Early three-family classification Though
Edward Drinker Cope had originally recognized
Elasmosaurus as a plesiosaur, in an 1869 paper he placed it, with
Cimoliasaurus and
Crymocetus, in a new order of
sauropterygian reptiles. He named the group Streptosauria, or "reversed lizards", due to the orientation of their individual vertebrae supposedly being reversed compared to what is seen in other vertebrate animals. He subsequently abandoned this idea in his 1869 description of
Elasmosaurus, where he stated he had based it on Leidy's erroneous interpretation of
Cimoliasaurus. In this paper, he also named the new family Elasmosauridae, containing
Elasmosaurus and
Cimoliasaurus, without comment. Within this family, he considered the former to be distinguished by a longer neck with compressed vertebrae, and the latter by a shorter neck with square, depressed vertebrae. In subsequent years, Elasmosauridae came to be one of three groups in which plesiosaurs were classified, the others being the
Pliosauridae and
Plesiosauridae (sometimes merged into one group). In 1874
Harry Seeley took issue with Cope's identification of
clavicles in the shoulder girdle of
Elasmosaurus, asserting that the supposed clavicles were actually scapulae. He found no evidence of a clavicle or an
interclavicle in the shoulder girdle of
Elasmosaurus; he noted that the absence of the latter bone was also seen in a number of other plesiosaur specimens, which he named as new elasmosaurid genera:
Eretmosaurus,
Colymbosaurus, and
Muraenosaurus.
Richard Lydekker subsequently proposed that
Elasmosaurus,
Polycotylus,
Colymbosaurus, and
Muraenosaurus could not be distinguished from
Cimoliasaurus based on their shoulder girdles, and advocated their synonymization at the genus level. Seeley noted in 1892 that the clavicle was fused to the coracoid by a suture in elasmosaurians, and was apparently "an inseparable part" of the scapula. Meanwhile, all plesiosaurs with two-headed neck ribs (the Plesiosauridae and Pliosauridae) had a clavicle made only of
cartilage, such that
ossification of the clavicle would turn a "plesiosaurian" into an "elasmosaurian".
Samuel Wendell Williston doubted Seeley's usage of neck ribs to subdivide plesiosaurs in 1907, opining that double-headed neck ribs were instead a "primitive character confined to the early forms".
Charles Andrews elaborated on differences between elasmosaurids and pliosaurids in 1910 and 1913. He characterized elasmosaurids by their long necks and small heads, as well as by their rigid and well-developed scapulae (but atrophied or absent clavicles and interclavicles) for forelimb-driven locomotion. Meanwhile, pliosaurids had short necks but large heads, and used hindlimb-driven locomotion.
Refinement of plesiosaur taxonomy Although the placement of
Elasmosaurus in the Elasmosauridae remained uncontroversial, opinions on the relationships of the family became variable over subsequent decades. Williston created a revised taxonomy of plesiosaurs in a
monograph on the
osteology of reptiles (published posthumously in 1925). He provided a revised
diagnosis of the Elasmosauridae; aside from the small head and long neck, he characterized elasmosaurids by their single-headed ribs; scapulae that meet at the midline; clavicles that are not separated by a gap; coracoids that are "broadly separated" in their rear half; short ischia; and the presence of only two bones (the typical condition) in the epipodialia (the "forearms" and "shins" of the flippers). He also removed several plesiosaurs previously considered to be elasmosaurids from this family due to their shorter necks and continuously meeting coracoids; these included
Polycotylus and
Trinacromerum (the
Polycotylidae), as well as
Muraenosaurus,
Cryptoclidus,
Picrocleidus,
Tricleidus, and others (the
Cryptoclididae). In 1940
Theodore White published a hypothesis on the interrelationships between different plesiosaurian families. He considered Elasmosauridae to be closest to the Pliosauridae, noting their relatively narrow coracoids as well as their lack of interclavicles or clavicles. His diagnosis of the Elasmosauridae also noted the moderate length of the skull (i.e., a mesocephalic skull); the neck ribs having one or two heads; the scapula and coracoid contacting at the midline; the blunted rear outer angle of the coracoid; and the pair of openings (fenestrae) in the scapula–coracoid complex being separated by a narrower bar of bone compared to pliosaurids. The cited variability in the number of heads on the neck ribs arises from his inclusion of
Simolestes to the Elasmosauridae, since the characteristics of "both the skull and shoulder girdle compare more favorably with
Elasmosaurus than with
Pliosaurus or
Peloneustes." He considered
Simolestes a possible ancestor of
Elasmosaurus.
Oskar Kuhn adopted a similar classification in 1961.
Samuel Paul Welles took issue with White's classification in his 1943 revision of plesiosaurs, noting that White's characteristics are influenced by both preservation and
ontogeny. He divided plesiosaurs into two superfamilies, the
Plesiosauroidea and
Pliosauroidea, based on neck length, head size, ischium length, and the slenderness of the humerus and femur (the propodialia). Each superfamily was further subdivided by the number of heads on the ribs, and the proportions of the epipodialia. Thus, elasmosaurids had long necks, small heads, short ischia, stocky propodialia, single-headed ribs, and short epipodialia. Pierre deSaint-Seine in 1955 and
Alfred Romer in 1956 both adopted Welles' classification. Per Ove Persson, however, considered Welles' classification too simplistic, noting in 1963 that it would, in his opinion, erroneously assign
Cryptoclidus,
Muraenosaurus,
Picrocleidus, and
Tricleidus to the Elasmosauridae. Persson refined the Elasmosauridae to include traits such as the crests on the sides of the neck vertebrae; the hatchet-shaped neck ribs at the front of the neck; the fused clavicles; the separation of the coracoids at the rear; and the rounded, plate-like pubis. He also retained the Cimoliasauridae as separate from the Elasmosauridae, and suggested, based on comparisons of vertebral lengths, that they diverged from the Plesiosauridae in the Late Jurassic or Early Cretaceous. In 2009, F. Robin O'Keefe and Hallie Street synonymized the Cimoliasauridae with the Elasmosauridae, noting that most of the diagnostic traits previously established to distinguish them are also found in elasmosaurids.
Modern phylogenetic context Carpenter's 1997 phylogenetic analysis of plesiosaurs challenged the traditional subdivision of plesiosaurs based on neck length. He found that
Libonectes and
Dolichorhynchops shared characteristics such as an opening on the palate for the
vomeronasal organ, the plate-like expansions of the
pterygoid bones, and the loss of the
pineal foramen on the top of the skull, differing from the pliosaurs. While polycotylids had previously been part of the Pliosauroidea, Carpenter moved polycotylids to become the
sister group of the elasmosaurids based on these similarities, thus implying that polycotylids and pliosauroids evolved their short necks independently. F. Robin O'Keefe likewise included polycotylids in the Plesiosauroidea in 2001 and 2004, but considered them more closely related to the Cimoliasauridae and Cryptoclididae in the
Cryptocleidoidea. Some analyses continued to recover the traditional groupings. In 2008 Patrick Druckenmiller and Anthony Russell moved the Polycotylidae back into the Pliosauroidea, and placed
Leptocleidus as their sister group in the newly named
Leptocleidoidea; Adam Smith and Gareth Dyke independently found the same result in the same year. However, in 2010 Hilary Ketchum and Roger Benson concluded that the results of these analyses were influenced by inadequate sampling of species. In the most comprehensive phylogeny of plesiosaurs yet, they moved the Leptocleidoidea (renamed the Leptocleidia) back into the Plesiosauroidea as the sister group of the Elasmosauridae; subsequent analyses by Benson and Druckenmiller recovered similar results, and named the Leptocleidoidea–Elasmosauridae grouping as
Xenopsaria. The content of Elasmosauridae also received greater scrutiny. Since its initial assignment to the Elasmosauridae, the relationships of
Brancasaurus had been considered well supported, and it was recovered by O'Keefe's 2004 analysis However, Ketchum and Benson's analysis instead included it in the Leptocleidia, Their analysis also moved
Muraenosaurus to the Cryptoclididae, and
Microcleidus and
Occitanosaurus to the Plesiosauridae; Within the Elasmosauridae,
Elasmosaurus itself has been considered a "wildcard taxon" with highly variable relationships. and in 2008 Druckenmiller and Russell placed it as part of a
polytomy with two groups, one containing
Libonectes and
Terminonatator, the other containing
Callawayasaurus and
Hydrotherosaurus. In 2020, O'Gorman formally synonymized Styxosaurinae with Elasmosaurinae based on the inclusion of
Elasmosaurus within the group, and also provided a list of diagnostic characteristics for the clade. In 2021 a new topology placed
Cardiocorax as a sister taxon of
Libonectes, representing an older lineage of elasmosaurids in the Maastrichtian.
Topology A: Benson
et al. (2013) An alternative hypothesis suggested by F. Robin O'Keefe and Hallie Street in 2009 is that aristonectines instead belong to a
family—
Aristonectidae—which is unrelated to elasmosaurids and is instead the sister group to
Polycotylidae within the larger clade "Cryptocleidoidea". ==Paleobiology==