I've just read the description of Rapetosaurus and have some comments.
"Currently we recognize two titanosaur taxa in the Maevarano Formation that are distinguished, in part, by their caudal vertebrae. Deperet's T. madagascariensis syntype includes both caudal morphologies and is therefore considered a nomen dubium."
Ughh. Damn the current trend to trash old species just to name a new one. They should designate one of the "Titanosaurus" madagascariensis caudals the lectotype and name one of the new genera "genus A" madagascariensis. "T." madagascariensis is obviously not nomen dubium if its remains can be referred to two other species. It's simply a chimaera and must be split.
"Rapetosaurus also offers the best opportunity yet to resolve the phylogenetic position of two controversial Mongolian sauropods, Nemegtosaurus and Quaesitosaurus."
I agree here, Nemegtosaurus and Quaesitosaurus are definitely titanosaurids. I was a believer in diplodocoid nemegtosaurids before, but Rapetosaurus is just too similar to nemegtosaurs and has too many titanosaurid postcranial characters. Well Upchurch, you did your best :-)
A rather large phylogenetic analysis was performed (228 characters, 16 taxa), with the following result-
Good job finally including Titanosaurus. They call the (Titanosaurus + Alamosaurus + Opisthocoelocaudia + Saltasaurus + Neuquensaurus) clade Saltasaurinae. However, I thought Saltasaurinae was defined as all taxa closer to Saltasaurus than Opisthocoelocaudia. I think Titanosaurinae should have priority over Saltasaurinae if the latter is used like Curry Rogers and Forster do. Also, Titanosauridae should be somewhere, why have people forgotten it? Making Rapetosaurus, Nemegtosaurus and Quaesitosaurus diplodocoids adds 29 steps. I dislike the inclusion of Euhelopus in the Titanosauria. Wonder what happens when Mamenchisaurus or Rebbachisaurus are included. When Mamenchisaurus is added to Wilson and Sereno's matrix, it clades right below Omeisaurus, but only 2 more steps are needed to make it go into the Titanosauria with Euhelopus. Eight more steps make the (Omeisaurus, Mamenchisaurus, Euhelopus) clade monophyletic and fall below neosauropods.
Below is a summary of the basic facts about Rapetosaurus. It is not a details segment, as I have not added a description or relationships section. Perhaps I'll write one later.
Rapetosaurus Curry Rogers and Forster 2001
R. krausei Curry Rogers and Forster 2001
Etymology- "David Krause's giant lizard", after Rapeto, a mischevious giant in Malagasy folklore.
Anembalemba Member of Maevarano Formation, Madagascar
Maastrichtian, Late Cretaceous
Holotype- (UA 8698) maxillae, nasals, lacrimal, jugal, quadrate, pterygoids, partial basioccipital, paroccipital process, dentary, surangular, angular, twenty-four teeth
Referred- (FMNH PR 2184-2192, 2194, 2196, 2197, 2210) exoccipital, opisthotic, laterosphenoid, supraoccipital, frontals, prefrontals, surangular, parietal, squamosal, mid-caudal centrum
(FMNH PR 2209) (juvenile) 75% complete skeleton
(UCB 92829) mid-caudal centrum
Diagnosis- expanded antorbital fenestra extends over tooth row; preantorbital fenestra positioned posterior to antorbital fenestra; subnarial foramen anteriorly positioned and dorsoventrally elongate; jugal process of maxilla posterodorsally elongate and narrow; frontals with median dome; quadrate with V-shaped quadratojugal articulation; supraoccipital with two anteriorly directed median parietal processes; pterygoid with extremely shallow basipterygoid articulation and dorsoventrally expanded anterior process; basipterygoid processes diverge only at distal extremes; dentary with 11 alveoli that extend two-thirds the length of the element; gracile cylindrical teeth with high-angle planar wear facets; 16 cervical vertebrae with constricted neural canals and continuous pre- and postspinal coels devoid of pre- or postspinal laminae; cervicalneural spines with proximal bifurcation and three pneumatized coels bounded by discrete laminae; 11 dorsal vertebrae with deep lateral pleurocoels; dorsal neural spines with strong pre- and postspinal laminae in deeply excavated anterior and posterior coels; dorsals with median interpre- and interpostzygapophyseal laminae; middle and posterior dorsals with divided spinodiapophyseal lamina; six sacral centra with deep lateral pleurocoels; all caudal centra procoelous with convex ventral margin lacking excavation; anterior caudal centra broad transversely and anteroposteriorly compressed; middle-posterior caudal centra with constant length:width ratio; anterior-middle caudal neural spines with spinoprezygapophyseal, prespinal and postspinal laminae on rectangular and anteriorly positioned neural arches; chevrons throughout 80% of tail; iliac peduncle of ischium comprises one-quarter of acetabulum; ischial peduncle of ilium low and poorly developed; pubis more than twice as long as ischium; scapula and coracoid with equal glenoid contribution; scapular blade not distally expanded; humerus/femur length quotient 0.80; radius and ulna with oblique interosseus ridges.
Those who want the pdf, contact me offlist.