Update (10/22/18): Dr. Ken Carpenter has recently published a new paper supporting the view below (and cites me favorably), but I would also be remiss to not recongize Dr. Andrea Cau for having thought up this idea 2 years before me. Sadly, he was not cited in Carpenter's paper. He and I both were unaware of Cau's work.
The last time I wrote about the size of Amphicoelias, I still used Diplodocus as a comparison. One of the comments that was made was that my size estimate was likely wrong, as Amphicoelias was probably a basal diplodocoid, not a diplodocid proper. After a little investigation, it turned out that two phylogenetic analyses have been published that included Amphicoelias, and both found Amphicoelias to be a basal diplodocoid. Whitlock (2011) was one of the studies, and I reproduce the phylogenetic diagram here (with Amphicoelias highlighted by yours truly):
As can be seen here, Amphicoelias is recovered more derived than Amazonsaurus, but more basal than the most basal rebbachisaur, Histriasaurus. This placement has weak support, and as Whitlock comments, "The recovery of Amphicoelias as a basal diplodocoid agrees with the findings of Rauhut et al. (2005), although the latter result may hinge on one or two key scorings. As mentioned above, re-scoring Amazonsaurus in Rauhut et al.’s (2005) matrix results in the placement of Amphicoelias in a large polytomy with essentially all other eusauropods."
On top of that, it is unclear from Whitlock's (2011) analysis, whether A. altus is only scored, or if A. fragillimus is included as well. So this analysis may only apply to A. altus. But no matter, as when I did some comparisons of the overall morphology and proportions of A. fragillimus, it turns out it agrees quite well with rebbachisaurs.
One of things I have noticed about A. fragillimus was that its neural arch is proportionally tall when compared to diplodocids such as Diplodocus, Barosaurus, Apatosaurus and others, as well as in comparison to the dicraeosaurs. Using the dorsal of Rebbachisaurus garasbae, I completed a new reconstruction of the dorsal of A. fragillimus (see above). The gray parts are directly traced from Rebbachisaurus and you can see how well the neural arch proportions match. Obviously, the neural spine in Rebbachisaurus is considerably wider proportionally than that of A. fragillimus, but the neural spine of Limaysaurus tessonei matches quite well, being remarkably thinner side-to-side than in Rebbachisaurus. A thinner neural spine is also seen in Amazonsaurus.
Amazonsaurus dorsal from Carvalho et al (2003)
Another thing that seems to support the hypothesis that A. fragillimus was a rebbachisaur or rebbachisaur-grade diplodocoid is the apparent extreme pneumaticity of the vertebra, with the etymology of its species name meaning "very fragile". Rebbachisaurs such as Nigersaurus and Tataouinea demonstrate extreme pneumaticity in their skeletons, for instance.
How big is a rebbachisaur-ized A. fragillimus?
As shown in the second image above, the reconstructed total height of the vertebra of A. fragillimus is just under 243 cm tall (this happens to be quite close to the reconstructed height when I previously used Diplodocus as a comparison). Compare this to the dorsal vertebrae height of Limaysaurus, at just over 121 cm tall. Assuming fairly similar proportions, A. fragillimus is about twice as large as Limaysaurus in linear dimensions. My recent reconstruction of Limaysaurus gives a total length of 14.3 m, and a mass of about 6.2-6.5 tonnes (depending on the assumed tissue density). This means a total length of about 28.6 m for A. fragillimus, and a mass of about 51-53 tonnes.
Key: blue figure -A. fragillimus, yellow figure - Supersaurus, green figure - Rebbachisaurus, gray scale figure - Limaysaurus
As seen in this comparison I mocked up, A. fragillimus was probably only slightly shorter in length than Supersaurus (silhouette of Supersaurus based on skeletal from Scott Hartman: www.palaeocritti.com/_/rsrc/12…). The lengths come out the same in the image, but in a fair comparison, the neck and tail would be horizontal (with no dorsal or ventral bending) in the Supersaurus outline, which would probably "add" a few more meters to the "length" of the Supersaurus. For comparison, Supersaurus was estimated to be 33-34 m long and mass 35-40 tonnes by Lovelace et al. (2007), but a GDI mass estimate based on the multi-view skeletal therein gives me a volume of about 32.6 m^3 and a mass of about 26 tonnes (assumed the same density they did of 0.8).
So, a rebbachisaur-like A. fragillimus was probably the biggest diplodocoid, but not the longest, and is smaller than the largest titanosaurs, although not by much. (Also note that the length of the hindfeet of a scaled of rebbachisaur-like A. fragillimus are about 144 cm long, which is only slightly shorter than the biggest Broome & Plagne sauropod tracks which are up to 150 cm long, meaning they might not have been record-breakers either, possibly 60 tonnes or less if they were rebbachisaurs.)
Carvalho IDS, Avilla LDS, Salgado L. 2003. Amazonsaurus maranhensis gen. et sp. nov. (Sauropoda, Diplodocoidea) from the Lower Cretaceous (Aptian-Albian) of Brazil. Cretaceous Research 24: 697–713.
Lovelace DM, Hartman SA, Wahl WR. 2008. Morphology of a specimen of Supersaurus (Dinosauria, Sauropoda) from the Morrison Formation of Wyoming, and a re-evaluation of diplodocid phylogeny. Arquivos Do Museu Nacional, Rio De Janeiro 65: 527–544.
Whitlock, JA. A phylogenetic analysis of Diplodocoidea (Saurischia: Sauropoda). Zoological Journal of the Linnean Society, 2011, 161, 872–915.