Giant lizards occupied herbivorous mammalian ecospace during the Paleogene greenhouse in Southeast Asia

Mammals dominate modern terrestrial herbivore ecosystems, whereas extant herbivorous reptiles are limited in diversity and body size. The evolution of reptile herbivory and its relationship to mammalian diversification is poorly understood with respect to climate and the roles of predation pressure and competition for food resources. Here, we describe a giant fossil acrodontan lizard recovered with a diverse mammal assemblage from the late middle Eocene Pondaung Formation of Myanmar, which provides a historical test of factors controlling body size in herbivorous squamates. We infer a predominately herbivorous feeding ecology for the new acrodontan based on dental anatomy, phylogenetic relationships and body size. Ranking body masses for Pondaung Formation vertebrates indicates that the lizard occupied a size niche among the larger herbivores and was larger than most carnivorous mammals. Paleotemperature estimates of Pondaung Formation environments based on the body size of the new lizard are approximately 2–5°C higher than modern. These results indicate that competitive exclusion and predation by mammals did not restrict body size evolution in these herbivorous squamates, and elevated temperatures relative to modern climates during the Paleogene greenhouse may have resulted in the evolution of gigantism through elevated poikilothermic metabolic rates and in response to increases in floral productivity.


Introduction
Modern terrestrial herbivore ecosystems are dominated by mammal faunas that originated with the evolution of ungulate folivores during the middle Eocene [1]. Conversely, herbivory is comparatively rare among extant squamates [2]. Squamates do not efficiently metabolize plant matter compared with mammals [3], and digestion requires elevated body temperatures which are correlated to large body size [2,[4][5][6] and restriction to tropical climates for most taxa [7].
The relative roles of physiology and ecological pressures from mammals as constraints on upper body size limits of herbivorous lizards are unknown, however. Direct and indirect interactions with ungulates and carnivorans are known to limit distribution and densities of carnivorous squamates [8][9][10] and the largest extant herbivorous reptiles only occur in insular, mammal-free habitats [3,11], suggesting competitive exclusion or predation pressure may limit maximum body sizes. Conversely, squamate body size can be affected by ambient temperature and food resources [3,12], and maximum body sizes of extant taxa may be limited by Holocene climatic maxima.
Fossil squamates generally demonstrate similar size and diversity patterns as extant herbivores during the Cenozoic, but the squamate fossil record is poorly sampled and generally restricted to the mid and high latitudes of North America and Europe [13][14][15][16]. The relative paucity and geographical restriction of the squamate fossil record confounds efforts to examine the historical relationship between body size and environment relative to faunal competition, climate and historical contingency in poikilothermic herbivores.
Here, we describe a giant acrodontan lizard from the rich, low-latitude vertebrate fossil record of the late middle Eocene Pondaung Formation of central Myanmar that includes a diversity of eutherian mammals, turtles, squamates and crocodylians recovered from siliciclastic sediments representing fluvial depositional environments [17][18][19][20][21]. Analysis of the new acrodontan's inferred diet and estimated body mass in the context of the co-occurring fauna and in comparison to modern vertebrate communities allows us test the relative influences of mammalian competition versus climate regime as a regulating mechanism of herbivorous reptile body size by examining herbivore community structure in past and present vertebrate ecosystems and by estimating minimum paleotemperatures necessary to support a giant poikilothermic herbivore based on the mass-specific metabolic relationship between body size and climate in living herbivorous lizards.  assigned to the taxon on the basis of size; NMMP-KU 0092, partial left dentary; NMMP-KU 1923, partial left dentary; NMMP-KU 1924 -1926, partial right dentaries [20] (figure 1e-p).

(d) Locality and horizon
The type locality is UCMP V96009, a locality number used to designate a stratigraphically low purple mudstone underlying red beds at Thandaung kyitchaung [18,22]

Material and methods (a) Phylogenetic analysis
Molecular and morphological data provide disparate hypotheses of the interrelationships of acrodontans, which limit the ability to resolve the phylogenetic status of fossil taxa [24]. To determine the interrelationships of Barbaturex to extant acrodontans sensu [25], we coded all preserved characters for the only extensive morphological phylogenetic analysis of constituent taxa [26]. Because Barbaturex remains preserve only a small fraction of described characters (5/122), we estimated the phylogenetic position of the taxon by optimizing character distributions onto tree topologies derived from combined morphological and molecular sequence data [2,27] (figure 2) and only molecular sequence data [33,34] (see the electronic supplementary material, figure S2). Phylogenetic position of Barbaturex was estimated by optimizing character distributions using Mesquite v. 2.75 [35]. Character codings for Barbaturex based on the matrix of [26] are as follows: 57(1), 58(0), 59(0), 65(2), 66(0).

(b) Body size estimation
We estimated maximum body size measured as SVL in mm for Barbaturex by reduced major axis linear regression of natural log transformed measurements of SVL onto natural log transformed dentary lengths measured in a straight line from the anterior tip of the element to the posterior margin of the lateral coronoid process in mm for extant acrodontans (see the electronic supplementary material,

(c) Body mass comparisons
To examine the status of Barbaturex within the Pondaung vertebrate fauna, we ranked body masses of terrestrial herbivores, omnivores and carnivores and compared them to ranked masses for extant faunas that include the largest extant herbivorous squamates. Body masses were obtained from literature references (see the electronic supplementary material, tables S2-S5). Taxonomic and geographical range data for all extant mammals for comparisons with living herbivorous squamates is from [37], and body masses are from [38]. Maximum body mass reports and estimates for the largest extant herbivorous squamates, as well as Pondaung Formation mammalian body masses, are listed in the electronic supplementary material. Island endemic squamates [39,40] from faunas lacking folivorous mammals could not be compared with the Pondaung fauna, and completely arboreal, nocturnal and granivorous mammal taxa were not included in this analysis as their ecologies are not directly comparable to the examined squamates.

Leiolepis
We derived paleotemperatures as minimum mean annual temperature (MAT) for the Pondaung Formation from body size estimates of Barbaturex based on the relationship between SVL and minimum MAT for the largest living herbivorous squamates (see the electronic supplementary material, table S6) using the metabolic scaling equation from [42]: MAPT = MAT + 3a10 0 C log 10 (SVL B /SVL C ) log 10 Q 10 ; where MAPT is mean annual paleotemperature, SVL B is SVL for Barbaturex, SVL C is SVL for Cyclura nubila, MAT is minimum mean annual temperature within the geographical range of C. nubila (24.68C), a is the metabolic scaling exponent of 0.33 [41], and Q 10 is a mass-specific metabolic rate of 2 -3 for reptiles [43]. We use C. nubila because it is the largest extant herbivorous lizard [5,7] and scaling the size-temperature model on it best fits the size-temperature distributions for other herbivorous taxa ( figure 4).

Results and discussion
We assign Barbaturex to crown Acrodonta relative to priscagamines and more fragmentary stem taxa based on the character combination of reduced numbers of pleurodont anterior teeth, acrodont cheek dentition with deep, continuous interdental grooves and reduction of the splenial to the posterior region of the dentary [25,26,[44][45][46]. Character optimization results in monophyly of Barbaturex þ Uromastycinae for both morphological and molecular topologies, with tree lengths one to two steps shorter than all other alternate topologies based on the morphological þ molecular scaffold (figure 2) and two to four steps shorter on the molecular scaffold (see the electronic supplementary material, figure S2). Character support for this hypothesis includes the absence of the posterior mylohyoid foramen and the ventral orientation of the Meckelian groove at the anterior tip of the dentary (figure 1p). The presence of two pleurodont dentary teeth [32] and a shortened parietal process of the frontal are additionally shared by the clade consisting of Uromastycinae, Barbaturex, and Leiolepis (figure 2). The stratigraphic occurrence of Barbaturex is consistent with our hypothesis of interrelationships (figure 2). The oldest fossil records of unambiguous crown acrodontans consist of uromastycines from the early Eocene of Europe and Asia [31,47], Barbaturex in the late middle Eocene of Asia and possibly the lineage including extant Leiolepis from the late Eocene of North America [24]. The first occurrence of the clade including agamines, draconines and amphibolurines may be early middle Eocene [32], but the late middle Eocene age of the Pondaung record precedes the first occurrences of crown members of this clade, which are early Neogene in age, consistent with molecular divergence estimates [48].
We estimate a SVL of 981 mm + 107 mm, and a mean body mass of 26.7 kg (range of 36.9 -18.9 kg) for Barbaturex based on the relationship of dentary length to body size in extant taxa. Barbaturex was over twice as long as the largest extant agamid [7], and estimated body sizes are larger than all extant and known fossil terrestrial lizards with the exception of extant Varanus komodoensis [49] and extinct V. priscus and Chianghsia nankangensis [50 -53].
Body size, dental morphology and phylogenetic relationships of Barbaturex allow for inference of feeding ecology. Large body size is correlated to herbivory in squamates [5,54], except for the largest varanids, which are carnivores [10,55,56]. Barbaturex lacks dental adaptations for carnivory including recurved, serrated and laterally compressed teeth [57]. Instead, it possesses an acrodont dentition with precise shearing occlusion as indicated by continuous wear facets on mandibular dentition (figure 1). The same occlusal mechanism facilitates herbivory in extant agamids [58,59], and tooth crown morphology in Barbaturex is most similar to herbivorous adult Hydrosaurus. Herbivorous agamids will consume animal protein [60], and omnivorous agamids consume both plants and insects [61]. It is probable that Barbaturex would have opportunistically preyed on invertebrates; however, prey size scales with body size in carnivorous lizards, including iguanians [62] and large body size in Barbaturex probably precluded insectivory as a major component of feeding ecology, at least in mature individuals. The phylogenetic interrelationships of Barbaturex relative to crown agamids are also consistent with herbivorous feeding habits. Optimization of feeding habits on crown acrodontan phylogeny demonstrates that Barbaturex is nested within an omnivorous to herbivorous clade as the sister taxon to fully herbivorous Uromastycinae and bracketed by omnivorous Leiolepis (see figure 3 and electronic supplementary material, figure S2).
Body mass of Barbaturex falls approximately in the middle of size ranges for the Pondaung ungulate guild (figure 3a), and is larger than estimated body masses [63] for the smaller perissodactyls and most artiodactyls. The position of Barbaturex within the body mass distribution of the Pondaung vertebrate fauna is unique relative to extant herbivorous squamates. In all three examined modern faunas, there is no overlap in body mass between ungulate herbivores and squamates. Instead, squamate body masses are far smaller and fall within a range of carnivorous mammals, omnivores and nonruminant herbivores for all modern faunas, including those from open, xeric environments (figure 3b), and both dry and wet forested environments (figure 3c,d). Difference in body size of Barbaturex relative to extant lizards cannot be explained by differences in ungulate body mass because Pondaung ungulates are both larger and smaller than extant taxa. Instead, body mass overlap between Barbaturex and Pondaung Formation mammals indicates that direct competitive exclusion or predation pressure did not restrict body size of these herbivorous squamates, despite differences in metabolic rate and dietary efficiency between poikilothermic and homeothermic herbivores. Similarly, indirect depression of biomass and diversity in extant squamates by environmental modification of ungulates does not appear to have been present in the Pondaung ecosystem based on both size and numbers of recovered specimens of Barbaturex.
Based on the relationship of maximum body size to minimum MAT in extant herbivorous squamates, Barbaturex at 981 mm SVL would require minimum MATs of 27.0-28.48C (range ¼ 26.0-29.98C for SVL range of 874-1088 mm) to maintain efficient metabolism ( figure 4). The late middle Eocene was an interval of cooling from the Middle Eocene climatic optimum [64], but included ice-free poles and extremely warm sea surface temperatures (SSTs) of 22.4-20.58C at 658S [65] during the temporal interval spanning the radiometric age estimates of the Pondaung Formation [23]. Model latitudinal SST gradients for the middle Eocene indicate higher SSTs by 68C relative to modern at a paleolatitude of 138 N [65, figure 3], consistent with MAT differences of approximately 2-58C for Myanmar in the region of locality UCMP V96009 between the late middle Eocene and modern [66].

rspb.royalsocietypublishing.org Proc R Soc B 280: 20130665
Elevated middle Eocene MATs would have allowed for the evolution of large body sizes for a given mass-specific metabolic rate, as inferred for other giant early Paleogene squamates [42], and would have resulted in greater floral productivity than modern ecosystems at low latitudes [67]. Larger body sizes produce increased thermal inertia and may have resulted in elevated temperature-dependent metabolic processes, including digestive efficiency and nutrient uptake [68]. Increased plant productivity affects body size in extant herbivorous iguanians [69], and the comparatively wider range of Pondaung Formation ungulate body mass relative to modern faunas also suggests high primary productivity (figure 3).
Convergent gigantism in Barbaturex and other Cenozoic squamates [42,53] as components of diverse vertebrate ecosystems demonstrates a greater past ecological breadth and diversity than expected from surveying extant herpetofaunas. These discoveries indicate that hypotheses of competitive advantage in extant mammals due to elevated metabolic processes are probably artefacts of modern climate and should not be used as models for inferring historical patterns of diversification and dominance in non-analogue deep time climates.
We thank the National Museum in Yangon, Myanmar for access to fossils, David Blackburn, Jens Vindum (CAS), Kevin Seymour (ROM), James Mead (ETSU) and Chris Bell (TMM) for access to comparative specimens. Chris Bell additionally provided access to obscure literature. Greg Brown (University of Nebraska State Museum) expertly cast, molded, and retrodeformed NMMP-KU specimens. K. Lindsay Eaves read and improved the manuscript. We acknowledge the late Donald E. Savage for the discovery of the first specimen of Barbaturex morrisoni and thank Jack Conrad and an anonymous reviewer for formal review.   (d) Ctenosaura similis, Central America. Colours are: black, ungulates; grey, carnivorous mammals; green, squamates; white, insectivores and non-ungulate herbivores.