Anatomy, relationships and palaeobiogeographic implications of the first Neogene holomorphic stingray (Myliobatiformes: Dasyatidae) from the early Miocene of Sulawesi, Indonesia, SE Asia

The early Miocene stingray † Trygon vorstmani represented by a single specimen collected from the fish-bearing limestones of the Tonasa Formation of SW Sulawesi, Indonesia, is redescribed here in detail. This taxon exhibits a unique combination of features that clearly support the presence of a new genus, † Protohimantura gen. nov. and its assignment to the whiptail stingrays (Dasyatidae) of the subfamily Urogymninae. The morphological and phylogenetic affinities of † Protohimantura gen. nov. with the living whiprays suggest a close association of this taxon with tropical shallow-water habitats hypothesized for the SW Sulawesi palaeoenvironment during early Miocene. Moreover, this occurrence, which also represents the first holomorphic stingray specimen from the Neogene, provides new insights into the role of the Indo-Australian Archipelago for the evolutionary history of fishes associated with reefs in the context of the shift of the marine biodiversity hotspot across the globe during the last 50 million years.

other myliobatiforms, including variably depressed circular to rhombic discs not more than 1.3 times as broad as long, an angular to obtuse and sometimes very elongated snout, absent caudal and dorsal fins, greatly elongated and slender to whip-like tail with one to four long venomous spines, and a skin ranging from being completely smooth to covered -to varying extents -with small dermal denticles and thorns (Cappetta for Himantura, and the subfamily therefore consists of seven genera: Brevitrygon, Fluvitrygon, Fontitrygon, Himantura, Maculabatis, Pateobatis and Urogymnus. On the contrary, the amphi-American 'Himantura' schmardae (Werner, 1904) has been recently included in a different genus, Styracura, and recognized as closely related to the freshwater stingray family Potamotrygonidae (Carvalho, Loboda & Da Silva, 2016). The representatives of the subfamily Urogymninae have unique characters within the Dasyatidae, such as the absence of skin folds on the tail, presence of a well-developed band of densely packed heart-shaped denticles on the disc with sharply defined margins and relatively narrow base of the tail with an almost circular cross-section (Last et al., 2016a). Moreover, with the exception of the southern Atlantic Fontitrygon, the subfamily appears to be restricted to the Indo-Pacific area (Compagno & Roberts, 1982;Manjaji, 2004;Last et al., 2016a).
Although the fossil record of dasyatids is well-represented, it is heavily biased toward isolated teeth, dermal denticles and caudal spines (Cappetta, 2012  1926; Murray et al., 2015). Remains of elasmobranchs, conversely, are absent in Sumatra and only a single, partial skeleton was recovered from shallow-marine limestones from the early Miocene of SW Sulawesi (de Beaufort, 1926). The goal of this paper is to present a morphological and systematic revision of this Neogene stingray, which was only described very cursorily as †Trygon vorstmani by de Beaufort (1926). The character combination distinguishes the specimen readily from all other dasyatids, therefore representing a new genus of whiptail stingrays of the subfamily Urogymninae. Palaeogeographic implications, based on the analysis of fossil occurrences of this subfamily, provide new insights into the role of the Indo-Australian Archipelago (IAA) for the evolutionary history of whiprays in the context of the shift of marine biodiversity hotspots across the globe during the last 50 million years.

GEOLOGICAL SETTING
The specimen that forms the focus of this study was collected by Professor H. A. Brouwer in 1923 in limestone outcrops near the village of Patoenoeang Asoe E, in the Maros District of SW Sulawesi, Indonesia (de Beaufort, 1926) (Fig. 1). The yellowish fossil-bearing, micritic and laminated limestones of this area belong to the uppermost early Miocene strata of the Tonasa Formation from which other remains of bony fishes and terrestrial flora were recovered (see: Brouwer & de Beaufort, 1923;Brouwer, 1924a;Bartstra, 1977;Tyler, 1997). The Tonasa Formation in the Pangkajene area (where Patoenoeang Asoe E is located) consists of an up to 600-m-thick sequence of shallow-water carbonates deposited from the early or middle Eocene to the middle Miocene in a widespread area of carbonate production known as Tonasa Carbonate Platform (Wilson, 1996(Wilson, , 2000Wilson, Bosence & Limbong, 2000). In SW Sulawesi, the Tonasa Carbonate Platform developed as part of a transgressive sequence to the west of a volcanic arc and is overlain by middle to upper Miocene volcanic rocks of the Camba Formation (Wilson, 2000;Wilson et al., 2000). In the area of Patoenoeang Asoe E, the upper part of the carbonate succession was deposited in a moderate-energy, shallow-water context within the photic zone, as inferred from the presence of larger and small benthic foraminifera, coralline algae, fragmented echinoids, corals and alveolinids (Wilson & Bosence, 1997;Wilson, 2000). The presence of the benthic foraminifer Flosculinella sp. suggested an early Miocene age for the strata (Wilson, 2000), as already hypothesized by Brouwer (1924b) who tentatively referred this sequence to the Burdigalian (about 20.4 to 16.0 Mya). Although lithologies, stratigraphy and tectonic evolution of this area have been extensively documented, mostly in order to study its hydrocarbon resources (see: Wilson, 1996Wilson, , 2000Wilson et al., 2000), the palaeontology and evolutionary significance of the fossil organisms (including bony and cartilaginous fishes) have been poorly investigated so far.

MATERIAL AND METHODS
The single specimen in part and counterpart was collected during a road construction near Patoenoeang Asoe E at the beginning of the 20th century. The specimen, which is housed in the collections of the Naturalis Biodiversity Center Leiden, The Netherlands, and labelled with the repository number RGM 624420, was examined using a stereomicroscope equipped with camera lucida drawing arm. Casts of the embedded teeth and dermal denticles were prepared using silicon compound and epoxy resin, and studied and photographed with a Hitachi S-3500N Scanning Electronic Microscope (SEM) at the University of Bristol. Measurements were taken to the nearest 0.1 mm. Osteological and tooth terminology primarily follows Nishida (1990), Lovejoy (1996), Herman et al. (1998,1999,2000) and Carvalho et al. (2004). Morphometric terminology is adopted and modified from Compagno & Roberts (1982,1984)  u r n : l s i d : z o o b a n k . o r g : a c t : C 1 D 0 B 6 9 3 -3 0 0 C -42C5-9019-30211ACED3BF Type species: †Trygon vorstmani de Beaufort, 1926.
Etymology: From the Ancient Greek word prōto, meaning 'first', 'foremost', 'earliest form of', and Himantura, one of the living whipray genera, thus indicating a possible close relationship between both taxa.
Diagnosis: A whipray characterized by the following combination of characters and body proportions: eye small; interorbital width/eye diameter ratio of 3.5; nasal capsule width/neurocranial length ratio of 0.7; nasal capsule length/neurocranial length ratio of 0.2; anteroposterior fontanelle/neurocranial length ratio of 0.8; scapulocoracoid width/lateral face length ratio of 2.2; 55 propterygial radials; 17 mesopterygial radials; mid-dorsal surface of disc covered by heart-shaped denticles arranged in an antero-posteriorly directed patch having sharply defined outlines; teeth with semiovoid or subhexagonal crown with a second transverse keel; lingual and labial crown ornamentation absent.

Description
The specimen examined is represented by a single partial skeleton lacking part of the external margin of the pectoral disc and the posterior portion of the body, including the tail (Fig. 2). However, the anterior portion of the body is quite complete and preserves several anatomical structures that identify this specimen as a new genus of the family Dasyatidae. Measurements and meristics for †Protohimantura vorstmani are summarized in Table 1. The body preserved in the main slab measures 161 mm from the anterior margin of the disc to the last preserved vertebra, just posteriorly to the second synarcual. Comparing this size to that of modern whiprays, it is therefore likely that the individual could have reached a total length of about 50-60 cm, comparable to the size of an adult individual of most living urogymnines (see: Last et al., 2016a). The high calcified bones corroborate the hypothesis of an adult stage for the specimen. In Fig. 2 the specimen is displayed in ventral view, as suggested by dermal denticles showing their ventral surface, and by the counterslab showing the gill arch skeleton (Fig. 3). The head is relatively close to the anterior margin of the disc and preserves traces of the eye. The eye is small, with the length of eyeball being 3.5 units in interorbital width. The central portion of the disc is covered with an antero-posteriorly directed band of denticles.
Neurocranium: The neurocranium is antero-posteriorly elongate, longer than wide, with its greatest width at  the level of nasal capsules (Fig. 2). The rostral cartilage is absent, resembling the condition of adult stingrays (e.g. Compagno, 1977;Miyake et al., 1992). The nasal capsules are antero-posteriorly short, transversely broad and ovoid in shape. Their anterior margin is rounded and biconvex with a small and triangular anterior median indentation. The nasal capsule width and length are about 70% and 20% of the neurocranial length, respectively. The internasal plate between the two capsules appears antero-posteriorly elongate and extremely narrow. The preorbital processes are small, posteriorly directed, and protrude by the posterolateral aspect of nasal capsules. The supraorbital process is small, triangular in shape and located just anteriorly to the postorbital process. The orbital region is longer than wide. The specimen preserves traces of an eye as a brown-coloured carbon film contoured by its optic capsule (Fig. 4A). The eyeball is ovoid in shape, slightly antero-posteriorly elongated, and possibly consists of an accumulation of melanosome-like microbodies containing molecularly preserved traces of melanin . The neurocranium is narrower at the level of the otic region, with its least width being about 30% of the total neurocranial length. The otic capsules are short. Although the specimen shows the ventral side in the main slab, it is possible to recognize (possibly due to taphonomic compression) the outline of the fronto-parietal fontanelle, which is antero-posteriorly elongated and covers about 75-80% of the neurocranial length; its posterior margin is concave and does not show any indentation. The precerebral fontanelle is difficult to examine. The postorbital processes are broad and shelf-like. The antorbital cartilage is very long and laterally narrow (Fig. 2). Its maximum width is at the level of the articulation with the postero-lateral aspect of the nasal capsule and extends posteriorly close to the hyomandibula at the level of the jaw joint. The antorbital cartilage is simple, not branched, posteriorly directed and articulates with propterygium.
Jaws: Both jaws are poorly preserved and their outline is difficult to describe. However, the palatoquadrate appears labio-lingually compressed, smaller and narrower than the Meckel's cartilage. The Meckel's cartilages are stouter and broader than the palatoquadrate. It is not possible to recognize the medial symphyseal process nor the anterior processes of the Meckel's cartilage, which are present in Himantura and Dasyatis according to Underwood et al. (2017). Wing-like processes, which project laterally from close to the lower jaw symphysis in pelagic stingrays, are absent in †Protohimantura gen. nov. The lateral oral . †Protohimantura vorstmani (de Beaufort, 1926) from early Miocene of Sulawesi, Indonesia. A, RGM 624420, holotype, counterslab. B, reconstruction, dermal denticles omitted. Scale bars 10 mm. Abbreviations: amp, anterior medial plate; cb, ceratobranchial; hyo, hyomandibula; met, metapterygium; pmp, posterior medial plate; pro, propterygium; ps, pseudohyoid; sca, scapulocoracoid; syn1, cervicothoracic synarcual; syn2, thoracolumbar synarcual; vc vertebral centra. diastema appears to be wider than the occlusal width. Antimeres of both upper and lower jaws are narrower and separated at symphysis.
Hyoid and gill arches: The hyomandibulae are preserved only as outline in the main slab (Fig. 2). They appear laterally compressed and narrow at about midlength, slightly arched and with a concave inner margin. The distal end of the hyomandibula articulates with the lower jaw through a strong and stout terminal portion. In the counterpart (Fig. 3), the hyomandibula appears strongly calcified and its proximal portion at the articulation with the otic region is enlarged and stouter than its mesial part. There is no trace of the angular cartilages typical of potamotrygonids or secondary hyomandibular cartilages as in  all the other ceratobranchials articulate with small rami along the lateral margin of the anterior portion of the medial plate. The last two ceratobranchials appear ankylosed but not fused in their proximal portion. The fifth ceratobranchial articulates with the anterior margin of scapulocoracoid. The preservation of the counterslab is so optimal that it is possible to recognize the filamentous branchial rays associated with the ceratobranchials (Figs 3A, 4B), although their number on each ceratobranchial is difficult to discern. Pharyngobranchials and extrabranchials are not preserved in the available material.
Synarcuals and vertebral column: Both anterior (cervicothoracic) and posterior (thoracolumbar) synarcual cartilages are preserved and strongly calcified. Anteriorly, the cervicothoracic synarcual articulates with the occipital condyles of the chondrocranium. Its medial crest, whose exposure in the main slab resulted from taphonomic compression, runs anteroposteriorly along almost its entire length (Fig. 2). The cervicothoracic synarcual possesses lateral stays, which are tab-like, located at about midlength and project perpendicularly laterally, although this might be due to taphonomy, since they should be dorsally directed forming a U-shape structure, as in all myliobatiforms (e.g. Aschliman et al., 2012a). It is not possible to detect the number of fused vertebrae that constitute the first synarcual, or the foramina. The thoracolumbar synarcual is as long as the cervicothoracic synarcual. It articulates anteriorly with the first synarcual but contrary to this latter, the thoracolumbar synarcual is relatively simpler, triangular in shape and tapers posteriorly. About 30 unfused individual vertebral centra can be seen along its entire length. The vertebral centra are strongly calcified, subrectangular in shape and antero-posteriorly short. However, since the posterior part of the body is not preserved, the number of vertebrae forming the vertebral column of †Protohimantura is unknown. Ribs are most likely absent as in all myliobatiforms (McEachran, Dunn & Miyake, 1996; Aschliman et al., 2012a).
Pectoral fins and girdle: The scapulocoracoid consists of a single straight and robust transverse structure, located ventral to the synarcual, and between the basibranchial copula and the articulation between the two synarcuals. Its width is about twice the length of its lateral margins. Anteriorly, the scapulocoracoid articulates with the fifth pair of ceratobranchials. The scapular fossa (or foramen) cannot be recognized in the available material, as well as the ball and socket articulation between scapular process and first synarcual. The suprascapulae are not exposed in the available material. Laterally, the scapulocoracoid articulates with the internal skeleton of the pectoral fins. The propterygium is long and arched, and extends to the anterior disc margin. The propterygium gradually tapers distally. It is distally segmented and the first small segment is adjacent to the anterior margin of the nasal capsule, resembling the condition seen in Dasyatis brevis, Hypanus longus and Fluvitrygon signifer (see: Garman, 1913; Lovejoy, 1996; Aschliman et al., 2012a). The proximal portion of the propterygium is enlarged and articulates with the anterior portion of the lateral margin of the scapulocoracoid, and with the anterior mesial margin of the mesopterygium. The mesopterygium is small and subtriangular in shape; it is a single, non-fragmented element and its external margin is more or less straight, not fused to radials. The mesopterygium is shorter than the proand metapterygium. The metapterygium is poorly and incompletely preserved, lacking its distal portion in the main slab. However, it appears to be long, arched and tapers posteriorly. It is more slender than propterygium.
The metapterygium appears to be a single element, at least in its proximal part. All pectoral radials articulate directly with the pterygia. Although it is not possible to detect the total number of pterygial radials, there are about 55 propterygial and 17 mesopterygial radials. We also counted 25 metapterygial radials but this number is far from being real, due to the lack of the distal portion of the metapterygium. Each radial is composed of at least nine segments. However, since the external margin of the disc is incompletely preserved, it is possible that the number of segments was much higher. For the same reason, the number of bifurcation of each radial before reaching the edge of the pectoral fin margin is unknown. The distalmost radials of †Protohimantura gen. nov. are calcified in chain-like patterns, forming the so-called 'catenated calcification', which is typical of batoids with undulatory swimming mode, including all myliobatiforms with the exception of Plesiobatis, Gymnura and pelagic stingrays (Schaefer & Summers, 2005).
Dentition: Teeth of †Protohimantura gen. nov. are minute and not in pavement-like arrangement, as in myliobatids. The dentition is probably gradient monognathic heterodont with low-crowned teeth,  which decrease in size toward the commissure. Sexual and ontogenetic heterodonties are unknown. The tooth morphology is consistent with that of Himantura uarnak figured by Herman et al. (1998, pls 8-9). In occlusal view (Fig. 5A, C) the crown is semi-oval or subhexagonal in shape, broader than long. The crown has an inwardly bent, low, transverse keel, which divides the crown into distinct labial and lingual parts. A second transverse keel, running more or less parallel to the main keel, is present on teeth of †P. vorstmani, resembling the condition seen in Himantura, Trygonoptera, Urobatis and Urolophus among myliobatiforms (Herman et al., 1999(Herman et al., , 2000, and supports its sister-group relationship with Himantura in this study. Lingual and labial ornamentations seemingly are absent, conversely to the condition seen in teeth of the extant Fluvitrygon signifer (Fig. 6). The basal view of the crown (Fig. 5D) shows a broad and slightly convex crown rim at the outer part, which gradually narrows to half its width at the inner part. The crown-root junction is located in a shallow depression in the centre of the basal surface of the crown. The root is of holaulacorhizous type with two lobes, which are triangular in shape in basal view (Fig. 5D). The root base has a welldeveloped and deep median groove that encloses a single large central foramen. There are no inner or outer foramina discernable. The root is lower than the crown. Due to the poor preservation of the specimen, it was not possible to count the number of tooth rows.
Dermal denticles: The mid-dorsal surface of the disc is covered by a dermal armour consisting of numerous heart-shaped denticles arranged in an incomplete ovoid, antero-posteriorly directed patch, whose outline is sharply defined ( The crown appears to be flat or slightly globular in appearance. Denticles have a sub-circular basal plate without well-differentiated peduncle. The pectoral fins and the region of the disc anterior to the nasal capsules are sparsely covered with smaller denticles with blunt crowns (Fig. 7D). There are no thorns in the specimen examined, although we do not exclude that thorns, as well as one or more stings, might have been present in the not preserved tail.

phylogenetiC analysis
The analysis of 102 morphological characters coded for 29 taxa produced only two most parsimonious trees with the same length of 214 steps and with the same relatively high consistency index (CI = 0.65) and retention index (RI = 0.79), which are depicted in Fig. 8. A complete list of synapomorphies for each node is listed in  One of the main results of our phylogeny is the recovery of a dichotomous nature of myliobatiforms (excluding Hexatrygon), which is consistent, at least in part, with the hypothesis of Nishida (1990), but never recovered in more recent morphological or molecular analyses. This might be due to the use of a large dataset, including 102 morphological characters taken from different works (Herman et al., 1998(Herman et al., , 1999 Table 2).
The monophyly of the clade Dasyatoidea (including all remaining stingrays except Hexatrygon) is supported only by one character, which is the spiracularis split into lateral and medial bundles, with the medial bundle inserting on to the posterior surface of Meckel's cartilage and the lateral bundle inserting onto the dorsal edge of the hyomandibula (ch. 88[1]). This group is entirely composed by members having the general condition of the disc shape (rhomboidal, quadrangular or oval) and 'catenated' calcification of radials, which reflect their undulatory swimming mode and benthic habits (Schaefer & Summers, 2005). The dasyatoid diversification seems to be achieved by the step-by-step adding of characters. The pair formed by Urolophus+Trygonoptera (family Urolophidae) is sister to all dasyatoid stingrays, and its monophyly as detected by Carvalho et al.  Although the relationship between freshwater potamotrygonids and Styracura is almost certainly true, based on morphological, molecular and chrono/geographic evidence, our phylogeny does not recognize this genus as a genuine member of the family. This can be due to the fact that Styracura lacks some characters of the lateral-line, and pectoral and pelvic fin skeleton of potamotrygonids, which on the contrary resemble some dasyatids (Carvalho et al., 2016). The placement of Styracura  (1982,1984) and Nishida (1990) is not supportive of the clade because it is present also in Styracura, potamotrygonids and †Heliobatis, and, therefore, is considered homoplastic. The sister-group relationship between †Protohimantura and Himantura is supported by two characters, which are the presence of a second transverse tooth keel (ch. 99[1]), and mid-dorsal surface of disc covered by heart-shaped denticles arranged in an antero-posteriorly directed patch having sharply defined outlines (ch. 102[1]). This pair forms the sister-group of all remaining dasyatids, whose monophyly is supported by the presence of a caudal fin reduced to tail-folds (ch. 34   Table 2.

Comparison and relationships
Although represented by an incomplete specimen, the morphological analysis of †Protohimantura vorstmani has revealed the presence of several characters that support unquestionably its inclusion in the order Myliobatiformes, including the absence of rostral cartilage, the presence of a very broad and shelf-like postorbital process, and a second (thoracolumbar) synarcual (see: Compagno, 1977; Carvalho et al., 2004; Aschliman et al., 2012a). Although it was not possible to detect the absence of ribs, it is most likely that †Protohimantura might have shown this character, as will all other myliobatiforms. The placement of the Sulawesi stingray within the derived monophyletic clade of the Dasyatidae is supported by a combination of several plesiomorphic characters, e.g. absence of angular cartilages (present in potamotrygonids), first segment of propterygium adjacent to anterior margin of antorbital cartilage or anterior to margin of nasal capsule (posterior to mouth, between mouth and antorbital cartilage, or adjacent to the nasal capsule in non-dasyatids dasyatoids), an external margin of the mesopterygium that is more or less straight and which is not fused to radials (undulated, not fused to radials in Gymnura; highly sinuous, fused with articulating radial elements in Urolophidae; e.g. Carvalho et al., 2004). Moreover, the absence of features characterizing Gymnura and pelagic stingrays (short orbital region with more anteriorly placed supraorbital and postorbital process, secondary hyomandibular cartilages, symphysial fusion of upper and lower jaws, fourth and fifth ceratobranchials fused to each other, lateral expansion of radials in pectoral region, first segment of propterygium adjacent to anterior margin of antorbital cartilage or anterior to margin of nasal capsule, 'crustal' calcification pattern of radials, wing-like body shape, with pectoral fins greatly expanded and different arrangement of teeth on jaws; e.g. Carvalho  , 2016a). In this perspective, the presence of a well-developed denticle band on the disc formed by heart-shaped placoid scales in †Protohimantura gen. nov. corroborates its sister-group relationship with Himantura and its inclusion in the almost entirely Indo-Pacific subfamily Urogymninae. †Protohimantura gen. nov. differs from the other urogymnine genera by having a unique combination of   ., 2016a). In this perspective, the presence of †Protohimantura gen. nov. in the early Miocene limestones of Sulawesi might suggest a close affinity of this taxon with the tropical shallow-water habitats associated with corals, as hypothesized for the uppermost sequence of the Tonasa Formation (Wilson, 1996, 2000; Wilson et al., 2000).
The fossil record of whiptail stingrays of the family Dasyatidae is extensive and well documented. The earliest putative known fossil dates back to the Hauterivian, early Cretaceous, and was included in the genus Dasyatis by Underwood et al. (1999). However, Cappetta (2012) hypothesized that it might represent a different genus, since the simpler morphology of these teeth is quite different from Dasyatis, which has been often used as basket/repository genus for many early fossil teeth with 'dasyatoid' morphology (Underwood et al., 1999; Cappetta, 2012). Except for †Protohimantura vorstmani, the fossil record of whiprays of the subfamily Urogymninae is only represented by isolated teeth that have been all included in the genus Himantura (Fig. 9). However, the paucity of fossil whiprays probably represents an artefact, since the teeth of 'Himantura' are very similar to those of . Although the presence of sampling biases must be considered, it seems evident that the spatial and temporal dynamics of Urogymninae identifies an eastward movement of the fossil occurrences from the Tethys during the Eocene, to the Arabian Peninsula and IAA in the Miocene, and almost all species occurring mostly in IAA today (Fig. 9). A very similar pattern was highlighted for another dasyatid subfamily, the Hypolophinae, whose fossil occurrences indicate a pre-Bartonian origination for the group in western Neotethys, followed by a rapid and widespread colonization of the proto-Mediterranean Sea, western Atlantic and Indo-Pacific during the late Paleogene-early Neogene (Adnet et al.,  2018). Spatial and temporal dynamics of Urogymninae and Hypolophinae appear to be consistent, at least in part, with the 'hopping-hotspots' model hypothesized by Renema et al. (2008), who evidenced that the location of the main marine centre of palaeobiodiversity has moved across the globe during the last 50 million years, triggered by plate tectonics from the Tethys during the Eocene, to the Arabian Peninsula and IAA during the Miocene, before leaving a single hotspot in the IAA in the present days (Renema et al., 2008; Leprieur  et al., 2016). In fact, it has been also suggested that the IAA acted as a region of accumulation and survival from the Palaeocene to the Oligocene, before acting as a centre of origin during the Miocene and, most recently, as a centre of expansion and export from the Pliocene onward ( In this perspective, after their Tethyan origin, the IAA may have acted as a refuge area for whiptail stingrays of the subfamily Urogymninae from the early Miocene, whose representatives later disappeared at least from Tethys. Subsequently, from the late Miocene, the IAA might have acted as a centre of speciation and then, starting in the Pliocene, as a centre of export toward the Mediterranean (with the Pliocene Himantura sp. from northern Italy) and Atlantic realms (with the liv-

CONCLUSIONS
Although the early Miocene stingray from Sulawesi lacks portions of the posterior body, including the tail and the characteristic spines, several features are preserved and allow identification as a new representative of the family Dasyatidae, subfamily Urogymninae, and the creation of a new genus, †Protohimantura. A monophyletic family Dasyatidae is recovered based on the parsimony analyses. The phylogenetic analysis recovered a dichotomous nature of the relationships of the Myliobatiformes, which might reflect a phylogenetic signal in the nature of calcification of their pectoral radials, in their body shape and, consequently, in their swimming style. The analysis of the fossil record of the Urogymninae seems to suggest that the modern distribution of whiprays is the final result of their spatial dynamics across the Palaeogene and consistent, at least in part, with the eastward shift of the marine centre of palaeobiodiversity across the globe during the last 50 million years.