First fossil of Varanus Merrem, 1820 (Squamata: Varanidae) from the Miocene Siwaliks of Pakistan

ABSTRACT In the ‘50s of the XX century, a German expedition lead by Richard Dehm collected a large amount of fossil remains from northern Pakistan. Among these was an isolated trunk vertebra of a lizard, which is here referred to Varanus sp. The collecting site of this specimen is not precisely known, but it most likely comes from middle to early late Miocene Siwalik sediments of the Chinji Formation. This is the first published record of a fossil lizard from the Neogene of Pakistan and adds to the very patchy record of Varanus Merrem, 1820 in Asia. It further supports previous reconstructions of a warm climate for the middle to early late Miocene of the Siwaliks.

Fossils of crocodylians and turtles will be the subject of future dedicated works, but we here add to the squamate fauna by describing the only other fossil of this group of reptiles that we were able to find in the SNSB-BSPG collections: an isolated lizard vertebra. No precise locality was available for this fossil, but some information can be retrieved from the labels (Fig. 1). It was collected by members of Dehm's expedition on the first day of February 1956 ( Fig. 1A), somewhere between Marianwala and Kas. Both these villages are located south of Chinji (Fig. 2), in an area where the Chinji Formation crops out. Numerous collecting sites are marked in the surroundings of these two villages in the original maps from the expedition (AV, pers. obs.), but neither on the maps nor on the original diaries any lizard vertebra is noted despite the original label clearly states that the fossil was recognised as such (Fig. 1A). The Chinji Formation is currently considered middle to early late Miocene in age (11.2-14.2 Ma; Barry et al. 2002;Patnaik 2016).

MATERIAL AND METHODS
The subject of this study is a single vertebra, numbered SNSB-BSPG 1956 II 2010. The vertebra was photographed with a Leica M205 microscope equipped with the Leica Application Suite V 4.10 at the Università degli Studi di Torino. Photographs of the labels were taken with a Samsung WB252F Digital Camera. The terminology used for the vertebral laminae follows that used by Tschopp (2016).
description SNSB-BSPG 1956 II 2010 (Fig. 3) is a large and elongated trunk vertebra, with a procoelous centrum. It is moderately well preserved, but the anterior part and most of its left lateral surface are covered by a hard sandstone matrix that was not removed in order not to damage the fossil. In ventral view (Fig. 3A), the centrum is subtriangular, with posteriorly converging and straight lateral margins. The ventral surface is flat, without keel. In spite of the matrix covering, the ventral margin of the anterior cotyle is clearly located posteriorly compared to the dorsal margin (Fig. 3A); thus, the anterior cotyle was facing anteroventrally. In posterior view (Fig. 3B), the centrum is dorsoventrally compressed. The posterior condyle is strongly eroded, but a distinct precondylar constriction is still recognisable in ventral view (Fig. 3A). The width of the centrum at the constriction is 5 mm, whereas the anterior cotyle can be estimated as about 8 mm wide. This would indicate a constriction/cotyle width ratio of 0.63. On the right lateral surface (Fig. 3C), the base of the synapophysis is visible, even though the rest of the latter is not preserved anymore. Nevertheless, it was dorsoventrally elongated. The vertebra has no postzygoprezygapophyseal lamina, but a very poorly distinct posterior centrosynapophyseal lamina is present (Fig. 3C). In dorsal view (Fig. 3D), a distinct interzygapophyseal constriction is visible. Being covered by sediment, the prezygapophyses are not clearly visible. The postzygapophyses are better exposed. Zygapophyses are subcircular to suboval and dorsally inclined of about 45°. Each prezygapophysis is separated by the pars tectiformis of the neural arch by a deep groove (Fig. 3D). The pars tectiformis does not bear a zygosphene, and no zygantrum is visible on the posterior edge of the neural arch. The posterior half of the latter is raised compared to the pars tectiformis ( Fig. 3E, 3F). The neural spine is well developed, even though broken along its entire length. It starts at the anterior margin of the neural arch and is laminar for most of its length ( varanid. This structure has been also recognised, but not named, by Smith et al. (2008), who reported its sharp development in Varanus and Saniwa, but not in Heloderma. In contrast, it appears to be absent in other lizard groups, such as lacertids (Tschopp 2016) (Hoffstetter 1964;Head 2005), which is dated between 10.1 and c. 3.5 Ma, late Miocene to early Pliocene, by Barry et al. (2002). Given that, fossils coming from this locality should be younger than the herein studied vertebra supposedly coming from the Chinji Formation. Unfortunately, we were unable to find Varanus remains from Winnewala in the SNSB-BSPG collection, where they should be stored, and consequently to confirm this occurrence and potentially make comparisons with SNSB-BSPG 1956II 2010 In  (Falconer 1868;Lydekker 1886Lydekker , 1888Hocknull et al. 2009;Sen et al. 2017), as well as the Quaternary of India, Indonesia, and Thailand (Falconer 1868;Lydekker 1886Lydekker , 1888Hooijer 1972;Patnaik et al. 2008;Hocknull et al. 2009;Suraprasit et al. 2016). Most of the Quaternary remains are referred to extant species, whereas all the Neogene ones are not identified at the species level. The only exception to this frame is Varanus sivalensis, an extinct species described from the Pliocene to Early Pleistocene of India (Falconer 1868;Lydekker 1886Lydekker , 1888Hocknull et al. 2009). However, Hocknull et al. (2009 casted doubts on the real taxonomic identity of the three specimens assigned to this taxon, a humerus and two vertebrae: the authors stated that the humerus, which has Pliocene age, might indeed show sufficient differences to sustain its own taxonomic status, whereas the dorsal vertebrae, found in Early Pleistocene deposits, are comparable with Varanus salvator in their morphology and size. Considering all of this, our knowledge of the evolutionary history of Varanus in Asia is still rather patchy and clearly needs more in-depth studies based on the recovery of further material. This bears significance not only for the history of this clade in this continent, but also for its European and African representatives. The biogeographical origin of the genus Varanus is still a debated issue (Villa et al. 2018), with both African (Fuller et al. 1998;Ast 2001;Holmes et al. 2010) and Eurasian (Amer & Kumazawa 2008;Conrad et al. 2012;Ivanov et al. 2018) origins proposed as possible alternatives. Nevertheless, the origin is to be found in the Paleogene, as indicated by the oldest Varanus remains coming from the Eocene of Egypt (Holmes et al. 2010). The absence of confidently diagnosable Varanus species prior to the early Miocene and, in Asia, even to pre-Quaternary times, however, hinders a better understanding of this problem. Detailed identification GEODIVERSITAS • 2022 • 44 (7) of the Asian fossil monitor lizards could have a direct role in unveiling the origin of the populations that inhabited Europe at least from the early Miocene (Delfino et al. 2013) to the Middle Pleistocene (Georgalis et al. 2017). A recent revision of European monitors by Villa et al. (2018) suggested the presence of at least two lineages in Europe, the second one (represented by Varanus marathonensis from the latest middle to the late Miocene) originating from an Asiatic ancestor. It would be, thus, interesting to comprehend if and which of the Asian Varanus, including the one from Pakistan here described, could be related to this dispersal wave.
This cannot be stated considering only morphological characters, given that known Asian remains of Varanus are represented in most cases by poorly-significant vertebral material. Nevertheless, Siwalik snakes could be potentially of help in drawing some hypotheses about relationships between European and Pakistani squamate faunas. Most Siwalik snake genera and species are not shared with Europe (Head 2005), thus suggesting ecological barriers and different biogeographic units. Based on stratigraphic distribution of snake fossils, Head (2005: 30) stated that "partition of European and South Asian biogeographic theaters occurred no later than middle Miocene". However, it should be also noted that the pattern reported for snakes does not reflect the mammalian palaeobiogeographic dynamics (Patnaik 2016): as a matter of fact, mammal faunal exchanges were indeed present between the Indian subcontinent and both Africa and likely Europe in the considered time interval. Further data are therefore needed to shed light on the biogeographic relationships of Neogene lizards from Pakistan and the Indian subcontinent as a whole.
In palaeoenvironmental terms, Varanus indicates a warm palaeoclimate (mean annual temperature ranging from 14.8°C to 28.1°C according to Böhme 2003) for the Chinji Formation. This partially agrees with the landscape reconstructions based on palaeobotanical studies (Srivastava et al. 2014;Patnaik 2016) suggesting dominance of evergreen to deciduous forests reflecting warm and humid conditions with high rainfall during the middle Miocene of the Siwaliks. The Siwalik sediments in Pakistan deposited in a fluvial environment (Barry et al. 2002;Head 2005). As far as the Chinji Formation is concerned, the riverine system originated fine-grained sandstone deposits that are still present as a covering on at least some of the fossil remains, such as the herein studied Varanus vertebra but also other fossils stored in the SNSB-BSPG (AV, pers. obs.). The most abundant squamate found in the whole Pakistani Siwalik sequence is A. dehmi (Hoffstetter 1964;Head 2005), which is considered a fully-aquatic animal. This is particularly evident in the Chinji and Nagri Formations, but subsequently, there is a rise in the occurrence of terrestrial and semi-aquatic taxa from around 10 Ma onwards (Dhok Pathan Formation;Head 2005). Based on rough correlations between monsoonal precipitation patterns, the transition towards a C4-dominated vegetation, and the increasing diversity in Siwalik snakes, Head (2005) suggested a possible link between the latter and increasing seasonality in the area and diversification of the available habitats, which would have favoured the onset of a more diverse reptilian fauna. Varanus currently displays a wide variety of ecological adaptation and includes species strongly linked with water environments (e.g., V. salvator, which nowadays is widely spread in Southeast Asia; Pianka et al. 2004;Eidenmullen 2007). Acquiring information on the ecological preference of the species represented by SNSB-BSPG 1956 II 2010 is impossible, even though the finding of only one remain in the Pakistani Siwaliks may suggest a low relation with water for the represented monitor when considering the pattern showed by snakes. Further studies of squamate fossils from the area are strongly needed to enhance our knowledge of their distribution, evolution, and ecology in the Neogene of Pakistan.

Acknowledgements
The study of SNSB-BSPG 1956 II 2010 benefited significantly from the help of Gertrud Rössner (Munich), who assisted AV in defining the possible provenance and age of the specimen and shared information about Dehm's expedition to Pakistan. We also thank Madelaine Böhme (Tübingen) for discussions about the Winnewala Varanus occurrence, the two reviewers, Davit Vasilyan (Porrentruy) and Martin Ivanov (Brno), for their valuable comments that improved an earlier version of the manuscript, as well as the editor Emmanuel Côtez (Paris). This work originated during a visit to the SNSB-BSPG collection in 2017 supported by an EAVP Research Grant provided to AV by the European Association of Vertebrate Palaeontologists. Oliver Rauhut (Munich) is thanked for help while visiting the collection back then. During part of the development of this work, AV was funded by the Alexander von Humboldt Foundation through a Humboldt Research Fellowship. MD is supported by Fondi di Ateneo dell'Università di Torino (2018-2019), Generalitat de Catalunya (CERCA Program), and Spanish Agencia Estatal de Investigación (CGL2016-76431-P, AEI/FEDER, EU). The original maps used for Fig. 2 are freely available at d-maps (https://d-maps.com/).