Comments on Therrien and Henderson's new paper

[Michael Mortimer <mickey_mortimer111@msn.com>]

While reading through the new paper in JVP on size estimates in theropods, I 
came across several points which I think deserve comment-

The authors find a relationship between skull and body length in theropods 
(excluding therizinosaurs, oviraptorosaurs and ornithomimosaurs), then test 
it using Sinraptor dongi, Velociraptor, Carnotaurus, NGMC 91 and Mei (none 
of which were used to model the relationship originally).  They find-

"The body length estimates for Sinraptor and Velociraptor are extremely 
close to the published length for the specimens considered (<3.3%), which 
reinforce the validity of SK-BL."

And yet the published length of Sinraptor dongi is just another estimate, 
since only a few proximal caudals were preserved.  Same for Velociraptor, as 
the specimen used (AMNH 6515) is just a skull and phalanges.

"Interestingly, the body length of Mei long and of the small feathered NGMC 
91 is overestimated (36% and 85%, respectively) by the SK-BL regression 
whereas that of Carnotaurus is underestimated (28%). These latter results 
reflect the facts that juvenile theropods have larger heads relative to 
their body size than adults (Ji et al., 2001; Xu and Norell, 2004) and that 
abelisaurids have shorter heads relative to their body size than other 
theropods (Bonaparte et al. 1990). Consequently, the SK-BL regression should 
not be used for abelisaurids or juvenile individuals."

Er, maybe it's because Mei and NGMC 91 both have much shorter tails (3.17 
and 3.6 times femoral length respectively) than any other theropods used, 
especially small ones.

Indeed, their figure 2 shows sillouettes of Compsognathus, Coelophysis, 
Allosaurus and Tyrannosaurus scaled to a standard total length to illustrate 
how relative skull length supposedly increases with increased total length.  
What should be obvious from the picture is that this is due to the 
increasingly shorter tails of the taxa!  Compsognathus' tail is >6.4 times 
its femoral length, Allosaurus' is ~5.1, and Tyrannosaurus' is ~4.6.

The dataset shows a bias for short-tailed larger theropods (7 tyrannosaurids 
vs. 3 carnosaurs, Dilophosaurus and Ceratosaurus) and long-tailed smaller 
theropods (2 coelophysids, 3 compsognathids and Deinonychus vs. 1 
troodontid).  Note also that tyrannosaurs (especially the larger 
tyrannosaurines, especially the largest Tyrannosaurus) are known to be 
comparatively more heavily built than other theropods.  Examining the 
dataset, their Allosaurus fragilis specimen is not YPM 1930 (the fragmentary 
holotype), but is USNM 4734 (the topotype).  I really don't see why 
Carnotaurus, Huaxiagnathus, Juravenator, Mei, NGMC 91 and Sinraptor 
hepingensis are included in the "incomplete" dataset.  They're more complete 
than Acrocanthosaurus, Deinonychus, and the Compsognathus holotype, which 
are all in the "complete" dataset.  Indeed, the Deinonychus specimen is just 
a partial skull.

The authors go on to estimate the length and mass of Spinosaurus, based on 
the new large snout (MSNM V4047) described by Del Sasso et al. (2005).  They 
say-

"A note of caution concerns the dimension of the reconstructed Spinosaurus 
skull. First, the proposed skull restoration (Dal Sasso et al. 2005:fig. 5B) 
is a composite reconstruction with the front and back halves being from 
different genera. Because skull shape varies among spinosaurids (i.e., the 
shape of the rostrum, the relative dimensions of the maxillae and 
premaxillae, and the shape of posterior region of the skull differ among 
spinosaurids; Fig. 3), there is potential for overestimating the length of a 
composite spinosaur skull."

Figure 3 shows the skulls of Baryonyx, Suchomimus and Irritator, with known 
parts outlined darkly.  Yet not only is the old reconstruction of Baryonyx 
used (from Charig and Milner, 1997), but the authors failed to realize the 
reconstruction of Suchomimus (from Sereno et al., 1998) is mostly based on 
Baryonyx (except the premaxilla and maxilla)!  Furthermore, reconstructing 
Spinosaurus based on the elongate baryonychine morphology instead of 
Irritator would result in a longer skull length if anything.

First, the authors test their skull vs. length and skull vs. mass 
relationships with more complete spinosaurids.  They get results close to 
previous estimates, claiming this proves the method works for spinosaurids.  
Yet Suchomimus' skull is only known from the snout, while the proximal snout 
of Baryonyx is missing.  Even worse, the incomplete skeleton of Suchomimus 
is headless!  The snout and other cranial elements (quadrates and dentaries) 
are referred specimens with unknown scaling relative to the postcrania!  So 
we don't know the skull length of either taxon or the skeletal length 
relative to skull length in Suchomimus, making it impossible to confirm 
their method works on spinosaurids.

Then they estimate Spinosaurus' length and mass based on their equations and 
Dal Sasso et al.'s 1.75 meter skull length estimate.  They get a much 
shorter (14.34 m) and heavier (20.9 tons) animal than Dal Sasso et al. did.  
Why?  I'd say because of what I wrote above regarding their biased sample of 
large theropods.  The large theropods are mostly short-tailed and massive 
tyrannosaurids, and the largest are especially massive tyrannosaurines.  
Spinosaurus probably had a much longer tail than tyrannosaurids and was more 
lightly built, yet because it's skull is longer than any theropod used in 
the model, it will be forced into the trendline formed by the largest 
tyrannosaurids.

Yet it gets worse, as the authors carpriciously revise Spinosaurus' skull 
length to 1.5 meters, resulting in a total length of 12.57 meters, "a value 
very close to the body length of the largest theropods known (e.g., 
Tyrannosaurus, Carcharodontosaurus, Giganotosaurus)."  "Consequently, the 
body length and body mass estimates for this individual are subject to 
change and could be considerably lower?possibly 12.57 m and 12 tons?within 
the range of the largest theropods known."  The Spinosaurus skull 
reconstruction is rather crammed posteriorly as it is, with the 
laterotemporal fenestra extending below the orbit and such.  I don't see how 
you could make a realistic reconstruction 1.5 meters long.  Look at figure 4 
in Dal Sasso et al. (2005) and see if you think MSNM V4047 was only 22% 
larger than Suchomimus.

"It is generally accepted that large hadrosaurs, including those smaller 
than the 17-m-long Shantungosaurus, were primarily quadrupedal 
(Brett-Surman, 1997; Dilkes, 2001), which suggests
that bipedalism at extremely large body size is impractical. In light of the 
results presented above, we conclude that it is doubtful that a bipedal 
theropod with a mass exceeding 20 tons could have existed. Although it was a 
large theropod, Spinosaurus aegyptiacus was probably no larger than the 
currently known largest tyrannosaurids and carcharodontosaurids, a size 
close to the biomechanical limit for strictly bipedal animals (Henderson, 
2005)."

Or maybe it suggests that spinosaurids weren't hadrosaurs.  Hadrosaurs of 
any size were primarily quadrupedal.  It's just as meaningless as showing 
large sauropods were quadrupedal.

In conclusion, while it would be nice to have a formula to estimate theropod 
length and weight with based on skull size, I believe it's wishful thinking. 
 Therrien and Handerson show you can get decent results if you limit taxon 
selection, but it's much better to use related taxa and scale to known 
proportions.  Even better, find allometric relationships within known taxa 
(e.g. tyrannosaurids).

Mickey Mortimer