The Miocene tortoise Testudo catalaunica Bataller, 1926, and a revised phylogeny of extinct species of genus Testudo (Testudines: Testudinidae)

Transcription

1 Zoological Journal of the Linnean Society, 2016, 178, With 12 figures. The Miocene tortoise Testudo catalaunica Bataller, 1926, and a revised phylogeny of extinct species of genus Testudo (Testudines: Testudinidae) 1 Institut Catal a de Paleontologia Miquel Crusafont, Universitat Aut onoma de Barcelona, Edifici ICTA-ICP, Carrer de les Columnes s/n, Campus de la UAB, Cerdanyola del Vall es, Barcelona, Spain 2 Dipartimento di Scienze della Terra, Universita di Torino, Via Valperga Caluso 35, Torino, Italy 3 87, 1er 5a, Sant FOSSILIA Serveis Paleontol ogics i Geol ogics, S.L., Carrer Jaume I num. Celoni, Barcelona, Spain Received 4 July 2015; revised 1 February 2016; accepted for publication 7 February 2016 We provide a taxonomic review of the extinct testudinid Testudo catalaunica, based on published and unpublished material from several Miocene (late Aragonian and early Vallesian) sites of the Vall es-pened es Basin (north-east Iberian Peninsula). We show that Testudo catalaunica irregularis is a junior subjective synonym of T. catalaunica, and further provide an emended diagnosis of the latter based on newly reported material. Contrary to some recent suggestions, this emended diagnosis discounts an alternative attribution of T. catalaunica to Paleotestudo. The latter is merely recognized as a subgenus of Testudo, based on a cladistic analysis that assessed the phylogenetic position of all extant and most extinct species of Testudo currently recognized as valid (including T. catalaunica). Our phylogenetic analysis (which recovers the molecular phylogeny of extant Testudo s.l.) supports a taxonomic scheme in which the three extant subgenera of Testudo are represented in the fossil record. Testudo s.s. is retrieved as the sister taxon of Testudo (Agrionemys) + [Testudo (Paleotestudo) + Testudo (Chersine)]. The extinct Testudo (Paleotestudo) is therefore the sister taxon of the Testudo (Chersine) clade. The latter subgenus reveals as the most diverse clade of Testudo s.l. in the fossil record, with T. catalaunica + Testudo steinheimensis constituting a subclade distinct from that including Testudo hermanni The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 178: doi: /zoj ADDITIONAL KEYWORDS: Agrionemys Chersine fossil Testudines Testudinids Testudo steinheimensis. INTRODUCTION Originally erected by Linnaeus (1758) to include all chelonians known by then, the genus Testudo Linnaeus, 1758, has had a complex taxonomic history. It was subsequently applied in a more restricted usage for tortoises (e.g. Fitzinger, 1826; Boulenger, 1889; Siebenrock, 1909; Williams, 1952; Mertens & *Corresponding author. david.alba@icp.cat 312 tortoises Paleotestudo Taxonomy Wermuth, 1955), until Loveridge & Williams (1957) restricted it to the five extant Western Palaearctic (mostly circum-mediterranean) species: Testudo graeca Linnaeus, 1758 (type species); Testudo kleinmanni Lortet, 1883; Testudo marginata Schoepff, 1793; Testudo hermanni Gmelin, 1789; and Testudo horsfieldii Gray, Although the use of Testudo s.l. for these extant species and other extinct closely related taxa is still common nowadays (Corsini, B ohme & Joyce, 2014; Turtle Taxonomy Working Group, 2014), through the years several authors 1, MASSIMO DELFINO1,2, JOSEP M. ROBLES1,3 and DAVID M. ANGEL H. LUJAN 1 ALBA *

2 A REVISED PHYLOGENY OF EXTINCT TESTUDO 313 have proposed splitting Testudo into several genera, such as Agrionemys Khosatzky & Mlynarski, 1966; Protestudo Chkhikvadze, 1970; Protagrionemys Chkhikvadze, 2001; and Eurotestudo Lapparent de Broin et al., 2006a. There is consensus that Protestudo (type species Testudo bessarabica Riabinin, 1918) and Protagrionemys (type species Testudo turgaica Riabinin, 1926) are junior subjective synonyms of Agrionemys (Auffenberg, 1974; Lapparent de Broin, 2000, 2001; Per al a, 2002; Lapparent de Broin et al., 2006a,b,c; Fritz & Bininda-Emonds, 2007; Bertolero et al., 2011; Turtle Extinctions Working Group, 2015). In turn, Lapparent de Broin et al. (2006a) considered that the type species of both Chersine Merrem, 1820, and Medaestia Wussow, 1916, was T. graeca, so that a new genus name, Eurotestudo, was required for T. hermanni. However, Fritz & Bininda-Emonds (2007) showed that the type species of both Chersine and Medaestia is, in fact, T. hermanni, and so Eurotestudo is an objective junior synonym of Chersine (see also Fritz & Kraus, 2008). Currently, an alternative taxonomy for Testudo s.l., distinguishing three distinct genera, is also widely used (Turtle Taxonomy Working Group, 2014): Testudo s.s. for the clade of T. graeca; Chersine Merrem, 1820, for T. hermanni; and Agrionemys Khosatzky & Mlynarski, 1966, for T. horsfieldii. The distinction of three genera for the five extant species of Testudo s.l. is solidly grounded on their molecular phylogeny (Fritz & Bininda-Emonds, 2007). Unlike the previous analysis by Parham et al. (2006), which found Testudo s.l. to be paraphyletic relative to Indotestudo Lindholm, 1929, Fritz & Bininda-Emonds (2007) found that Testudo s.l. constitutes a monophyletic clade, in which two distinct subclades can be distinguished: on the one hand, T. hermanni + T. horsfieldii; and, on the other, (T. kleinmanni + T. marginata) + T. graeca; i.e. Chersine + Agrionemys and Testudo s.s. spp., respectively. Although the distinction of several genera seems rather unnecessary in the light of the species diversity of this clade (e.g. Loveridge & Williams, 1957; Fritz & Bininda-Emonds, 2007; Corsini et al., 2014), the above-mentioned genus-group names are potentially useful for formally distinguishing the various subclades of Testudo at the subgenus rank (Fritz & Bininda-Emonds, 2007), especially when the more numerous extinct species are also taken into account. Here we therefore recognize Chersine and Agrionemys as distinct subgenera to include those extinct species of Testudo that are more closely related to T. hermanni and T. horsfieldii, respectively. The genus-group name Chersus Wagler, 1830, is available for the subclade including T. marginata (its type species) and T. kleinmanni, having been previously employed by some authors either as the name of a distinct subgenus (Gmira, 1993, 1995) or clade (Parham et al., 2006; Vlachos & Tsoukala, 2016). However, we follow most recent authors (e.g. Turtle Taxonomy Working Group, 2014) in considering that T. marginata and T. kleinmanni belong to the same (sub)genus as T. graeca, so that Testudo (Chersus) becomes a junior subjective synonym of Testudo (Testudo). Unfortunately, morphology-based cladistic analyses incorporating extinct species of Testudo s.l. have yielded somewhat contradictory results compared to one another as well as relative to the molecular phylogeny. For example, concerning extant species, neither Lapparent de Broin et al. (2006b,c) nor Corsini et al. (2014) recovered the sister-taxon relationship between T. hermanni and T. horsfieldii attested by Fritz & Bininda-Emonds (2007), although Corsini et al. (2014) at least recovered each of the three extant lineages (forming a basal polytomy with the extinct Testudo antiqua Bronn, 1831). The most recent analysis by Vlachos & Tsoukala (2016: fig. S1A), when unconstrained by the molecular results of Fritz & Bininda-Emonds (2007), did not resolve the internal phylogeny of extant Testudo s.l., as the consensus tree of their unconstrained analysis found a large basal polytomy and failed to show the monophyly of Testudo with the exclusion of Indotestudo. The systematics of extinct Testudo s.l. is further complicated by the purported distinction of an extinct genus, Paleotestudo Lapparent de Broin, 2000, which branches well within Testudo s.l. (at the base of Chersine) according to Corsini et al. (2014), but from a more basal position according to Vlachos & Tsoukala (2016). Previous phylogenetic analyses did not include the extinct Testudo catalaunica Bataller, 1926, originally described based on material from the Valles-Penedes Basin in north-east Spain and included in Paleotestudo by some authors (Lapparent de Broin et al., 2006b,c; Perez-Garcıa & Murelaga, 2013). Based on a preliminary account of the new material described here and attributed to T. catalaunica, Lujan et al. (2014) noted that this species does not fit the diagnosis of Paleotestudo and argued that it should be retained in Testudo. Here we describe in detail and figure this new material and, on its basis, we provide an emended diagnosis of T. catalaunica. In order to assess the systematic position of this taxon amongst the members of the clade Testudo, we further provide a cladistic analysis of most of the extinct species of this genus, and compare our results with those previously obtained by Fritz & Bininda-Emonds (2007) in the case of extant Testudo, and those recently derived by Corsini et al. (2014) and Vlachos & Tsoukala (2016) regarding extinct species of Testudo. Based on our results, we provide a revised taxonomic scheme of both extant

3 314 A. H. LUJ AN ET AL. and extinct members of genus Testudo (Table 1) and re-evaluate the validity of Paleotestudo as a distinct taxon. MATERIAL AND METHODS STUDIED SAMPLES AND NOMENCLATURE The previously unpublished specimens of T. catalaunica newly reported in this paper, together with those previously described (including the type material), are listed in Table 2 and described in detail in the next section. All these specimens come from the Valles- Penedes Basin and constitute the basis for the emended diagnosis provided here for this species. This material was also employed to code T. catalaunica in a data matrix used to perform a cladistic analysis (see next subsection for further details). The comparative sample for the remaining taxa is specified in Appendix 1. Some of the extant taxa, particularly T. graeca, have a wide distribution range, with local populations being morphologically distinct and distinguished at the subspecies rank (Fritz et al., 2007, 2009); however, the comparative sample employed in our study includes specimens from Europe, Asia, and Africa and is thus representative of the species as a whole. With regard to extinct Testudo, additional species are nomenclaturally valid (Table 1), but the insufficient information available from the literature precluded their inclusion in the data matrix of the present study. Anatomical shell nomenclature used throughout the descriptions is based on Zangerl (1969). Given the arched shape of the carapace, it is worth specifying that, whereas the description of shell contour is in dorsal view, the description of plates and scutes shape is perpendicular to their surface (and, hence, not necessarily in truly anatomical dorsal view, but also in anterolateral, posterolateral, or lateral views, depending on the case). PHYLOGENETIC ANALYSIS A cladistic analysis based on maximum parsimony was performed with PAUP* 4.0 for Unix (Swofford, 2003), with the search command branch-and-bound, based on a taxon-character data matrix of 19 taxa and 37 characters (Table 3, Appendix 2). All characters were treated as unordered and were parsimony informative. Clade robusticity was assessed by means of bootstrap analysis ( replicates) and Bremer support indices. The following metrics were computed for the most parsimonious trees: consistency index (CI); retention index (RI); and rescaled consistency index (RC). Character polarity was determined using the outgroup method, with the genera Malacochersus Lindholm, 1929 and Indotestudo Table 1. Taxonomy of the genus Testudo adopted in this paper. Type species of subgenera are denoted with an asterisk, whereas extinct species are denoted with a dagger. Within each subgenus, species are ordered according to year of description. Species included in the cladistic analysis are in bold type, whereas taxa of dubious taxonomic validity (potentially synonyms of other taxa or nomina dubia) are denoted with a question mark. For further details on the taxonomic validity of the listed species, see Danilov (2005), Lapparent de Broin et al. (2006c), and Fritz et al. (2009) Genus Testudo Linnaeus, 1758 Subgenus Testudo Linnaeus, 1758 Testudo (Testudo) graeca Linnaeus, 1758* Testudo (Testudo) marginata Schoepff, 1793 Testudo (Testudo) marmorum Gaudry, 1862 Testudo (Testudo) kleinmanni Lortet, 1883 Testudo (Testudo) cernovi Khosatzky, 1948 Testudo (Testudo) kucurganica Khosatzky, 1948 Testudo (Testudo) semenensis Bergounioux, 1955 Testudo (Testudo) burtschaki Chkhikvadze, 1975 Testudo (Testudo) eldarica Khosatzky & Alekperov in Alekperov, 1978 Testudo (Testudo) kenitrensis Gmira, 1993 Testudo (Testudo) oughlamensis Gmira et al.,2013 Testudo (Testudo) brevitesta Vlachos & Tsoukala, 2016 Subgenus Agrionemys Khosatzky & Mlynarski, 1966 Testudo (Agrionemys) horsfieldii Gray, 1844* Testudo (Agrionemys) bessarabica Riabinin, 1918 Testudo (Agrionemys) turgaica Riabinin, 1926 Testudo (cf. Agrionemys) bosporica Riabinin, 1945 Testudo (Agrionemys) paraskivi Kusnetsov, 1972 Subgenus Chersine Merrem, 1820 Testudo (Chersine) hermanni Gmelin, 1789* Testudo (Chersine) antiqua Bronn, 1831 Testudo (Chersine)?escheri Pictet & Humbert, 1856 Testudo (Chersine)?mellingi (Peters, 1868) Testudo (Chersine) pyrenaica Deperet & Donnezan, 1890 Testudo (Chersine) amiatae Pantanelli, 1893 Testudo (Chersine)?amberiacensis Deperet, 1894 Testudo (Chersine) kalkburgensis Toula, 1896 Testudo (Chersine) promarginata Reinach, 1900 Testudo (Chersine) lunellensis Almera & Bofill, 1903 Testudo (Chersine) catalaunica Bataller, 1926 Testudo (Chersine) steinheimensis Staesche, 1931 Testudo (Chersine) csakvarensis Szalai, 1934 Testudo (Chersine)?szalaii Młynarski, 1955 Testudo (Chersine) burgenlandica Bachmayer & Młynarski, 1983 Testudo (Chersine) opisthoklitea Schleich, 1981 Testudo (Chersine) rectogularis Schleich, 1981 Subgenus Paleotestudo Lapparent de Broin, 2000 Testudo (Paleotestudo) canetoniana Lartet, 1851 * Testudo (Paleotestudo) angustihyoplastralis (Gemel, 2002)

4 A REVISED PHYLOGENY OF EXTINCT TESTUDO 315 Table 2. List of fossil specimens of Testudo catalaunica included in this study. This work denotes those specimens that are first described in this paper Catalogue no. Locality Age (Ma) Anatomical description Figure(s) Description MGSB25324a SQ Almost complete shell 2A F, 3A F Bataller (1926), Bergounioux (1938) Without number SQ Partial right scapula 2G, 3G Bataller (1926) (lost) ( Specimen A ) Without number SQ Left peripheral 7 2H, I, 3H, I Bataller (1926) (lost) ( Specimen B ) MGSB31642 SQ Right anterior shell 2J L, 3J L, 10A Bataller (1926) portion MGM1561 SQ Partial shell 2M V, 3M V Bergounioux (1958) MGSB25324b SQ Partial plastron 2W, X, 3W, X Perez-Garcıa & Murelaga (2013) MGM236M SQ Partial plastron 2Y D, 3Y D Perez-Garcıa & Murelaga (2013) IPS30074 ACM/C1-ET Partial shell 4A, B, 5A, B This work IPS30073 ACM/C1-E Partial carapace 4C, 5C, 10B This work IPS30052 ACM/C1-E Posterior portion of 4D, 5D This work carapace IPS30061 ACM/VIE-E Partial plastron 4E, 5E This work IPS4417 ACM/C1-E Partial shell 4F, G, 5F, G This work IPS4423 ACM/C1-E Posterior portion of 4H, I, 5H, I This work carapace IPS4411 ACM/C1-E Complete shell 4J N, 5J N This work IPS4420 ACM/C1-E Complete shell 4O S, 5O S This work IPS4415 ACM/C1-E Anterior shell portion 4T, U, 5T, U This work IPS4419 ACM/C1-E Complete shell 4V Y, 5V Y This work IPS32957 ACM/C1-E Almost complete shell 4Z, A, 5Z, A This work MGB V61-3 SQ Anterior plastral lobe 6A, B, 7A, B This work IPS2073 SQ Partial shell 6C, D, 7C, D This work IPS33180 CB Nuchal and left 6E, F, 7E, F This work peripherals 1 3 IPS36396a CB Almost complete shell 6G N, 7G N This work IPS36396b CB Proximal fragment of 8A, B This work right femur IPS30905 CM Proximal fragment of 8C, D This work left femur IPS87219 CM Proximal fragment of 8E, F This work left femur IPS16441 HPS Proximal fragment of left femur 8G, H This work ACM, Abocador de Can Mata; C1, Cella 1; CB, Castell de Barbera; CM, Can Mata indeterminate; HPS, Hostalets de Pierola Superior; SQ, Sant Quirze; VIE, Vial Intern d Explotacio. See main text for abbreviations. Lindholm, 1929 being employed as such. Ingroup taxa include the five extant species of Testudo as well as 12 extinct species of the same genus, including Testudo canetotiana Lartet, 1851 (which is the type species of Paleotestudo). The data matrix employed was partially based on previously published morphological descriptions (Loveridge & Williams, 1957) and cladistic character statements (Lapparent de Broin et al., 2006b,c; Corsini et al., 2014; Vlachos & Tsoukala, 2016), and partially coded anew by the authors. In the descriptive and phylogenetic sections of the paper, character and character states have been reported within brackets by stating the character number preceded by # and afterwards the character state number preceded by a colon (e.g. [#1:0] is character 1 state 0). The list of literature and comparative material employed to code the data matrix is reported in Appendix 1.

5 316 A. H. LUJ AN ET AL. Table 3. Data matrix employed in the cladistic analysis performed in this paper for the genus Testudo. Malacochersus and Indotestudo are included as outgroups. Extinct species are denoted by a dagger and type species of each subgenus are in bold type. Missing data are represented by a question mark. A equals (0,1). Taxon/character Malacochersus tornieri ? Indotestudo elongata Testudo (Testudo) graeca A Testudo (Testudo) marginata A 1 Testudo (Testudo) marmorum 1 1? 1 0 0? ? Testudo (Testudo) kleinmanni A Testudo (Testudo) kenitrensis 1 1? ? Testudo (Testudo) oughlamensis 1 1? 1 0 0? A Testudo (Testudo) brevitesta 1???? 0? 1? 0??? 1? 0?? 1? 0? ?? ??? Testudo (Agrionemys) horsfieldii Testudo (Agrionemys) bessarabica 1?? 0 0 1? 0 0? ? 0? 0 1? Testudo (Paleotestudo) ? canetoniana Testudo (Chersine) steinheimensis 1 1? 0 1 0? A A A Testudo (Chersine) catalaunica ? A 1 A A Testudo (Chersine) antiqua 1 1? 0 1 0? A A 1 A Testudo (Chersine) burgenlandica ? Testudo (Chersine) pyrenaica ? ? Testudo (Chersine) lunellensis ? A Testudo (Chersine) hermanni A

6 A REVISED PHYLOGENY OF EXTINCT TESTUDO 317 ABBREVIATIONS Institutional abbreviations BSP, Bayerische Staatssammlung f ur Pal aontologie und Geologie, Munich, Germany; UCBL1, Collections de Geologie de l Universite Claude Bernard Lyon 1, France; MTM, Magyar Termeszettudomanyi Muzeum (Hungarian Natural History Museum), Budapest, Hungary; ICP, Institut Catala de Paleontologia Miquel Crusafont, Universitat Autonoma de Barcelona, Spain; IPS, acronym of the ICP collections; MGB, Museu de Geologia, Museu de Ciencies Naturals de Barcelona, Spain; MGB V, acronym of specimens from Villalta s collection currently housed at the MGB; MGSB, Museu de Geologia del Seminari de Barcelona, Spain; BMNH, Natural History Museum [formerly British Museum (Natural History)], London, UK; MHNT, Museum d Histoire Naturelle de Toulouse, France; MNHN, Museum national d Histoire naturelle, Paris, France; MNHN.F, Paleontological Collections of the MNHN; MNHN.ZA, Comparative Anatomy Collections of the MNHN; MNHN.RA, Zoology Collections of Reptiles and Amphibians of the MNHN; MNHN.REP, Reptile Collection currently of the MNHN.F; NMW, Naturhistorisches Museum Wien, Austria; SMF, Senckenberg Museum Frankfurt, Germany; SMNS, Staatliches Museum f ur Naturkunde Stuttgart, Germany. Locality abbreviations ACM, Abocador de Can Mata (els Hostalets de Pierola); C1, Cella 1 (ACM sector); CB, Castell de Barbera (Barbera del Valles); CM, Can Mata indeterminate (els Hostalets de Pierola); HPS, Hostalets de Pierola Superior (els Hostalets de Pierola); SQ, Sant Quirze (Sant Quirze del Valles); VIE, Vial Intern d Explotacio (ACM sector). AGE AND GEOLOGICAL BACKGROUND Here we describe all the available fossil remains of T. catalaunica from the Valles-Penedes Basin (northeast Iberian Peninsula; Fig. 1), a NNE SSW-orientated half-graben limited by the Littoral and Pre-littoral Catalan Coastal Ranges that was generated by the rifting of the north-west Mediterranean region during the Neogene (Cabrera et al., 1991; Roca & Guimera, 1992). Except for some early and middle Miocene shallow marine and transitional sequences, most of the basin infill consists of marginal alluvial fan sediments with a rich fossil record of early, late middle, and late Miocene terrestrial vertebrates (Agustı, Cabrera & Moya-Sola, 1985; Casanovas- Vilar et al., 2011a). The following correlations with Mammal Neogene (MN) units are available for the localities that have yielded remains of T. catalaunica (Moya-Sola et al., 2009; Casanovas-Vilar et al., 2011b; Alba et al., 2011: ACM (MN7+8, late Aragonian, middle to late Miocene), CB (MN9, early Vallesian, late Miocene), CM (MN7+8 or MN9, late Aragonian to early Vallesian, late Miocene), HPS (MN9, early Vallesian, late Miocene), and SQ (MN7+8, late Aragonian, middle Miocene). Both CM and HPS, in the area of els Hostalets de Pierola, are not palaeontological sites, but instead loosely defined fossiliferous areas and so, owing to lack of precise stratigraphical control, it is not possible to determine their age more accurately (Agustı et al., 1985; Alba et al., 2006, 2011). This contrasts with the situation for the various localities at ACM (Alba et al., 2006, 2007, 2009, 2010, 2011), within the same area, in which magnetostratigraphical control (Moya-Sola et al., 2009; Casanovas-Vilar et al., 2011b) allows a much more accurate dating (Alba et al., 2011; D. M. Alba, unpublished data; see Table 2 for further details). Regarding CB, it was traditionally correlated to MN7+8 and considered roughly coeval with SQ (e.g. Casanovas-Vilar et al., 2011b), but based on currently available evidence it seems preferable to correlate it to MN9 (see discussion in Alba & Moya- Sola, 2012). SYSTEMATIC PALAEONTOLOGY ORDER TESTUDINES LINNAEUS, 1758 SUBORDER CRYPTODIRA COPE, 1868 SUPERFAMILY TESTUDINOIDEA BATSCH, 1788 FAMILY TESTUDINIDAE BATSCH, 1788 GENUS TESTUDO LINNNAEUS, 1758 SUBGENUS CHERSINE MERREM, 1820 TESTUDO (CHERSINE) CATALAUNICA BATALLER, 1926 (FIGS 2 9) Main synonyms 1926 Testudo catalaunica Bataller: 11 28, figs 1 6 (original description) Testudo catalaunica Bataller, 1926: Bergounioux: 283, figs Testudo catalaunica var. irregularis Bergounioux: 167, figs 11 12, pl. 29a, b (original description). 2006b Paleotestudo catalaunica (Bataller, 1926): Lapparent de Broin: c Paleotestudo catalaunica (Bataller, 1926): Lapparent de Broin: Paleotestudo catalaunica (Bataller, 1926): Perez-Garcıa & Murelaga: 14. Lectotype MGSB25324a, almost complete shell (Figs 2A F, 3A F; see also Bataller, 1926: pl , 3.3).

7 318 A. H. LUJ AN ET AL. Figure 1. Schematic geological map of the Valles-Penedes Basin, showing the location of sites that have yielded remains of Testudo (Chersine) catalaunica. ACM, Abocador de Can Mata; CB, Castell de Barbera; CM, Can Mata indeterminate; HPS, Hostalets de Pierola Superior; SQ, Sant Quirze. Paralectotypes MGSB31642, right anterior shell portion (Figs 2J L, 3J L; see also Bataller, 1926: pl. 4.3). Specimen A, partial right scapula without catalogue number (Bataller, 1926: pl. 1.1; see also our Figs 2G, 3G) and currently lost. Specimen B, left peripheral 7 without catalogue number (Figs 2H, I, 3H, I; see also Bataller, 1926: pl. 4.1, 4.2) and currently lost. Emended diagnosis Medium-sized species of Testudo (c cm adult shell length) with a slightly elongated, elliptical shell lacking hypo-xiphiplastral hinge. Shell bridge not elevated. Vertebral scutes 2 4 approximately as wide as long. Vertebral scute 5 wider than the remaining vertebrals. Anterior and posterior borders of marginal scutes 5 parallel to those of peripheral plates 5 6. Anterior plastral lobe with a markedly trapezoidal and nonprotruding anterior contour; dorsal epiplastral lip wider than long and little overhanging, without a conspicuous gular pocket. Unreduced neural series; neural plate 1 elongated with a marked posterior constriction (narrowing behind the posterior sulcus of vertebral 1), and neural plates 2 8 no more than twice wider than long. Lack of contact between peripheral plate 6 and hyoplastron. Cervical scute present and well developed both dorsally and viscerally. Suprapygal plates 1 2 unfused. Supracaudal scute undivided. Pygal plate straight (noncurved) and trapezoidal without anterolateral borders. Gular scutes without ventral relief. Femoral (greater and lesser) trochanters medially well developed (with the intertrochanteric fossa being invisible in ventral view). Differential diagnosis Testudo (Chersine) catalaunica autapomorphically differs from other species of Testudo s.l. by the shape of neural plate 1, which markedly narrows behind the posterior sulcus of vertebral 1. It differs from species of Testudo s.s., Testudo (Agrionemys), and Testudo (Paleotestudo) in the anterior and posterior borders of marginal scute 5 being parallel to those of peripheral plates 5 6, and the trapezoidal (instead of rounded) anterior plastral lobe. It further differs from species of the two former subgenera but not Testudo (Paleotestudo) in the medially well-developed femoral trochanters, and the vertebral scutes 2 4 and neural plates 2 8 being no more than twice wider than long; from species of Testudo s.s. in the lack of a hypo-xiphiplastral hinge, vertebral scute 5 being wider than vertebrals 1 4, and the little overhanging dorsal epiplastral lip without a conspicuous gular pocket; from species of Testudo (Agrionemys) by the not elevated shell bridge, the slightly elongated (instead of rounded) shell contour, and the dorsally better developed cervical scute; and from Testudo (Paleotestudo) in the lack of a curved pygal plate as well as posteroventrally directed peripheral

8 A REVISED PHYLOGENY OF EXTINCT TESTUDO 319 Figure 2. Previously published shell and postcranial remains of Testudo (Chersine) catalaunica from Sant Quirze. (A F) almost complete shell MGSB25324a (lectotype) in dorsal (A), ventral (B), and right lateral (C) views, together with details of the nuchal in dorsal view (D), the pygal in external view (E), and the entoplastron in ventral view (F). (G) partial right scapula Specimen A (paralectotype) in posterior view (reproduced from Bataller, 1926). (H, I) partial left peripheral 7 Specimen B (paralectotype), in external (H) and internal (I) views (reproduced from Bataller, 1926). J L, right anterior shell portion MGSB31642 (paralectotype) in dorsal (J), ventral (K), and right lateral (L) views. M V, partial shell MGM1561 (holotype of Testudo catalaunica irregularis) in dorsal (M), ventral (N), and right lateral (O) views, together with details of the anterior plastral lobe in left lateral view (P), the nuchal in dorsal (Q) and visceral (R) views, the pygal in external (S) and internal (T) views, and the entoplastron in ventral (U) and visceral (V) views. (W X), partial plastron MGSB25324b in ventral view (W), together with details of the entoplastron in ventral view (X). Y D, partial plastron MGM236M, in visceral (Y) and ventral (Z) views, together with details of the anterior plastral lobe in left lateral (A ) and anterior (B ) views, and the entoplastron in ventral (C ) and dorsal (D ) views. plates 8 11 and pygal plate (which are autapomorphic of the latter subgenus). Compared with other species of Testudo (Chersine), it differs from all except T. steinheimensis in the trapezoidal anterior plastral lobe; it further differs from T. antiqua in the nonpointed anterior plastral lobe and in the unreduced neural plate series; from Testudo burgenlandica, Testudo lunellensis, and T. hermanni in the absence of extensive contact between peripheral 6 and hyoplastron; from T. burgenlandica and Testudo pyrenaica in the better developed cervical scute both dorsally and viscerally; from T. pyrenaica and T. hermanni in the gular scutes without ventral relief; from T. lunellensis and T. hermanni in the unfused suprapygal plates 1 2, the undivided supracaudal scute, and the trapezoid pygal plate (without anterolateral borders); from T. hermanni in the unreduced neural plate series; and from T. (C.) steinheimensis in the lack of a protruding anterior plastral lobe (which is autapomorphic of the latter species), the unreduced neural plate series, and the slightly more overhanging dorsal epiplastral lip with (poorly developed) gular pocket. Type locality Sant Quirze (= Trinxera del Ferrocarril; MN7+8, middle Miocene). Distribution Late Aragonian to early Vallesian (MN7+8 MN9) of the Valles-Penedes Basin (north-east Iberian Peninsula), including the localities of SQ (MN7+8), CB (MN9), HPS (MN9), CM (MN7+8), and various localities of the ACM series (MN7+8). PRESERVATION AND MEASUREMENTS The preservation and measurements of the specimens of T. catalaunica included in this study are provided here on an individual basis. In turn, a joint

9 320 A. H. LUJ AN ET AL. Figure 3. Schematic drawings corresponding to the fossil remains depicted in Figure 2. Thick lines correspond to scute sulci, dashed lines denote plate sutures, and oblique lines denote missing portions or sediment encrusting the fossils. morphological description is subsequently provided in the next subsection. Sant Quirze (type locality) MGSB25324a (lectotype): Nearly complete shell (Figs 2A F, 3A F; see also Bataller, 1926: pl , 3.3), 17.5 cm in length and 13.5 cm in width. The interior of the shell and most of the cracks on its outer surface have been sealed with plaster. The left peripherals 1 4 are missing and most of the distal portions of costals 1 3 are eroded. The posterior portion of the carapace is severely distorted, and both suprapygals are missing. The plastron is generally well preserved, except for the very deteriorated left hyo-hypoplastron, the incomplete left xiphiplastron, the broken anterior margin of the left epiplastron, and the missing right epiplastron. There are no signs of distortion, although some plates (particularly on the left hypo-xiphiplastron) are slightly dislocated owing to cracks. Specimen A (paralectotype): Partial right scapula without catalogue number (Figs 2G, 3G; see also Bataller, 1926: pl. 1.1) that is currently lost. The completely preserved scapular process was 3.6 cm in height and the preserved portion of the acromion 1.7 cm in length (Bataller, 1926). Specimen B (paralectotype): Complete left peripheral 7 without catalogue number (Figs 2H, I, 3H, I; see also Bataller, 1926: pl. 4.1, 4.2) and currently lost. Based on Bataller s (1926) pictures, this specimen was 2.5 cm long and 5.0 cm wide. All of the intermarginal sulcus was discernible in both external and internal views. MGSB31642 (paralectotype): Right anterior carapace portion, including neurals 1 4, right costals 1 4, and right peripheral plates 1 7 (Bataller, 1926: pl. 4.3; see also our Figs 2J L, 3J L). The preserved portion of the carapace, 17.3 cm long and 10.7 cm wide, is highly fractured and part of right costal 4 and peripheral 7 have been partially reconstructed with plaster. Missing from the carapace is the nuchal plate and most of neural plate 4. The sutures between peripheral plates 4 5 and sulci between marginal scutes 5 6 are not completely visible. The preserved plastron fragment mainly consists of the right hyoplastron, which is slightly rotated toward to the left relative to its original anatomical position. MGM1561 (holotype of Testudo catalaunica irregularis): Partial shell missing all the neural and costal plates (Figs 2M V, 3M V; see also Bergounioux, 1958: figs 11, 12, pl. 29). The preserved anterior portion of the carapace, which is distorted towards the left, measures 18.1 cm in length and 12.5 cm in width. Both right and left peripherals 3 6 are damaged. The plastron is nearly complete, only missing some fragments from the right hypoplastron

10 A REVISED PHYLOGENY OF EXTINCT TESTUDO 321 Figure 4. New shell remains of Testudo (Chersine) catalaunica from Abocador de Can Mata. (A, B) partial shell IPS30074 in dorsal (A) and ventral (B) views. (C) partial carapace IPS30073 in dorsal view. (D) posterior portion of carapace IPS30052 in posterior view. (E) partial plastron IPS30061 in ventral view. (F, G) partial shell IPS4417 in dorsal (F) and ventral (G) views. (H, I) posterior portion of carapace IPS4423 in dorsal (H) and ventral (I) views. J N, complete shell IPS4411 in dorsal (J), ventral (K), left lateral (L), right lateral (M), and posterior (N) views. O S, complete shell IPS4420 in dorsal (O), ventral (P), left lateral (Q), right lateral (R), and posterior (S) views. (T, U) anterior shell portion IPS4415 in dorsal (T) and ventral (U) views. V Y, complete shell IPS4419 in dorsal (V), ventral (W), left lateral (X), and posterior (Y) views. (Z, A ), almost complete shell IPS32957 in dorsal (Z) and ventral (A ) views. and xiphiplastron. Most of the scute sulci are well preserved, although some sutures are sealed with plaster (as in the hyo-hypoplastra contact). MGSB25324b: Nearly complete but highly fractured plastron (Figs 2W, X, 3W, X; see also Perez-Garcıa & Murelaga, 2013), 18.1 cm long and 11.7 cm wide. The right xiphiplastron is missing, the lateral portion of the right hypoplastron is eroded, and the posterior portion of the right hyoplastron is severely damaged and reconstructed with mortar. The specimen is mounted on a plaster support that hides the visceral surface. MGM236M: Nearly complete but poorly preserved plastron (Figs 2Y D, 3Y D ; see also Perez-Garcıa & Murelaga, 2013), 18.7 cm long and 10.7 cm wide. The posterior tip of the left xiphiplastron is broken, the lateral edges of both hypoplastra are very damaged, the right axillary buttress is completely eroded, and both inguinal buttresses are considerably damaged. Both the ventral and visceral sides are visible, and in spite of the deficient restoration with plaster, the morphology of sutures and scute sulci can be adequately ascertained, except that it is not possible to observe that the humeropectoral, pectoroabdominal, and abdominofemoral sulci join medially.

11 322 A. H. LUJ AN ET AL. Figure 5. Schematic drawings corresponding to the fossil remains depicted in Figure 4. Thick lines correspond to scute sulci, dashed lines denote plate sutures, and oblique lines denote missing portions or sediment encrusting the fossils. MGB V61-3: Anterior-most portion of a plastral lobe (Figs 6A, B, 7A, B), 8.1 cm long and 10.8 cm wide. The anterior and lateral margins of the right epiplastron are eroded, and only the anterior portions of the hyoplastra are preserved. Both the ventral and visceral surfaces are to a large extent covered by a grey concretion that partially hides the sutures and sulci, as well as some plates. IPS2073: Poorly preserved and little informative partial shell, 11.2 cm long and 8.9 cm wide (Figs 6C, D, 7C, D). Dorsally it only preserves the central portion of the carapace, and the sutures are not visible because the plate elements are heavily fused. Most of the plastron is lacking, although the left hypoplastron is relatively well preserved. The anterior and posterior openings of the shell are completely sealed and reconstructed with plaster. Els Hostalets de Pierola (ACM, CM and HPS) IPS30074: Dorsoventrally compressed partial shell (Figs 4A, B, 5A, B), 18.3 cm long and 12.5 cm wide, that preserves a small central portion of the carapace, most of the peripheral plates, and a large portion of the plastron (except for the xiphiplastra and the left hyoplastron), although the hypoplastra are very fractured. Most of the costals, neurals, and suprapygal 1 are missing, and the left peripheral 7 is artefactually protruding from the shell contour. IPS30073: Left portion of carapace (Figs 4C, 5C), 23.2 cm long and 14.5 cm wide, which, albeit very distorted and slightly fractured, displays the plate elements close to its anatomical position thanks to the matrix infill. As the plates are slightly disarticulated, all the sutures are clearly visible. The anterior portion of the carapace is partly covered by a hard concretion.

12 A REVISED PHYLOGENY OF EXTINCT TESTUDO 323 Figure 6. New fossil shell remains of Testudo (Chersine) catalaunica from Sant Quirze (SQ) and Castell de Barbera (CB). (A, B) anterior plastral lobe MGB V61-3 from SQ in visceral (A) and ventral (B) views. (C, D) partial shell IPS2073 from SQ in dorsal (C) and ventral (D) views. (E, F) nuchal and left peripheral 1 3 IPS33180 from CB in dorsal (E) and visceral (F) views. (G N) almost complete shell IPS36396a from CB in dorsal (G) and ventral (H) views, together with details of the nuchal in dorsal (I) and visceral (J) views, the pygal in external (K) and internal (L) views, and the entoplastron in ventral (M) and visceral (N) views. Figure 7. Schematic drawings corresponding to the fossil remains depicted in Figure 6. Thick lines correspond to scute sulci, dashed lines denote plate sutures, and oblique lines denote missing portions or sediment encrusting the fossils. IPS30052: Small posterior fragment of carapace (Figs 4D, 5D), 2.5 cm long and 5.3 cm wide, which only preserves vertebral 4, an anterior portion of vertebral 5, and right and left pleurals 3 4. The preserved plate elements are partially damaged and their surface altered such that only some sulci are partly visible. IPS30061: Anterior fragment of plastron with some misplaced costal plates attached to its posterior portion (Figs 4E, 5E). A variably thick concretion, together with sediment matrix, masks most of the external and internal surfaces, except for the posterior-most preserved portion of the plastron and the dorsal epiplastral pad (which has been prepared and is clearly visible). The posterolateral margin of the right epiplastron is missing, and the entoplastron and left epiplastron are quite chipped.

13 324 A. H. LUJ AN ET AL. Figure 8. New fossil postcranial remains of Testudo (Chersine) catalaunica from Castell de Barbera (CB), Hostalets de Pierola Superior (els Hostalets de Pierola, HPS) and Can Mata indeterminate (CM). (A, B) proximal fragment of right femur IPS36396b from CB in dorsal (A) and ventral (B) views. (C, D) proximal fragment of left femur IPS30905 from CM in dorsal (C) and ventral (D) views. (E, F) proximal fragment of left femur IPS87219 from CM in dorsal (E) and ventral (F) views. (G, H) proximal fragment of left femur IPS16441 from HPS in dorsal (G) and ventral (H) views. IPS4417: Partial shell, 13.0 cm in length and 11.1 cm in width, preserving most of the carapace but only the posterior half of the plastron (Figs 4F, G, 5F, G). The anterior and posterior edges of the carapace are seriously damaged, and on the right it further displays a hole mostly filled by matrix. The pygal and the left posterior peripheral plates are partially eroded, and most of the nuchal plate is missing. The epiplastra, hyoplastra, and entoplastron are missing, but the preserved posterior portion of the plastron is relatively well preserved. IPS4423: Partial shell, 5.8 cm long and 9.1 cm wide, preserving the central portion of the carapace and most of the posterior portion of the hypoplastra (Figs 4H, I, 5H, I). The visceral surfaces are not visible owing to the matrix infill. The preserved right lateral portion of the carapace is poorly preserved and so only neurals 3 4 and the left costals 4 5 can be recognized. IPS4411: Nearly complete shell, 12.9 cm long and 10.7 cm wide, preserving most of the carapace and the whole plastron (Figs 4J N, 5J N). The right lateral margin of the carapace is slightly cracked (so that the sulci are only partly visible) and the posterior portion of the carapace is severely crushed (so that the posterior profile is distorted). Owing to dorsoventral compression, the plastron is slightly misplaced inside the carapace. Exceptionally amongst ACM specimens, all the sutures of the neural series are clearly visible on the dorsal surface of the carapace. IPS4420: Almost complete shell, 12.5 cm long and 11.5 cm wide (Figs 4O S, 5O S). Only the right lateral and posterior portions of the carapace are poorly preserved, with the right peripherals 4 and 6 being almost eroded away. There is also a rectangular hole at about the middle of the carapace caused during the excavation of the specimen in the field. The plastron is complete except for the distal edges of the right hyo-hypoplastron. The neural sutures are only visible for neural 1, as the remaining neurals are heavily fused. IPS32957: Partial shell, 8.2 cm long and 6.1 cm wide, preserving the posterior portion of the carapace and most of the posterior portion of the plastron (Figs 4Z, A, 5Z, A ). Neurals 1 3, suprapygal 2, the pygal, and all the posterior peripheral plates are missing from the carapace, whereas most of the proximal portions of the left and right costals 1 4 and of left peripherals 4 7 are very cracked and chipped. Only the anterior edge of the epiplastra and xiphiplastra are missing from the plastron. IPS4419: Nearly complete and slightly distorted (laterally compressed) shell, 10.5 cm long and 7.5 cm wide (Figs 4V Y, 5V Y). The anterior edge of the carapace is severely damaged, most of the right

14 A REVISED PHYLOGENY OF EXTINCT TESTUDO 325 Figure 9. Reconstruction of the shell of Testudo (Chersine) catalaunica based on the specimens from the Valles- Penedes Basin described in this paper. (A) dorsal and (B) ventral views. Thick lines correspond to scute sulci, whereas dotted lines denote plate sutures. Ab, abdominal scute; An, anal scute; Ce, cervical scute; co, costal plate; ent, entoplastron plate; epi, epiplastron plate; Fe, femoral scute; Gu, gular scute; Hu, humeral scute; hyo, hyoplastron plate; hyp, hypoplastron plate; Ma, marginal scute; ne, neural plate; nu, nuchal plate; Pe, pectoral scute; per, peripheral plate; Pl, pleural scute; py, pygal plate; sp, suprapygal; Ve, vertebral scute; xi, xiphiplastron plate. peripheral plates are eroded, and several dorsal plates (especially neurals 2 and 6, left costal 4 and right costals 2 3) are poorly preserved. The plastron is complete except for the distal portion of both hyohypoplastra. IPS4415: Partial shell, 5.6 cm long and 6.7 cm wide, lacking the posterior portion of both carapace and plastron (Figs 4T, U, 5T, U). The nuchal and the left anterior peripheral plates are covered by a concretion that hinders discernment of the sutures and the relationships between the marginal scutes. The visceral surface of the carapace is not visible because it is covered by matrix. The hypoplastra, xiphiplastral, and the anterior edge of the epiplastra are missing. IPS30905: Proximal fragment of a left femur (Fig. 8C, D), 1.51 cm long and 0.83 cm wide. It preserves most the proximal portion of the shaft, the head, the greater and lesser trochanters, and the intertrochanteric fossa. The ventral surface of the preserved bone is partially covered by matrix. IPS87219: Proximal fragment of a left femur (Fig. 8E, F), 1.25 cm long and 1.33 cm wide. It only preserves the femoral head, the two trochanters, and the intertrochanteric fossa, which is completely filled by sediment. IPS16441: Partial left femur (Fig. 8G, H), 2.25 cm long and 1.73 cm wide, which preserves the femoral head, the two trochanters, the intertrochanteric fossa, and a larger portion of the shaft than the two preceding specimens. Castell de Barbera IPS33180: Associated but disarticulated nuchal plate and left peripheral plates 1 3 (Figs 6E, F, 7E, F). The posterior edge of the nuchal plate is broken

15 326 A. H. LUJ AN ET AL. and its posteroventral surface is partly covered by a concretion, but otherwise the plates are well preserved and the scute sulci are visible. IPS36396: IPS36396a is a nearly complete, but somewhat distorted shell, 1.65 cm long and 1.36 cm wide, preserving the whole plastron and most of the carapace (Figs 6G N, 7G N). The pygal and most of the posterior peripheral plates are misplaced to the left, so that the relationships between the costal scutes and the peripheral plates cannot be confidently assessed. The right costals 2 5 are also clearly displaced, overlapping the neural series. In contrast, the remaining fractures do not considerably dislocate the rest of the plates. IPS36396b is the only associated postcranial element, consisting of a proximal fragment of right femur, 0.91 cm long and 0.63 cm wide (Fig. 8A, B); the distal portion of the shaft is missing, and the intertrochanteric fossa is partially covered by sediment. DESCRIPTION Testudo (Chersine) catalaunica is a medium-sized tortoise with a slightly elongated shell (reaching cm in length and cm in width) that displays an elliptical dorsal profile [#15:1]. Nine specimens (MGSB25324a, MGM1561, IPS30074, IPS4417, IPS4411, IPS4420, IPS32957, IPS4419, and IPS36396a) are sufficiently preserved to show that the carapace as a whole (not the shell bridge) is elevated and well arched, with a height/width ratio of [#14:1]. In dorsal view, the shell does not widen behind peripherals 7 [#22:0]. Like in other species of Testudo, the carapace displays no sculpturing; merely, scute growth lines can be discerned in IPS36396a (Fig. 6G) thanks to its good state of preservation (this is unusual in fossil specimens). Carapace The nuchal plate is roughly hexagonal [#29:0] and wider than long [#37:0]. Its posterior edge (which contacts neural 1) is shorter than the anterior one, which is slightly pointed anteriorly but not markedly notched. Its posterolateral margins are generally quite straight, except in IPS36396a and IPS30066, where they are well sinuous (Figs 4C, 5C, 6G, 7G). On the visceral surface of the nuchal there is a moderately developed transverse thickening that corresponds to the edge of the marginal overlap. The cervical scute is well developed both dorsally and viscerally [#18:0] (Figs 6G, 7G); it is longer than wide and its length constitutes 26 38% of the nuchal plate length. In some specimens (MGSB25324a, IPS33180, and IPS36396a), the cervical lateral sulci are parallel to each other, such that its anterior and posterior borders are approximately of the same width; in other specimens (MGM1561, IPS4415, IPS4419, and IPS4423), the anterior edge is somewhat narrower than the posterior one. The neural plate series count can be determined in several specimens (MGSB25324a, IPS4411, and IPS4419), all of which display eight plates [#23:0]. These are no more than twice as wide as long [#20:1], and vary in shape from octagonal (neurals 2 and 4) to subsquare (neurals 3 and 5) to hexagonal (neurals 6 to 8). Irregularities in plate shape frequently occur in the posterior portion of the neural series, especially in neural 8, which is hexagonal but whose shorter lateral edges may be variably situated either posteriorly (IPS4419 and IPS4420: Fig. 5P, Q) or anteriorly (IPS4411 and IPS4417: Fig. 5F, N). Neural plates 1, 3, 5, and 8 are encroached by the intervertebral sulci. Neural plate 1 is subrectangular, longer than wide, and longer than the remaining neural plates (Figs 3A, J, 5C, F, O, V, 7H). It is transversely split into two (anterior and posterior) portions by the posterior sulcus of vertebral 1. The anterior portion of neural 1 is longer and wider than the posterior one and displays slightly convex lateral margins that slightly diverge posteriorly, whereas the lateral margins of the posterior portion are quite straight. As a result, neural 1 is markedly constricted posteriorly, i.e. its posterior portion abruptly becomes clearly narrower than its anterior portion at about the level of the posterior sulcus of vertebral 1, instead of progressively narrowing or widening as in other taxa (see also Figs 9A, 10). The morphology of the suprapygal plates is preserved in five specimens (MGM1561, IPS4411, IPS4417, IPS4420, and IPS32957). There are two suprapygal plates [#19:1], with suprapygal 1 embracing suprapygal 2, which together constitute an isosceles trapezoid [#1:1]. The pygal plate is trapezoidal, with quite straight lateral margins [#24:0]. In undistorted specimens, in lateral view this plate is posteroventrally directed [#28:0], like peripherals 8 11 [#8:0] (Figs 4F, J, L, Q, 6G). The external surface of the pygal is flat or slightly convex, whereas its internal surface is moderately concave. Marginal scutes 12 appear to be completely fused in all the preserved specimens, so that the supracaudal scute is not divided by a sagittal groove (Figs 2E, S, 3E, S, 4N, S, Y, 5N, S, Y, 6G, K, 7G, K; [#21:0]). The posterior edge of the pygal plate is roughly straight in all the specimens except for IPS36396a, in which it is slightly concave. The eight costal plates are trapezoidal and show the alternation typical of Testudo, with even costals displaying a wide proximal edge and a narrow distal edge, separated by odd costals with the opposite morphology (Figs 3A, 5C, F, J, L, N, 7G). The even costals

16 A REVISED PHYLOGENY OF EXTINCT TESTUDO 327 host all of the intervertebral sulci. Peripheral plates 1 3, together with the nuchal, form the anterior opening of the shell. Peripheral 1 is rectangular, whereas peripheral 2 is subtriangular, and neither of them protrude significantly from the carapace outline [#13:0]. Only two specimens (IPS33180 and IPS36396a; Fig. 6F, G, respectively) display moderately developed spikes at about the middle of the anterior edges of peripherals 1 and 2, coinciding with the contact between the marginal scutes, and the presence of protrusions on the remaining peripherals at the level of sulci between marginals is quite variable [#30:0,1]. Peripherals 3 7, which are barely visible in dorsal view, are fairly straight and completely involved in the shell bridge (the bony junction between the carapace and the plastron), which is not elevated when compared with the overall shell height [#6:0], as can be observed in MGSB25324, IPS4419, and IPS4420. In some well-preserved specimens (IPS36396a, IPS32957, and IPS4420), it can be ascertained that the external surface of peripherals 3 7 displays a weak longitudinal lateral ridge, and that peripheral 6 does not contact the hyoplastron [#17:0]. Internally, peripheral 7 displays an elongated scar of the dorsal projection of the hypoplastron (only visible in IPS36396a and MGM1561). Peripherals 8 10 are subrectangular, well developed posteroventrally, and partially visible in dorsal view [#8:0]. Peripheral 11, which is sutured to the pygal, costal 8, and suprapygals 1 and 2, is roughly triangular. Except for vertebral 5, the vertebral scutes are <1.4 times wider than long [#9:1] (Fig. 9A), as can be ascertained in various specimens preserving the vertebral series (MGSB25324a, IPS4411, IPS4417, IPS4419, IPS4420, and IPS36396a). Vertebral scute 1 is generally pentagonal, although in two specimens (IPS4420 and IPS32957) it is completely hexagonal, and in all instances it contacts marginal scute 1. Vertebral scute 2 is vaguely hexagonal, although slightly narrower anteriorly; thus deviating from a symmetrical hexagon (IPS36396a, IPS2073, MGSB25324a, MGSB31642, IPS4411, IPS4415, IPS4419, IPS4420). Vertebral scute 3 is hexagonal, with equal anterior and posterior spans except for IPS4420, in which the posterior edges are slightly shorter that the anterior ones (Fig. 5L). In the five specimens completely preserving vertebral 4 (MGSB25324a, IPS4411, IPS4417, IPS4419, IPS4420), it is narrower posteriorly than anteriorly. Vertebral scute 5 is trapezoidal [#27:1], much narrower anteriorly than posteriorly, and its lateral sulci descend through costal 8 to contact marginal 11 (Fig. 5F, L N, Q S, V, Y, Z). It is noteworthy Figure 10. Detailed schematic drawings of neural plate 1 in dorsal view of Testudo (Chersine) catalaunica from the Valles-Penedes Basin compared with those of selected representatives of Testudo s.l. and Indotestudo. Thick solid lines correspond to scute sulci (the posterior sulcus of vertebral 1), whereas dotted lines denote plate sutures (of neural 1 and surrounding plates); thinner solid lines and areas with oblique lines denote damaged or missing shell portions. (A, B) Testudo (Chersine) catalaunica MGSB31642 (A) and IPS30073 (B); (C) Testudo (Chersine) steinheimensis SMNS 80200; (D) Testudo (Chersine) pyrenaica UCBL ; (E) Testudo (Chersine) lunellensis IPS57549; (F) Testudo (Chersine) antiqua SMNS 4450; (G) Testudo (Chersine) burgenlandica NMW 1981/24/1; (H) Testudo (Paleotestudo) canetoniana SA 1857; (I) Indotestudo elongata MNHN.REP64; (J) Testudo (Testudo) graeca MNHN.REP66; (K) Testudo (Testudo) marginata MNHN.ZA ; (L) Testudo (Testudo) kleinmanni MNHN.ZA ; (M) Testudo (Agrionemys) horsfieldii MNHN.REP51; (N) Testudo (Chersine) hermanni MNHN.REP9.