The Good, the Bad and the Spinosaurus

So, after about a year I figured I’d get around to completing this. Truth was, I got distracted by other research, personal stuff, and little motivation to work on the project I was doing this is connected to. Namely, I wasn’t focusing heavily enough on the project in which I worked out how Spinosaurus aegyptiacus vertebrae might be arranged.

A recap is in order: First, in A Fistful of Spinosaurus, I argued that the dorsal vertebral spines of Spinosaurus aegyptiacus, based solely on the holotype, had an arrangement that differed strongly from other reconstructions in that it placed several centra and spines in different spots, and certainly moved the longest spine to the base of the tail. This was backed up by a reconstruction of Andrea Cau’s that I had missed (or had seen and forgotten, which is why I’ve taken a back seat in arguing it was my idea — I really don’t recall). Then, I provided a pro tem reconstruction of the skeleton based as it was on just the vertebrae and lower jaw parts of the holotype, excluding ribs and other bits that von Stromer had applied to the skeleton not mentioned in his original monograph. Scott Hartman has since produced an absolutely killer reconstruction that supercedes anything I did, so I leave that to his work.

Second, in For a Few Spinosaurus More, I attempted to argue that the holotype specimen of Spinosaurus aegyptiacus infers a different animal than the cranial material that has come from Morocco in recent years, especially that described by dal Sasso et al. in 2006. In it, they argue that their new specimen (MSNM V4047, the premaxillae, vomers and most of the maxillae and rostral nasals) represents an animal about 20% larger than the holotype.

Skull and mandibles of Spinosaurus aegyptiacus von Stromer reconstructed after dal Sasso et al. (left) and me (right). Scale bars indicate the actual sizes of the two key elements. Skull is reconstructed in generalization from von Stromer and dal Sasso et al., with a portion of the nasal crest simply merged with the snout fragment.

Their reconstruction further purports that the mandibular expansion fit the “subnarial” gap between premaxillary and maxillary tooth rows, and that their lower dentition of the dentary “rosette” expanded lateral to the upper rostrum, and even projected above the dorsal margin of the upper jaw in lateral view. I thought this unreasonable given the material, and determined instead that the jaw should not fit. I did not present the details of how this should work, but instead tried to make the case on the basis of comparisons with extant organisms in which rostral dentition is expanded. I never completed my argument, so I’ve gone back and started work on this again, as intended, after a bloody whole year. Yay. Or “booo” — because this took waaaaaay too long.

(Those last few sentences were written in August 2013. It is now 2014. Moreover, when I started this post it was about June 2012. This post had been delayed most because I was working on some concepting work and, well … sidetracks. I had a pair of intensely more important projects that took ALL of my time away, and while this post sat at the back of my mind I couldn’t get back to it.)

Before I commence with my argument, let me clarify: I frame this argument under the principle that I am not correct; it is not my assumption that I am right, nor that dal Sasso et al. or other reconstructors are wrong. Rather, it is my position that I cannot reconcile the reconstruction with morphology, and found that some things were not necessarily accurate. I do find that dal Sasso et al. used sound principles in reconstructing their skull, and that the design of the skull and reconstructions following this are sound. I disagree on some of their principles, but wish rather to test them. My preliminary reconstruction, presented in the last two posts, sought to do this, but were incomplete.

Principally, I found that the assumption that MSNM V4047 and BSP 1912 VIII 19 represented very similar taxa, that they were merely size adjusted, and the lack of differentiating characters used in their comparison problematic. Specifically, the issue of correspondence of different bones to similar-seeming taxa is frought with potential error, and dal Sasso et al. acknowledge this, where two elements with no overlapping parts are assumed to be the same taxon by various criteria; that whatever morphological differences between the two could be accounted for by size, such that the holotype mandible was merely of a much smaller animal; and that there were notable distinctions between the two which raised the potential specter of taxonomic variation (even if it would be dismissible as individual variation, for which no evidence ca be forthcoming given the material is of two different sets of bones).

This current post attempts to present the final part of my argument for why MSNM V4047 probably does not apply to the same species as BSP 1912 VIII 19, holotype of Spinosaurus aegyptiacus von Stromer (1915). This argument seeks to find a disjunct in how the mandible aligns to the newly described jaw, and thus how one might further reconstruct the upper jaw for Spinosaurus aegyptiacus, and possibly the lower jaw of the Moroccan form.

I use two premises:

1. The BSP mandible represents the natural morphology of the adult or skeletally mature animal, and that it is an accurate representation of the jaw of the animal even were it scaled 20%. Near mature or nearly-top-size giant theropods can vary in size by this amount, as in Tyrannosaurus rex, but also conversely can maintain similar cranial dimensions as opposed to dissimilar postcranial proportions, as this comparison of Sue (FMNH PR201) and Stan (BHIGR 3033) by Scott Hartman can attest. Against this premise, I have measured against a natural comparison of the specimens as they are now, and with the BSP mandible scaled up 20%.

2. The actual morphology of the jaw fragments considered here are preserved as they would be in life. If I were doing this full-tilt, I would be trying to figure out what the natural morphology of the elements would be without taphonomic distortion involved, and thus the actual attitude and orientation of the various bits. We are using the illustrations of dal Sasso et al. and von Stromer here, and nothing more.

I further used the following conditions to limit how far my reconstruction must go:

1. The lower teeth are lingual to the upper teeth when the jaw is closed, and the mesialmost dentary teeth are distal to the upper first premaxillary teeth. To affirm this, I attempted to determine, as in alligatorid crocodilians, whether there are tooth pits on the upper jaw into which the lower teeth might fit if they were interspersed between the upper teeth, or whether the upper and lower teeth were evenly spaced to allow them to “interlock” (as in, say, Dakosaurus maximus), or further crossed past one another, as in crocodylid crocodylians, or with pits on the outside of the jaws that allow the teeth to “slot” into the bone (as in Goronyosaurus nigeriensis Lingham-Soliar, 1999).

2. A: The rostral end of the mandible can only come as close to the upper jaw as a reconstruction of the first and/or second dentary crown can fit, when that tooth “occludes” the palate of the upper jaw, and B: the caudal end of the mandible can only come as close to the upper jaw as a reconstruction of the last dentary crown can fit, while C: these dimensions being determined as a function that the alveolar diameter preserved a maximal tooth size with a crown height (CH) as long <2.5x the diameter of the alveolus, so as to create an objective point from which to measure missing crowns.

3. Crown orientation in distal view (relative to crown base: maxilla ventrally, dentary dorsally) is oriented perpendicular to the long axis of the skull, where the mesial and distal carinae are parallel to the long axis of the skull. This pattern conforms to that seen in other large-bodied theropod dinosaurs in lateral crowns (non-mesial maxillary and dentary crowns, and certainly excluding premaxillary) where the carinae are arranged parallel to the jaw’s long axis. It is also confirmed for MSNM V4047. This premise is extended to BSP 1912 VIII 19. The principle governing this premise is important, as I will detail further down.

With this in mind, let us consider the material.

The Good – We Have Nearly the Entire Skull of a “Generalized” Spinosaurine Spinosaurid
(Unfortunately, it’s not all one specimen.)

The skull of our exemplars is build, as in the last post, from a combination of Irritator challengeri, Angaturama limai (as a probably synonym), Spinosaurus aegyptiacus (which preserves that part of the jaw that Irritator challengeri lacks), and MSNM V4047 (which bridges the upper skull of Irritator challengeri and Angaturama limai). Additional material includes fragments that are not necessarily relevant. I want to know the probably shape, proportions and length of the mandible, and its best fit to the mandible. This is to produce a cranial reconstruction in dorsal, lateral and ventral views.

Skull of SMNS 58002, holotype of Irritator challengeri Martill et al. (2001) in dorsal, left lateral, occipital and right lateral views.

The Bad – What Do You Mean, “It’s Not All One Specimen?”

See, I want to have a singular species with which to compare everything, and wish I had a singular specimen. I don’t. This makes my intended goal (comparing the reconstruction to something real) difficult. I have to make a lot of assumptions. Instead of creating a skull, modeled however prettily as I did last time, I must use a variety of associated analyses and try to composite them. Then, and only then, can I attempt to answer the question: “Does MSNM V4047 belong to Spinosaurus aegyptiacus?”

It’s not a total loss, though. I can use a variety of little techniques to help me align everything, and make testable hypotheses, then compare these to further data. The beauty of this is if we find more material, better specimens, we can then use this information to test against whatever the heck I’m doing. I get to find out not only that I did it right (or wrong) but that my conclusions were correct (or not) and what I could have done better. The process here is to make this analysis as self-correcting as possible (something I was not very successful with last time).

This analysis requires me to do several things. First, I need to align the mandible. As I mentioned here, alignment requires me to get the jaw in the right orientations in all axes. But since I am using an illustration done in only two perspectives (presumably directly lateral and directly dorsal) and I cannot correct these (the material is not available, being dust) there is an intrinsic flaw. Nonetheless, I try to do my best. One of the things I need to do is to figure out what the intermandibular angle is between the dentaries. This will tell me how wide the jaw is at each alveolus, and I can compare this directly to MSNM V4047 at each point along the jaw. To do this, I need to have a general idea of the shape of the mandible as if complete, and how wide apart the mandibles were at the quadrate/articular jaw articulations; for both of these, I will lean heavily upon Irritator challengeri, but for fairness I will also use the “baryonychines” Baryonyx walkeri and Suchomimus tenerensis to make sure I cover all of my bases.

Second, once I have the mandible length and then posterior width set, and thus the angle between the two halves and thus their width along the length, I can now work on the length and width of the upper jaw … which is a lot easier. While it still requires some fudging due to uncertainty in assuming that all of our “spinosaurines” are alike, given our “baryonychines” it is probably a safe bet that similar proportions prevail. This then sets the jaw model, and applying a further assumption (that all teeth are vertically oriented and not laterally splayed, which corrects some distortion of the jaws) helps enforce Premise #3 (above).

Third, I performed two separate fit analyses, in which the position of the mandible is checked against the upper jaw when the natural size is considered, and when I increased the total dimensions of the mandible by 20%. This includes width, which doesn’t increase the overall linguolabial dimension by any extreme amount, and in fact it is fairly negligible when relative position of the jaws is considered. Considerations on these analyses were: Where the position of the “terminal rosette” was in relation to the upper jaw, and what was the mandibular angle when enforcing Premise #1? That is, the posterior mandibular alveoli had to be lingual to the alveoli of the upper jaw, and in fact I had to enforce this along the entire length of the jaw. Fitness is considered on two levels: how well the mandible stays within the alveolar margin of the upper jaw in alveolar view, and how closely the curvature of the margins of the jaw fit one another. This last fit measure is meant to determine how much soft-tissue, if any, has to extend around the jaws to “close” the gaps in the jaw that must result due to occlusion of the mandibular teeth to the palate. I tested the fitness by also retracting the jaw to a point that conforms to that of dal Sasso et al., where the “terminal rosette” underlies the “subnarial gap” of the rostrum.

Fitting disparate specimens (which I had argued were disparate species) to one another is a tricky task. I consider the effect of this when dealing with other spinosaur taxa. Let’s consider this image, from Wikimedia Commons:

Cast of skull of Suchomimus tenerensis Sereno et al., 2001. Someone has overlain the photo with a layer highlighting the premaxillae-maxillae in red (the holotype).

Note the extreme slenderness of the mandible. In consideration of the posterior width of the skull, this may seem very, very narrow, but in comparison to Spinosaurus aegyptiacus, it isn’t. For one, this skull narrows over a longer length of skull than does our subject, while at the same time the dentaries are effectively parallel for most of their length. This doesn’t seem to be the case for Spinosaurus aegyptiacus, in which the mandible widens much further rostrally from the narrow mandibular symphysis, and it does so for a particularly interesting reason: Spinosaurus aegyptiacus likely had extremely large mandibular adductor musculature, and further anterior insertion on the jaw, than in “baryonychines.” This is true of Irritator challengeri as well, to be honest; there is nothing really to this aside from the fact that the posterior position of the mandibular widening results in a constraint on where the pterygoid processes of the palate can depress: the further to the rear the mandible widens, the further to the rear or the less space the pterygoid processes have to fit between the jaws. And fit they must. But I get ahead of myself.

The Ugl–I mean, the Spinosaurus

I used illustrations of the mandible from von Stromer (1915) and of the upper jaw from dal Sasso et al. (2006). Two sets of mandibular illustration were produced, after the elements were naturally scaled (using von Stromer’s provided maximum length measurement for BSP 1912 VIII 19). I mirrored the mandible in dorsal view for both measurements. A similar scaling for the jaws was produced for lateral views of the same. The alveolar view set were coordinated with respect to my premises, and the lateral view set were positioned with respect to this; conversely, the alveolar set were positioned with further respect to the lateral view, so that their positions were coordinated.

I used two different positions for the jaw, to consider how “fit” the jaw is in alveolar view: where the first alveolus of the mandible abuts the second alveolus of the premaxilla (it appears impossible to bring the mandible further rostral while conforming to my premises); and where the teeth of the “terminal rosette” are positioned roughly exclusively in the “subnarial gap” region. It is the latter that is illustrated by dal Sasso et al., and is being tested here.

There is greater correspondence in the shape of the margins of the jaw in lateral view when the mandible is “natural”; this is not the case when the jaw is 20% larger. Part of the effect this has is that the mandible “fits” dal Sasso et al.’s reconstruction, where the mesial dentition lies lateral to the “subnarial gap” when the calculated 20% increase is not taken into effect.

Illustration of the alveolar shapes and relative positions of the spinosaur rostrum fragments, based on BSP 1912 VIII 19 (after von Stromer, 1915) — in blue — and MSNM V4047 (after dal Sasso et al., 2006) — in red. The mandibles are considered “natural” for this figure. Top: Shapes and arrangement of the alveoli, with the mandible “fit” into the upper jaws as per Premise #3. Bottom: Relative rostral positions of the mandible in “narrow” gauge in rostralmost position, with the first dentary tooth posterior to the third premaxillary (largest) tooth, and in the position argued by dal Sasso et al. (2006).

I just did a post on overbites. In fact, it was the absence of a discussion on overbites that held this whole post back, but now that I have that, I can continue. I noted that overbites in those animals in which such a feature is unambiguously present seems to suggest a few factors. In eurhinosaur ichthyosaurs, the rostrum is much elongated and probably serves a distinct, billfish-like function or, as in pristid rays (sawfish) possible a sensory and defensive function, but the dentition in these overbitten portions is either absent or uniform. There are are some crocodilians (such as the basal neosuchian Sarcosuchus imperator and similar taxa, like the dyrosaurids) which have a region of the upper jaw that overbites the lower but not by much. Proterosuchids (which may not be a natural group) are basal archosauriforms and include Archosaurus rossicus and Proterosuchus fergusi, in which the premaxilla drastically overshoots the lower jaw and not just just have strongly recurved teeth that point backwards into the mouth, but preserve a distinct diastema between the maxillary and premaxillary dentition.

But these morphologies don’t seem to be the case in spinosaurines: The premaxilla has drastically differently sized teeth, and there is a distinct diastema, but the jaw maintains a linear relationship to the maxillary teeth, and the curvature of the jaw isn’t as severe as on the lower jaw’s alveolar margin. But there’s more. Theropod dinosaurs have jaws that fit INTO the maxillary arcade, and almost certainly entirely along its length. There is often quite a bit of room on the upper jaw for this, but in the case that some teeth enmesh between others on the opposite jaw, there tends to be wear facets. This is true for whales and marine reptiles (Thewissen et al., 2011; Young et al., 2012), there is little enough reason to assume it’s not true for terrestrial theropod dinosaurs. But I don’t think the dentition is necessarily so enmeshed here, for precisely that reason, yet if placed into the position of dal Sasso et al., the mandibular teeth do enmesh partially. Absence of dental pits in the maxilla also imply lack of a close fit of the jaws, almost certainly necessary for the mandible to reach the position dal Sasso et al. give it. The “terminal rosette” of the mandible, at the “corrected” 20% larger size results in there always being a point where the jaws must interlock. Dal Sasso et al. took this even further by suggesting the dentition of the lower jaw passed the ventral margins of the upper at the diastema, which is admittedly quite narrow, but what this ends up doing is pushing the teeth of the lower jaw into the maxilla, though they are fairly small. Unlike in Baryonyx walkeri (or Suchomimus tenerensis), in which the post-rosette dentary teeth are less than 40% the height of the rosette teeth, Spinosaurus aegyptiacus has post-rosette teeth only about 50% their height. So while the rostral teeth are very large, the posterior teeth are not tiny; the size difference is not as extreme except immediately behind the rosette, in which the alveolar diameter is less than 25% the rosette dentition.

Spinosaurine cranial fragments in lateral view, based on BSP 1912 VIII 19 (after von Stromer, 1915) and MSNM V4047 (after dal Sasso et al., 2006). Top two: “natural” size; bottom two: mandible increased in dimensions by 20%. The first and third images position the mandible rostrally so that the first dentary tooth is immediately caudal to the first premaxillary tooth (absolute maximal length). The second and fourth images position the mandible caudally so that the “terminal rosette” coincides with the “subnarial gap.”

Typically, the upper dentition merely passes the alveolar margin of the lower jaw, and terminate about at the position of the row of foramina that extend along the alveolar margin, which tracks the inferior alveolar nerve (maxillary branch of the of the trigeminal nerve which passes into the mandible) (Soares, 2002; George & Holliday, 2013). We probably cannot rely on this data point based on the problems that arise from taphonomic distortion in the fossil record: the mandible is often more closely appressed (or looser) than the “natural” close position might attest — but also, as seen in monitor lizards when the jaw is “shut” in fleshed head, the teeth of the lower jaw rarely abut the upper. It is merely that when the jaw is shut, it should generally be able to shut to a certain point and no further. The reconstruction of dal Sasso et al. has the largest maxillary teeth pass beyond the edge of the mandible, the mandibular “rosette” pass beyond the edge of the premaxilla. These factors are so unusual they invite criticism on their argument alone.

When the two specimens are associated at original size, the mandible of BSP 1912 VIII 19 is just narrow enough to fit between the dental arcade of the upper jaw of MSNM V4047 (below). The alveoli of the former are visible in dorsal view, but just barely, as long as the posterior dentary array are precisely medial to the maxillary array (as per above). Yet when the mandible is enlarged to 20%, to “fit” MSNM V4047, maintaining the position of the dentary teeth medial to the maxillary results in no real difference: the rosette remains unable to pass a tooth clearly vertically. Of course, this doesn’t dismiss the possibility of the teeth enmeshing with the mandibular ramus tilted along its long axis. Instead, the exteme difference in size of the upper and lower teeth prevent the lower teeth from passing easily any portion of the maxillary teeth, meaning the upper teeth firmly prevent and outward tilt of the upper teeth. The lack of an equitable orientation becomes more problematic when one considers the angle of the mandibular symphysis.

The mandibular symphyseal surface is unavailable for examination. What we have left is just the dorsal view of the element and von Stromer’s description, which is unrevealing. Using merely the dorsal view, which is taken with the alveoli in the transverse plane and probably perpendicular to the sagittal, thus parallel to the viewer, there is a broad contact along the dentary that extends the length of the rosette, and thus about 6 tooth positions. This region is broader at the fifth alveolus than anywhere else along its length. This constrasts with Baryonyx walkeri in which the symphyseal surface is short, high, and restricted to about 1 tooth position in length (Charig & Milner, 1997), although the bone in this region is partially eroded. Unlike BSP 1912 VIII 19, medial surface of the dentary is inflected laterally in this region, indicating the symphyseal angle is narrow, but separates the jaws posterior to the 2nd alveolus. BSP 1912 VIII 19, therefore, suggests a direct measurment of the angle of the mandible. When confined to the maxillary arcade, however, it is necessary to “clip” this surface, and so the illustration below provides two angles for the paired mandibles (again, the caveat being that it is being produced from von Stromer’s (1915) inkwash illustrations, however accurate they might be, and untestable now). The first is the minimum in which all dentary teeth are within the maxillary arcade (as above) — termed here the “narrow” gauge — and the second is the “aligned” gauge where the apparent symphyseal angle is preserved. The mandible suggests a much broader angle, a more triangular aspect, in dorsal view, with a broad gular region between mandibular rami suggesting a wider jaw. When scaling the mandible to 20% larger, the teeth of the mandible can pass lateral to the diastema, but only some of them, and when positioned caudally as in dal Sasso et al., even the first mandibular tooth may be able to pass the diastema.

Spinosaurine cranial fragments in lateral view, based on BSP 1912 VIII 19 (after von Stromer, 1915) and MSNM V4047 (after dal Sasso et al., 2006). C represents a conjoined alveolar view of both specimens. A, “natural” sizes of the fragments in rostral (left) and caudal (right) positions of the mandible, as determined by the the text, and “narrow” and “aligned” gagues, as determined by direction of dental carinae orientation. B, same, but with the mandible increased by 20% in all dimensions. C, “aligned” and “narrow” gauges of the mandible.

This results in improper assumptions that the two specimens might correspond to an effectively identical animal but for size.

Dental morphology between the two specimens was also compared. Unlike animals in which a distinct if even slight overbite might be present, the teeth of BSP 1912 VIII 19 are not curved, but are merely unserrated, straight, and keeled; those in MSNM V4047 are not well preserved, and only one crown is well preserved and close to what is assumed its full size: The left 2nd maxillary crown. This crown is curved along its mesial margin, whilst the distal margin is only slightly curved, but this curvature is distal from the base and caudal, meaning it is recurved, whereas in BSP 1912 VIII 19 the crown’s curvature are always toward the apex on both mesial and distal edges.

Dental orientation, unlike morphology, can be trickier to pin down. But when it comes to typical theropods — which as I understand spinosaurids aren’t easy to compare to — dental orientation tends to be the same between upper and lower jaws. And when different, there’s usually a good reason (such as a short radial array at the rostral end of the maxilla, as in Coelophysis bauri). Most importantly, the tips of the teeth point in the same way. This is due to the need to have the teeth approach parallel trajectories as the jaws close for the most efficient bite, most importantly with an unserrated, piercing morphology the teeth of both jaws shouldn’t diverge in angle when engaging prey. The maxillary teeth in MSNM V4047 are angled rostrally when sufficiently preserved, whilst the premaxillary teeth appear vertical as long as the upper jaw is held in a longitudinal orientation. This constrasts with BSP 1912 VIII 19, in which all but the very first dentary tooth appears to be vertical. If we assume the two specimens are the same species and using the jaw length of dal Sasso et al. (2006) as a guide, then the teeth of the upper and lower jaws wouldn’t orient on the same angle during closure of the jaw, preventing the teeth from fully functioning as a piscivorous, piercing apparatus. I’ll get into this particular “apparatus” at a later date, but it should be noted that for the most part, especially with animals with a radial-ish array of the lower jaw, the teeth usually oppose backwards curving teeth on the upper, which always makes the tips of the teeth angled towards the bite action. When there is a distinct difference, different functions for the teeth are employed. Teeth that splay outward, even in nominal “fish-eaters” (a good example is anglerfish), are not necessarily used for prey acquisition (are not piercers) but as a “trap array.” It was this reason I spent some time reconstructing Masiakasaurus knopfleri and Daemonosaurus chauliodus without the progrnathic dentition, as it made little enough sense at the time.

The likelihood then, as I see it, is that the two jaws are neither perfectly congruent (lacking application of the morphological overlap test) nor do they suggest a similar morphology to one another. The relative angle of the dentition, mandibular symphysis, arrangement of the teeth and their size, their curvature (or lack thereof) each suggest that we are dealing with two, distinct morphologies. With slightly curved, procumbent dentition and a narrow jaw in MSNM V4047, but straight and vertical dentition with a wider jaw in BSP 1912 VIII 19 the two specimens imply, at least, two species with dietary acquisition strategies: An eastern, straight-toothed macrophage, suited for eating relatively large estuarine/paludal fish, perhaps with a widening gape as suggested by helical quadrate/articular joints (especially as noted by Andrea Cau in this post on his blog [in Italian, but here’s an English translation via the Google]) to swallow, rather than process, very large prey; and a western, curve-toothed mesophage, suited to moderate large estuarine or near-shore fish. Thus, I disagree with dal Sasso et al. (2006) and the further use of MSNM V4047 as a proxy upper jaw for Spinosaurus aegyptiacus. More likely, the skull of this species would be both shorter and teeth than has been reconstructed

Charig, A. J. & Milner, A. C. 1997. Baryonyx walkeri, a fish-eating dinosaur from the Wealden of Surrey. Bulletin of the Natural History Museum of London 53: 11-70.
Dal Sasso, C., Maganuco, S., Buffetaut, E. & Mendez, M. A. 2005. New information on the skull of the enigmatic theropod Spinosaurus, with remarks on its sizes and affinities. Journal of Vertebrate Paleontology 25(4):888–896.
Fowler, D. W., Freedman, E. A., Scannella, J. B. & Kambic, R. E. 2011. The predatory ecology of Deinonychus and the origin of flapping in birds. PLoS ONE 6 (12): e28964.
George, I. D. & Holliday, C. M. 2013. Trigeminal nerve morphology in Alligator mississippiensis and its significance for crocodyliform facial sensation and evolution. The Anatomical Record 296: 670-680.
Soares, D. 2002. An ancient sensory organ in crocodilians. Nature 417: 241-242.
Schubert, B. W. & Ungar, P. S. 2005. Wear facets and enamel spalling in tyrannosaurid dinosaurs. Acta Palaeontologica Polonica 50 (1): 93-99. [PDF]
Thewissen, J. G. M., Sensor, J. D., Clementz, M. T. & Bajpai, S. 2011. Evolution of dental wear and diet during the origin of whales. Paleobiology 37 (4): 655-669.
von Stromer, E. 1915. Ergebnisse der Forschungreisen Prof. E. Stromers in den Wüsten-Ägyptens. II. Wirbeltier-Reste der Baharîje-Stufe (unterstes Cenoman). 3. Das Original des Theropoden Spinosaurus aegyptiacus nov. gen. et nov. spec. [Results of the explorations of Prof. E. Stromer in the Egyptian desert. II. Vertebrate remains of the Bahariya Formation (lowest Cenomanian). 3. The original of the theropod Spinosaurus aegyptiacus nov. gen. et nov. spec.] Abhandlungen der Königlichen Bayerischen Akademie der Wissenschaften Mathematisch-physikalische Klasse Abhandlung 28(4):1–32.
Young, M. T., Brusatte, S. L., Beatty, B. L., de Andrade M. B. & Desojo, J. B. 2012. Tooth-on-tooth interlocking occlusion suggests macrophagy in the Mesozoic crocodylomorph Dakosaurus. The Anatomical Record 265 (7): 1147-1158.