This post will discuss soemthing that a paper has already done in full — however, some conclusions about this paper’s result means that I will re-cover what it did, and make a different argument. That paper is Krause et al. 
This is a tale of two taxa, Majungasaurus crenatissimus  and Majungatholus atopus . The first was discovered at the Maeverano region on western Madagascar in the 1890′s, while the latter was discovered a bit afterward. Both are likely to be the same animal, as covered in , and this argument is not to rehash that or dispute it.
The problem here stems from the actions taken by Depéret , where upon naming the theropod dinosaur Megalosaurus crenatissimus, chose to name an animal represented by a series of fossils (FSL specimens 92.289-290, 306, and 343), but never designated any of these fossils as the holotype. One should note that this practice is hardly unique: neither Cope nor Marsh during the infamous “Bone Wars” were fully willing to designate the holotypes of specimens by original designation, and the hard-and-fast rules for type-setting were not established until the early decades of the 1900′s.
Lavocat , upon pointing this out and having recovered his own Maeverano specimen (a partial dentary with an unerupted crown (MNHN.MAJ 1), designated this specimen to be the neotype of crenatissimus, removing crenatissimus from Megalosaurus by forming the new genus Majungasaurus for it. This is our second issue: Lavocat had to have determined that the original material irregardless of their typification, were part of a designated series (Depéret had to have them in mind when he form the name), which tells us that when Lavocat attempted to re-establish the type specimen, he did so by setting aside a holotype and designated a neotype. As  states (pg. 6):
Although not explicitly stated, it appears that Lavocat (1955[...]) did not consider any of the specimens in Depéret’s type series to be worthy of designation as the lectotype.
Intriguingly, Sampson et al.  argued that neither the holotype (the FSL series) nor the neotype were adequate to serve as a type specimen, and used this argument to refer the newly recovered specimens, including complete skeletons, to Majungatholus atopus . Thus they presumed that Lavocat’s material was not comparable as a basis for taxonomy, as well as Depéret’s. The cranial material previously recovered, substantive seeming as a type for a pachycephalosaur , makes for a peculiar animal. It has, for example, allowed others  to presume other theropod dinosaurs from pachycephalosaur remains, although of course troodontids were also considered former pachycephalosaurs (I sense a trend…).
The problem, of course, is that one of these specimens must be a lectotype. The holotype will be that specimen, and the mandible will be the neotype. At this point, defining characteristics follow from the neotype, so it wouldn’t matter which of these specimens was the lectotype, especially if it weren’t one of the two, partial teeth. Sekeletal material is now known for phalanges, vertebrae, and teeth that can be compared to Depéret’s original series. Because a lectotype had never been established, it is even possible for the authors to find one of Depéret’s specimens that is the closest to being the most diagnostic (I nominate the caudal vertebra, FSL 92.289, on basis of being the most complete element preserved and not being a tooth). This would prevent, in the least, for Majungasaurus crenatissimus from being a tooth-based taxon.
But this continues the discussion, regardless. How diagnostic is the caudal vertebra, FSL 92.289? A caudal vertebra may be complete or diagnostic enough to form the basis of taxonomy, although some may differ on how much material is required to be diagnostic, or requiring an autapomorphy (or more), or having an autapomorphic suite (a series of otherwise non-autapomorphic features, i.e. apomorphies). As the caudal sequence in FMNH PR2100  is known up to a point that is congruent with the morphology of a part of Depéret’s type series, one can investigate this. Little is said in  about the diagnostic aspects of Majungasaurus crenatissimus vertebrae, although  cite (pg. 2):
(12) dorsal and caudal vertebral neural spines dorsally expanded (transversely and craniocaudally)
FSL 92.289  lacks evidence of the neural spine as shown, and this point in FMNH PR2100 the neural spines are thin and unexpanded. The vertebra is otherwise identical to that in FMNH PR2100 and probably came from the same kind of animal. Affirming it as Majungasaurus crenatissimus is not an issue; it is part of the type series, thus it MUST be Majungasaurus crenatissimus, barring evidence that the different elements come from different animals. Assuming the nature of a single abelisaurid taxon from Maeverano … a little tricker, but so far, based soley on sampling, the taxonomy of abelisaurids from the Malagasy region is secure: there is one taxon we know of, and its name is Majungasaurus crenatissimus.
But it could go back the other way: We can say that dental-based variation is tricky, and that even closely related, similar, but disparate large-bodied theropods exist in the same formation, at the same time. This occurs in the Dinosaur Park Formation in Alberta, Canada, where Albertosaurus libratus and Daspletosaurus torosus are both known (they are similar in size, although they differ, and primarily in the crania). The type series can also belong to a disparate group of animals, one feeding upon the other, although there is so far no evidence this has occured . Most large theropods are opportunistic carnivores, and they are thus compared to large predatory mammals who will resort to cannibalism in times of stress or for territorial reasons. Disparity among the material but confined to only two large abelisaurid taxa seems unlikely.
But it could have gone that way.
The moral here is not that I disagreed with anything these authors have said, but that their arguments shine a light on the issue of loose typification how taxonomy can be tricky. This is not only dependant on how one selects a type, but the quality and quantity of the material used to do so: What extent, what degree of preservation, what number of features, what quality of features … qualifies a holotype, a taxon, a referal?
 Krause, D. W., Sampson, S. D., Carrano, M. T. & O’Connor, P. M. 2007. Overview of the history of discovery, taxonomy, phylogeny, and biogeography of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar. Journal of Vertebrate Paleontology 27 (supplement to 2, SVP Memoir 8):1-20.
 Depéret, C. 1896. Note sur les Dinosauriens Sauropodes et Théropodes du Crétacé supérieur de Madagascar [Note on sauropod and theropod dinosaurs from the Upper Cretaceous of Madagascar]. Bulletin de la Société Géologique de France 21:176–194.
 Sues, H.-D. & Taquet, P. 1979. A pachycephalosaurid dinosaur from Madagascar and a Laurasia-Gondwanaland connection in the Cretaceous. Nature 279:633–635.
 Lavocat, R. 1955. Sur une portion de mandibule de Théropode provenant du Crétacé supérieur de Madagascar [On a partial theropod mandible coming from the Upper Cretaceous of Madagascar]. Bulletin du Muséum National d’Histoire Naturelle 27:256–259.
 Sampson, S. D., Witmer, L. M., Forster, C. A., Krause, D. W., O’Connor, P. M., Dodson, P. & Ravoavy, F. 1998. Predatory dinosaur remains from Madagascar: Implications for the Cretaceous biogeography of Gondwana. Science 280(5366):1048-1051.
 Naish, D. & Martill, D. M. 2008. Dinosaurs of Great Britain and the role of the Geological Society of London in their discovery: Ornithischia. Journal of the Geological Society, London 165:613–623.
 O’Connor, P. M. 2007. The postcranial axial skeleton of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar. Journal of Vertebrate Paleontology 27 (supplement to 2, SVP Memoir 8):127-162.
 Rogers, R. R.; Krause, D. W. & Curry-Rogers, K. 2007. Cannibalism in the Madagascan dinosaur Majungatholus atopus. Nature 422(6931):515–518.