After digging my hands in the dirt for five years at an Allosaurus site, I've found it's pretty evident that the traditional definition of species as espoused by Mayr of discreet interbreeding populations is, well, impotent at fueling much worthwhile discussion in paleontology.
Simply put, genotypes and phenotypes are not reliable measures of each other. Hopefully most of you out there are already familiar with this fact. However, I don't think that we can just cast aside the concept of species because then we've cut off our own legs, we have to have a fundamental unit to measure and test to gain any realistic data. Biologists can use contemporary populations, we in paleontology have a sparse random sample (arguable at that).
Let me use an analagy of cats in their "kind" - haha. But, for the purposes of metaphor, imagine every Lion, Cheetah, Leapord, and Panther to be found in the fossil record to represent a time span of 2-3 million years in Africa. Now, put 150 million years worth of dirt on their graves along with some selective fossilization, with leapords fossilizing well due to their reliance on trees, often growing in riparian zones. Panthers... damn, those jungles are not good for fossilization, no excesse minerals to permineralize the bones! Cheetah's and Lions... they're grassland species, is their enough water flow to fossilize them?
Anyway, you would have some 10 - 15 "complete" skeletons (i.e. 65 - 80% complete), and an assortment of scattered bits. Now, are they one species or several? How would you tell? The gracile cheetah could be confused with a lion becuase one could argue convincingly that they represent dimorphic patterns between males and females... which of course leads to assumptions about social life which would not reflect the actual relations of either species. Ultimately, their skeletons are so similar that the concept of species is problematic at best, and at worst downright misrepresentative and would result in a horribly re-contstructed ecosystem.
So, the term "species" should not have the same meaning in different disciplines of science. Instead. let's measure exactly within the boundaries of our hypothesis. In the case of allosaurs (which has a more spotty fossilization record than the scenario above), the debate centers on if their are two species, A. fragilis and A. jimmadseni, or just one highly variable species of A. fragilis. To enter this debate, a hypothesis is formed that the convex margin of the jugal defines A. fragilis while a straightened one defines A. jimmadseni in conjuction with smaller body size... okay, I won't bore you with jargon. The end result is that the term "species" is defined in a way that functions to make the hypothesis testable.In much the same manner, the term "species" has a specific meaning in biology that also is testable in the context of genetics.
Remember, the Linnaeus classification system of Kingdom, Phyla, Class, Order, Family, Genus, and Species came about before Darwin and Wallace's theory of evolution, which in turn came out before Mendel's work on genetics was made widely available. The concept of "Species" is unfotunately the apex of this trichotomy. But there is good news, a new classification system known as the Phylocode, already taught to many undergrads in bio 101, focuses more on larger groupings of organisms. Google it if you want more information on it (hey, since when did "google" become a verb?).
Anyway, if you've read this far, the take-home message is that species is better used as a functional term in context of a specific research goal... anything out side of that is going to cause serious confusion of all "kinds".
All the best,
Theus