The studied vertebra was collected from sediments belonging to the India Muerta Formation exposed at the intersection between Ruta Nacional 9 and access to Dique El Cadillal (also known as Celestino Gelsi), north of the city of San Miguel de Tucumán, Tucumán Province, northwestern Argentina (Fig. 1).

The India Muerta Formation (Choromoro Group) is represented by discontinuous outcrops scattered in the vicinities of Dique El Cadillal. The lithology of this unit is characterized by grey, coarse- to fine-grained sandstones, either massive or with cross-bedding, and brownish siltstones (both massive and bioturbated) representing a fluvial system of meandering rivers (Bossi et al., 1998).

The fossil record of the India Muerta Formation includes remains of Crocodylia, Testudines, and various mammals (Babot and Ortiz, 2008 and García López and Babot, 2015). The associated mammalian fauna was assigned to the late Miocene Huayquerian South American Land Mammal Age (SALMA) (Bossi et al., 1998).

The fossil specimen is deposited in the PVL. Comparative skeletons of extant snakes at UNMdP-O and fossil specimens belonging to the MACN and MLP collections were used to assist with fossil identifications. Measurements were taken using a digital caliper and they are in millimeters.

Institutional abbreviations are: MACN: Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Colección Nacional de Paleovertebrados, Ciudad Autónoma de Buenos Aires, Argentina; MLP: Museo de La Plata, División Paleontología de Vertebrados, La Plata, Argentina; PVL: Fundación Miguel Lillo, Colección Paleontología Vertebrados, San Miguel de Tucumán, Argentina; UNMdP-O: Universidad Nacional de Mar del Plata, Colección Herpetológica, Sección Osteología, Mar del Plata, Argentina.

Anatomical abbreviations are: cl: centrum length; cth: cotyle height; ctw: cotyle width; H: vertebra height; naw: neural arch width at interzygapophyseal constriction; pr–po: distance between anterior edge of prezygapophysis and posterior edge of postzygapophysis of the same side; pr–pr: distance between outer edges of prezygapophyses; zgh: zygosphene height.

Serpentes Linnaeus, 1758

Alethinophidia Nopcsa, 1923

Macrostomata Müller, 1831

Boidae Gray, 1825

Gaimanophis Albino, 1996

Type species. Gaimanophis tenuis Albino, 1996

Gaimanophis powelli sp. nov.

Fig. 2

Etymology. In honor to my colleague and friend Jaime Powell (†), collector of the studied specimen, for his contribution to the vertebrate paleontology of Argentina.

Holotype: PVL 6547, isolated mid-trunk vertebra.

Type locality and horizon. National Route 9 (26° 39′ 31″ S, 65° 14′ 00″ W), at the access to Dique El Cadillal (Celestino Gelsi), Tafi Viejo Department, north to the city of San Miguel de Tucumán, Tucumán Province, Northwestern Argentina (Fig. 1). India Muerta Formation. Huayquerian SALMA, late Miocene.

Diagnosis. Small boid snake of the genus Gaimanophis differentiated from G. tenuis in the larger size, the prezygapophyses less slanting, the neural spine shorter dorsally than ventrally, and the zygosphene straight bearing an anteriorly protruding tongue.

Measurements. cl = 5.60; cth = 3.60; ctw = 3.84; H = 8.46; naw = 7.90; pr–pr = 10.80; pr–po = 7.76; zgh = 0.80.

The specimen is a nearly complete mid-trunk vertebra, lacking the right postzygapophysis, most of the right zygantrum, part of the posterior edge of the neural arch and left zygosphene. The vertebra is rather eroded, with the condyle and paradiapophyses badly conserved. It is small, robust, wide, low and short, with the vertebral centrum shorter than the neural arch width (cl < naw). In anterior view, the zygosphene is well developed, dorsoventrally thin, straight and presumably wider than the cotyle. It has a small articulate surface conserved on the right side and a thin tongue protruding anteriorly. The prezygapophyses are small, emerging above the level of the neural canal floor, and scarcely inclined on the horizontal plane. Each prezygapophysis is separate from the neural canal by a short, deep notch. The distance between prezygapophyses surpasses greatly the height and length of the vertebra (pr–pr > H > pr–po); thus, the vertebra looks notably widened. The prezygapophyseal process is small, short, and robust. The neural canal is subtriangular, rather large, wide, high, and bearing only a ventral longitudinal crest. The cotyle is rather large, wider than high (ctw > cth), with a thin border. The ventral edge of the cotyle is broken. There are not paracotylar foramina. The paradiapophyses are rather small, low, not projected laterally; they do not reach the level of the ventral edge of the cotyle and have surfaces ventrolaterally oriented.

In posterior view, the neural arch is depressed and wide. The left zygantrum is deep, containing a well-developed articulate surface. Dorsally, the zygantrum is broken. The left postzygapophysis is rather small and scarcely slanting on the horizontal plane. The neural canal looks subtriangular. The condyle is extremely eroded, slightly wider than high.

In lateral view, the vertebra is higher than long, with the vertebral centrum clearly shorter than the neural arch (H > pr–po > cl). The neural arch is depressed. The neural spine is low and its basis is long. It has anterior and posterior edges slightly converging in a dorsal border; thus, the base of the spine is longer than the dorsal rim. Dorsally, the spine is somewhat eroded and broken, but its anterior edge is complete revealing its slight tallness. The articulate surface of the zygosphene is small, oval, and anterodorsally oriented. The interzygapophyseal crest is long, sinuous, and marked. Lateral foramina are present at each side. The subcentral ridges are moderately defined, anteroposteriorly developed along a straight line. Ventrally, the haemal keel surpasses the level of these ridges. The paradiapophyses are not well developed dorsoventrally, far from the prezygapophyseal surfaces and rather downwards oriented, exposing much of their articulate surfaces ventrally. Although the paradiapophyses are eroded, there is no evidence of strong differentiation into di- and parapophyseal surfaces. The condyle axis is slightly inclined on the horizontal.

In dorsal view, the vertebra is short, wide, with the neural arch wider than long (pr–pr > pr–po). The interzygapophyseal constriction is well defined at the middle, but moderately deep and anteroposteriorly short. Although the posterior border of the neural arch is partially broken, it is possible to note the presence of a strong and deep medial notch. The zygosphene has the anterior edge convex, forming a flat tongue, protruding anteriorly. The basis of the neural spine is long but the dorsal border is short, thinner anteriorly than posteriorly. The prezygapophyses are small, with short and narrow surfaces anterolaterally projected. The prezygapophyseal processes are hidden in dorsal view.

In ventral view, the vertebral centrum is short and wide, especially in its anterior part due to a strong divergence of the moderately defined subcentral ridges. The paradiapophyseal surfaces are small, clearly visible in ventral view. The left postzygapophysis has a small articulate surface, which is subtriangular and posterolaterally projected. The cotyle is ventrally visible because of the broken ventral border. The haemal keel is wide and moderately protruding. Its anterior part is widened, but does not reach the width of the cotyle. Posteriorly, it is thinner and finishes as a smooth and not well-defined keel, not reaching the condyle. The subcentral channels are not deep. The subcentral foramina are opened on the anterior part of the centrum.

The described vertebra is attributed to the extinct genus Gaimanophis based on the presence of a combination of characters present in the holotype of G. tenuis, which is the type species of the genus and the single species described up to now (Albino, 1996). These features are: small size; vertebra clearly wider than longer; neural arch depressed; low, thin and long neural spine; zygosphene thin, wider than the cotyle; rather large and subtriangular neural canal; prezygapophyses arising dorsally from the level of the neural canal floor and slightly inclined above the horizontal plane; short and thick prezygapophyseal process; vertebral centrum short and anteriorly notoriously widened; cotyle and condyle wider than high; absence of paracotylar foramina; and paradiapophyses small, not surpassing the ventral edge of the cotyle. The haemal keel of the known vertebrae of the type species of Gaimanophis varies according to the position in the column. It is clearly prominent and distinctive as a short hypopapophysis, or poorly developed and wide (Albino, 1996). The haemal keel of the studied specimen is defined but not protruding, smooth and rather wide anteriorly but thinner posteriorly. It probably corresponds to a vertebra of an intermediate position in the column, between vertebrae with short hypoapophysis and vertebrae with wide and flat keel.

In conclusion, most characters support the assignation of this vertebra to Gaimanophis. Nevertheless, some features between the Patagonian and northwestern specimens are different. The morphology of the zygosphene with an anterior tongue and straight in the middle part, as well as the neural spine shorter dorsally than ventrally, are distinctive characteristics of the studied vertebra, differing from the vertebrae assigned to G. tenuis in Albino (1996) . Moreover, all of the vertebrae referred to G. tenuis are smaller and the prezygapophyses are more inclined than in the new specimen. Thus, the vertebra from India Muerta Formation is referred to a new species of the genus, Gaimanophis powelli sp. nov.

Based on the Patagonian specimens of the genus, Albino (1996) tentatively assigned Gaimanophis to the Erycinae, a group presently absent in South America. The small size, absence of paracotylar foramina, depressed neural arch, and low neural spine are all characters found in this group. However, the erycines are characterized by the presence of caudal vertebrae with additional apophyses (Szyndlar and Rage, 2003), although it was demonstrated that snakes with erycine-like trunk vertebrae could possess caudal vertebrae without additional apophyses (Szyndlar and Böhme, 1996). Unfortunately, caudal vertebrae of Gaimanophis are unknown, so the assignation of this genus to Erycinae is not well supported.

Although absent from the Mesozoic, boid snakes are present in South American since the Paleocene (Albino and Brizuela, 2014). Until some years ago the Paleogene boid record came only from Argentina and Brazil, with few remains from Bolivia (Albino, 1987, Albino, 1990, Albino, 1992, Albino, 1993, Albino, 2011, Albino, 2012, de Muizon et al., 1983, Rage, 1991 and Rage, 2001), whereas the Neogene provided remains from Argentina and Colombia (Albino, 1992, Albino, 1996, Albino et al., 2013, Hecht and LaDuke, 1997 and Hoffstetter and Rage, 1977).

The Paleocene and Eocene climates were characterized by quite warm and tropical conditions extended north and south towards the poles. In this scenery, Paleogene boids were relatively diverse in South America. The Patagonian Paleogene boid fauna included small (Waincophis), medium-sized (Boa) and large forms (Chubutophis), but only small genera (Waincophis, Gaimanophis) persisted into the Miocene. The absence of large snakes in the Patagonian Miocene with respect to those from earlier periods would has been related to decreased temperatures, which is consistent with known changes in mean global annual temperatures reported for this epoch (Albino, 2012 and Albino and Brizuela, 2014). Nevertheless, the climatic conditions in the Patagonian region during the Miocene would have been yet temperate enough to permit the presence of small boids at latitudes around 44°S, where they do not exist today (Albino and Brizuela, 2014). There are no records of post-Miocene boids from Patagonia, but during the latest Miocene and Pliocene some boids still lived outside their present range (Albino, 1992, Albino and Carlini, 2008 and Albino et al., 2013). In addition, the genus Boa inhabited the Mesopotamian region of Argentina during the Late Pleistocene, but changes in the Paraná and Uruguay Rivers would have been caused its extinction from this region (Albino and Carlini, 2008). No other fossil boid has been reported for Argentina, except for the new species of Gaimanophis from the northwestern region described in this paper.

With respect to the Brazilian Paleogene boids, Rage (2001) recognized some extinct genera (Waincophis pressulus, W. cameratus, Hechtophis austrinus) as well as the oldest record of the extant Corallus (Corallus priscus). During the latest years, the record of fossil boids from Brazil has been enlarged with the description of remains collected from Neogene sediments outcrops in southwestern Brazilian Amazonia, including the extinct Waincophis and the earliest presence of the extant genera Eunectes and probably Epicrates (Hsiou and Albino, 2009 and Hsiou and Albino, 2010). Furthermore, the fossil record of boids from Colombia, which was limited to the extant Eunectes from the Miocene (Hoffstetter and Rage, 1977), has been increased with the description of a very large boid from the Paleogene (Titanoboa cerrejonensis, Head et al., 2009). The latest years also provided records of boid snakes from the late Miocene of Venezuela, which include the extant genus Eunectes (Head et al., 2006 and Hsiou and Albino, 2010).

The identification of a new species of the extinct boid Gaimanophis in the Dique El Cadillal increases the diversity of the genus into two species. Besides, it indicates a wider temporal and geographical distribution of the genus than the previously known. At present, Gaimanophis extends from the early Miocene of Patagonia (G. tenuis in Chubut province, Albino, 1996, indicated by an arrow in Fig. 1) to the late Miocene of northwestern Argentina (G. powelli in Tucumán province, this work).

Boids are currently slightly diverse in South America, containing five genera (Boa, Corallus, Epicrates and Eunectes) with 17 species distributed in tropical and subtropical environments from Colombia and Venezuela to Argentina, north to 34°S. Known since the Paleocene, fossils belonging to this group do not appear to be representative of the real past diversity in the continent. This would be in part consequence of the lack of enough studies focused on Cenozoic snakes of South America. Nevertheless, the most recent discoveries in Argentina, Brazil, Colombia and Venezuela demonstrate the importance of focusing future prospections for fossil snakes in the Neogene of little explored regions. In this sense, the recognition of a new species of boid in the late Miocene of northwestern Argentina increases the known diversity of this group in this territory.