus. And those basal forms are, themselves, quite distinct from the albertosaurines and tyrannosaurines. And where does Eotyrannus fit in? Heck, what about Tarbosaurus, which could either be a sister taxa to Tyrannosaurus or the ultimate Asian tyrannosaurine that evolved in parallel to the North American tyrannosaurini from some Alioramus-like ancestor?
Is the mammalian fossil record that much more complete that we can point to Hesperocyon and say, "Okay, this was the direct ancestor of Osbornodon?" That just seems ridiculous! And even within Osbornodon, there are different species, including O. fricki, O. iamonensis, and O. sesnoni. And the ones in the middle are considered transitional between the earliest and the latest, as though a smooth progression from one end of Osbornodon to the other occurred! It's maddenning!
As Scott Elyard has reminded me on many occasions, fossil ancestry can never actually be known. The best we can hope for is to find a sister-group relationship between any two animals. You cannot point to Daspletosaurus and call it the ancestor of Tyrannosaurus. Instead, the two are sister groups, both born from some unknown common ancestor.
And that brings me back to Thalassocnus, the marine ground sloth. Maybe those different species occur in a fairly linear progression through time, one seemingly replacing the other, but that can never be known. Look, let's pretend that Dromaeosaurus, Velociraptor, Deinonychus, and Utahraptor are all called Velociraptor. For fun. V. albertensis, V. mongolensis, V. antirrhoppus, and V. ostrommaysi. Much as we'd like to, we can't say that one just evolved into the next, which evolved into the next, which evolved into the "pinnacle" of Velociraptor.
The best we can do is create a long tree, like this (I don't have a scanner at my side):
V. albertensis + (V. mongolensis + (V. antirrhoppus + V. ostrommaysi)
And each new step would require its own specific name. The most inclusive group might be Velociraptoria, the next step up might be Velociraptoridae, and the most inclusive group, comprising of V. antirrhoppus and V. ostrommaysi, might be the Velociraptorinae. But it's not a "line of descent," it's a branching bush with unknown common ancestors.
So that's what I don't understand about mammal taxonomy. And that's my rant for the day. :-)