138 



M.S.Y. LEE AND J.D. SCANLON 



similar morphology, which is functionally correlated with 

 anguilliform swimming (Scanlon et ah, 1999; Lee et al., 1999). It 

 can thus be concluded that Mesoleptos was marine. This is also 

 supported by the morphology of the posterior ribs . While they do not 

 exhibit the histological features of true pachyostosis, they are never- 

 theless robust (in the MCSNT and HUJ specimens) and might have 

 served to reduce bouyancy, much like the pachyostotic ribs in other 

 marine reptiles. The type of M. zendrinii, from the Comen locality, 

 also comes from deposits dominated by marine fish (Gorjanovic- 

 Kramberger, 1892) and is associated with aigialosaurs, dolichosaurs 

 and pachyophiids. This also presumably applies to the MCSNT 

 specimen (although its collection details have not been recorded it is 

 probably from either Comen or Lesina). Marine habits of the present 

 specimen are also implied by the position of the 'Ein Jabrud locality 

 far from the palaeoshoreline (Scanlon et al., 1999), and the articu- 

 lated nature of the preserved elements suggesting in situ preservation. 

 The laterally compressed body and small limbs suggest that 

 Mesoleptos swam primarily by lateral undulation, holding its limbs 

 against its flanks (Carroll 1985; Lee 1999). In such forms, most of 

 the propulsion occurs by movements of the tail, and to some extent 

 the posterior region of the trunk. This is consistent with the observa- 

 tion that the posterior trunk region is most laterally compressed in 

 Mesoleptos (the tail is unknown). The forelimbs and hindlimbs, 

 however, were still large and well ossified enough to have been 

 functional. They may have been used for slow locomotion ('walk- 

 ing') along the seabed, where (with the help of buoyancy) they could 

 have supported the body. Alternatively, or additionally, they may 

 have been used for forays on the shore. 



PHYLOGENETIC RELATIONSHIPS OF 

 MESOLEPTOS 



All previous interpretations of the morphology and relationships of 

 Mesoleptos were based either on poorly preserved and inadequately 

 described material (the type), or on a composite of the type with the 

 referred Novak specimen (Gorjanovic-Kramberger, 1892) which is 

 clearly distinct from M. zendrinii in rib morphology. Gorjanovic- 

 Kramberger's inclusion of Mesoleptos in Varanidae was 

 'phenetically' based on its long ribs, as distinct from the shorter and 

 more uniform ribs of Aigialosaums (as then interpreted) and 

 dolichosaurs. However, he recognised it as marine in habits and thus 

 by no means a typical varanid. Nopcsa classified Mesoleptos doubt- 

 fully as an aigialosaur (1903), but later placed it in a separate 

 subfamily (Mesoleptinae) with Eidolosaurus, close to both 

 aigialosaurines and dolichosaurines within Dolichosauridae (1923). 



McDowell and Bogert (1954) returned Mesoleptos and 

 Eidolosaurus, again doubtfully, to Aigialosauridae, but also briefly 

 considered that they might be related to the living earless monitor, 

 Lanthanotus. Hoffstetter (1955) also retained Mesoleptos as a poss- 

 ible aigialosaurid, while recognising Eidolosaurus as a dolichosaur 

 and suggesting that Pachyvaranus might represent a distinct family. 

 Romer (1956) placed both Mesoleptos and Eidolosaurus, with 

 question marks, in Dolichosauridae. 



The current state of understanding of these groups is perhaps best 

 indicated by the fact that the systematic conclusion to Calligaris' 

 (1988) review was formed by a summary of Nopcsa's (1923) 

 classification, without substantial additions or revisions. That these 

 groups have been poorly studied recently is highlighted by Carroll 

 and DeBraga's (1992) statement that only five species had been 

 assigned to Aigialosauridae, and did not mention either Mesoleptos 

 (assigned to Aigialosauridae by Nopcsa, 1903, Camp, 1923, and 



McDowell and Bogert, 1954), Eidolosaurus (assigned by Nopcsa, 

 1923, and McDowell and Bogert, 1954) or Pachyvaranus (assigned 

 by Arambourg and Signeux, 1952). 



The relationships of Mesoleptos, therefore, remain unresolved. 

 While a robust assessment will have to await more complete material, 

 a preliminary analysis is undertaken here. Morphological informa- 

 tion from the MCSNT and HUJ specimens (based on examination of 

 specimens) and the type (based on published descriptions) was used 

 in order to evaluate its phylogenetic relationships. Mesoleptos was 

 added to the data matrix used in the most recent comprehensive 

 analysis of squamates (Lee 2002); this matrix includes 248 osteo- 

 logical characters, used here, and addresses recent criticisms of 

 various characters (Rieppel and Zaher 2000). Recently described (or 

 redescribed) elongate marine squamates were also included in this 

 matrix: Pachyrhachis (Lee and Caldwell, 1998), Pachyophis (Lee et 

 al., 1999), Adriosaurus (Lee and Caldwell, 2000) and dolichosaurs 

 (Coniosaurus and Dolichosaurus; Caldwell and Cooper, 1999; 

 Caldwell, 1999, 2000). Coniosaurus and Dolichosaurus are here 

 combined into a single taxon, Dolichosauridae sensu stricto, based 

 on the observations that the comparable parts of the two taxa appear 

 almost identical, they overlap stratigraphically, and as noted by 

 Caldwell (2000) one of the Coniosaurus species might be synony- 

 mous with D. longicollis. Character codings for all taxa (except 

 Haasiophis) in this matrix, including the marine fossil forms, are 

 based on direct examination of the material. As descriptions of the 

 remaining marine squamates discussed above are dated, and they 

 have yet to be restudied, they have not been included in the analysis. 

 The full matrix is presented elsewhere (Lee 2002) and only the 

 (new) character codings for Mesoleptos are listed here (Appendix 

 1 ). The full matrix (including Mesoleptos) used in this analysis has 

 been deposited in TreeBase (http://www.treebase.org/treebase/). 



The enlarged data matrix with Mesoleptos was analysed using the 

 heuristic algorithm of PAUP* (Swofford, 1999) employing 100 

 random addition sequences. Two analyses were performed, with 

 multistate characters ordered according to morphoclines where 

 possible, or with all multistate characters unordered, to see if the 

 phylogenetic analyses were contingent on assumptions of character 

 state transitions. The degree of support for each grouping was 

 ascertained by the support index (Bremer, 1988), calculated in 

 PAUP using batch commands generated by TreeRot Version 2b 

 (Sorenson, 2000). These commands were modified so that each 

 heuristic search employed 100 rather than 20 random addition 

 sequences. Nonparametric bootstrapping (1000 heuristic replicates 

 each employing 100 random addition sequences) was also used to 

 assess the robustness of each clade. As there were no fully specified 

 a priori hypotheses for Mesoleptos and all other squamates, 

 Templeton tests are inappropriate and were not performed (Goldman 

 era/., 2000). 



Phylogenetic affinities 



In the ordered analysis, three most parsimonious trees were found, 

 each of length 672, consistency index = 0.46, retention index = 0.7 1 . 

 The strict consensus is shown in Fig. 3A, along with nodal supports. 

 In the unordered analysis, 4 most parsimonious trees were found 

 when only branches with unequivocal character support were re- 

 tained, each of length 639, consistency index = 0.48, retention index 

 = 0.71. The strict consensus is shown in Fig. 3B. along with nodal 

 supports. 



The basic topologies of the ordered and unordered consensus 

 trees are similar to each other and largely unchanged from that the 

 previous study (Lee, 2002), so that diagnoses of all the clades within 

 Squamata are not repeated here. The characters diagnosing additional 



