INTRODUCTION

Sparassodonts, an extinct group of metatherians, were the predominant mammalian carnivores in South America from the Paleocene (Tiupampan or Peligran South American Land Mammal “Age,” or SALMA; Forasiepi and Rougier, 2009; Muizon et al., 2018) to the late early Pliocene (Chapadmalalan SALMA; Goin and Pascual, 1987; Prevosti et al., 2013). Sparassodont finds span the continent, from the La Guajira Peninsula of Colombia (Suarez et al., 2016) to the famous fossil beds of Santa Cruz Province, Argentina (Sinclair, 1906; Prevosti et al., 2012). The record of sparassodonts is heavily geographically biased, however, with most fossils coming from Argentina (Forasiepi, 2009) and a few localities in Bolivia (Villarroel and Marshall, 1982; Petter and Hoffstetter, 1983; Muizon, 1998; Forasiepi et al., 2006; Engelman and Croft, 2014; Engelman et al., 2015; Muizon et al., 2018) and Colombia (Marshall, 1977; Goin, 1997).

Of the small number of sparassodonts reported from Chile, most derive from the earliest middle Miocene Río Frías Formation at the locality of Alto Río Cisnes in the south (∼44.5° S). Marshall (1990) identified four sparassodont taxa from this locality, the hathliacynids Sipalocyon gracilis and Cladosictis patagonica and the borhyaenoids Prothylacynus patagonicus and Borhyaena tuberata. This sparassodont fauna is generally similar to slightly older sparassodont faunas from the Santa Cruz Formation (Marshall, 1990; Prevosti et al., 2012). More recently, Flynn et al. (2002) and Flynn et al. (2008) reported Cladosictis and cf. Sipalocyon in faunal lists of late early Miocene levels at Pampa Castillo in southern Chile (∼47° S) and Laguna del Laja in south-central Chile (∼37.5° S), respectively. Notably, all these occurrences span a relatively narrow temporal interval (late early Miocene–earliest middle Miocene; Flynn and Swisher, 1995; Flynn et al., 2002, 2008) and appear to represent taxa previously known from the Santa Cruz Formation of Argentina.

Here, we describe a new sparassodont from the Los Helados Locality of the Abanico Formation, near Estero Los Helados in the Tinguiririca River drainage in central Chile. Although the Abanico Formation is best known for producing early Oligocene mammal fossils (e.g., Flynn et al., 2003a; Hitz et al., 2006; Croft et al., 2008a; Bertrand et al., 2012; Bradham et al., 2015, and references therein), its assemblages span a wide age range, from Eocene to Miocene (Flynn et al., 2003b, 2012; Hitz et al., 2006; Croft et al., 2008b). The specimen described here is late Eocene in age, making it one of the few late Eocene sparassodonts known. It is also the first sparassodont from the Abanico Formation to be formally described (though another specimen is currently under study by our group; Engelman et al., 2017). The sparse representation of this group within the Abanico Formation is somewhat puzzling given the unit's overall richness in mammal fossils (Croft, 2006). This is the first taxon to be described from what promises to be an intriguing, but as yet little-studied, Andean fossil mammal fauna.

MATERIALS AND METHODS

We follow the SALMA chronology formulated by Flynn and Swisher (1995), as modified by Flynn et al. (2003a), Ré et al. (2010), Woodburne et al. (2014), and Krause et al. (2017).

Three samples of pyroclastic flow and fall deposits (CH-130, CH-131, and CH132B) that stratigraphically bracket the horizon that produced the specimen described here (SGOPV 6200) were dated via argon geochronology. Purified ∼30 mg separates of plagioclase were produced via standard density, magnetic separation, and hand-picking techniques. Samples were analyzed in the ⁴⁰Ar/³⁹Ar geochronology laboratory at the University of California, Santa Barbara, by incremental heating in a Staudacher-type resistance furnace. Isotopic analyses were obtained on a MAP 216 mass spectrometer, using the general procedures and system described by Gans (1997). The flux monitor used for all irradiations was Taylor Creek Rhyolite (Dalrymple and Duffield, 1988) with an assigned age of 28.27 Ma in order to make it compatible with an assumed age of 28.1 Ma on Fish Canyon Tuff Sanidine (another widely used standard). All errors given for our estimated (preferred) ages as reported throughout the text and in table 1 are ± 2σ (95% confidence).

GEOGRAPHIC AND GEOLOGIC SETTING

The specimen described below was recovered from the Abanico Formation in the Estero Los Helados, a northern tributary of the Río Tinguiririca immediately west of Río Azufre (fig. 1), in the Andean Main Range of central Chile. The Río Azufre drainage produced the early-diverging interatheriid notoungulate Antepithecus brachystephanus (Hitz et al., 2006), which suggests an early late Eocene age (Barrancan subage of the Casamayoran SALMA sensu Cifelli, 1985) for at least one level of the >∼1.5 km thick, continuous section of the Abanico Formation exposed in that region. The formation exceeds ∼3.6 km in composite thickness along the main trunk of the Tinguiririca drainage from roughly the longitude of Los Helados to Termas del Flaco (Mosolf, 2013; Mosolf et al., 2018), though its base is unexposed in the area. The oldest dated horizon in the Abanico Formation in the Río Tinguiririca drainage, or anywhere in the Andean Main Range, is roughly 75 Ma (i.e., Late Cretaceous; Mosolf, 2013; Mosolf et al., 2018), but the age of the oldest fossiliferous level in the unit has yet to be securely established. The lower portion of what is generally regarded as the Abanico Formation in the Tinguiririca drainage (an interval evidently devoid of fossils) has recently been reassigned to the Plan de los Yeuques Formation (Muñoz et al., 2018). This more restricted view of the Abanico Formation influences notions of its thickness and the age of its base, as the Los Lunes, Guanaco, and Guzmana members of the Abanico Formation (sensu Mosolf et al., 2018) would shift to the Plan de los Yeuques Formation. This debate has no bearing on the age or stratigraphic provenance of SGOPV 6200, however, as the specimen was collected from stratigraphically higher levels of the Abanico Formation.

FIG. 1.

Map of Chile (left) and central Chile (inset box) showing the location of Los Helados (LH, at arrow) and other selected Paleogene Chilean fossil mammal localities of the Abanico Formation. Gray area in inset box represents Abanico Formation outcrops. Abbreviations: Az, Azufre (middle Eocene?); Cp, Cachapoal (early Oligocene); LQ, Los Queñes (late Eocene? and late Oligocene?); Tn, Tinguiririca (early Oligocene); Tp, Tapado (middle Eocene?).

The stratigraphic succession of the Abanico Formation in the Los Helados drainage consists of a thick succession of ignimbrites, reworked pyroclastic flows, debris flows, ash and lapilli fall deposits, and fluvial sandstone and conglomerate (fig. 2). All sedimentary units are composed entirely of volcaniclastic detritus—including an abundance of pumice, vitric ash, and well-formed volcanic crystals—suggesting a local source of juvenile tephra and pyroclastic deposits. Units dip moderately westward or northwestward, and the exposed section exceeds 500 m in thickness. The detailed stratigraphy and U-Pb and ⁴⁰Ar/³⁹Ar geochronology of the entire section will be reported elsewhere (Gans et al., in prep.). Here, we describe briefly the geologic context and geochronologic constraints on the fossil-bearing horizon from which SGOPV 6200 was obtained.

FIG. 2.

Simplified stratigraphic column for the Los Helados area. Radioisotopic (⁴⁰Ar/³⁹Ar) dates discussed in the text are denoted in bold. Approximate U-Pb zircon ages for older samples are also illustrated but should be considered very preliminary.

This fossil was collected from a bench far up the southeastern side of the drainage from a pumiceous debris flow within a ∼50 m thick sequence of sandstone, conglomerate, and lapilli fall deposits. Excellent age brackets exist for SGOPV 6200, as the specimen was recovered from a level approximately midway between two primary volcanic deposits for which we have obtained reliable ages. A dacite ignimbrite (sample CH-130) approximately 15 m stratigraphically below the fossil yielded an ⁴⁰Ar/³⁹Ar plateau age on plagioclase of 37.13 ± 0.24 Ma (table 1). This ignimbrite is a pumiceous crystal-rich, weakly welded tuff with abundant fresh crystals of plagioclase, biotite, and hornblende and trace amounts of quartz. A lapilli and ashfall tuff approximately 20 m stratigraphically above the fossil (sample CH-131) yielded a reasonably flat ⁴⁰Ar/³⁹Ar age spectrum on plagioclase with an interpreted age of 35.86 ± 0.38 Ma (table 1). This deposit contains pristine glass shards and pumice lapilli with fresh crystals of plagioclase and minor pyroxene, indicating it is a primary fall deposit that has not been reworked. A slightly higher pyroclastic flow deposit (sample CH-132B) that displays some evidence for sedimentary reworking (abundant somewhat rounded andesitic cobbles, highly rounded pumice) yielded a slightly older plagioclase plateau age of 36.95 ± 0.28, suggesting it may have incorporated some older detritus. The simplest interpretation is that SGOPV 6200 is between 37 and 36 Ma, with maximum and minimum allowable age ranges of 37.4–35.5 Ma and 36.9–36.24 Ma, respectively. This inferred age partly overlaps the ∼38–37 Ma range estimated for Mustersan SALMA faunas at Gran Barranca in Argentina, based on isotopic dating of the El Rosado and Bed 10 tuffs (Bond and Deschamps, 2010; Madden et al., 2010; Ré et al., 2010; Dunn et al., 2013). Referral to the Musteran SALMA is supported by preliminary identifications of the associated fauna, which includes polydolopid metatherians, dasypodid xenarthrans, the archaeohyracid notoungulate Pseudhyrax, early-diverging interatheriids (“notopithecines”), and toxodontian notoungulates that are less hypsodont than those typical of early Oligocene (Tinguirirican SALMA) sites. Therefore, the stratigraphic level at Los Helados that produced SGOPV 6200 appears to pertain to the Mustersan SALMA, congruent with previous interpretations (Croft et al., 2008b; Flynn et al., 2012).

TABLE 1.

Summary data for ⁴⁰Ar/³⁹Ar analyses.

SGOPV 6200 is thus late Eocene in age, predating the earliest Oligocene Tinguiririca Fauna, which was recovered from deposits north and south of the Río Tinguiririca near the town of Termas del Flaco (∼22 km to the SSE) and is the basis for the Tinguirirican SALMA (Flynn et al., 2003a). Isotopic dates indicate that the Tinguirirican is early Oligocene in age; horizons bearing Tinguirirican fossils and those immediately bracketing them are no younger than ∼31.5 Ma (Flynn et al., 2003a), and correlative faunas have been determined to be no older than ∼33.5 Ma (Ré et al., 2010; Dunn et al., 2013). SGOPV 6200 likely postdates the Tapado Fauna from the main drainage of the Río Tinguiririca, which is geographically interposed between Estero Los Helados and Termas del Flaco. Although the stratigraphic section producing the Tapado Fauna has now been dated isotopically (Mosolf et al., 2018), the fossil localities and geochronologic results have yet to be fully integrated. Nevertheless, the composition of the Tapado Fauna suggests a middle Eocene age, similar to that inferred for the level producing Antepithecus at Azufre (Flynn et al., 2005, 2012; Croft et al., 2008b). SGOPV 6200 appears to be roughly equivalent in age to lower levels of Los Queñes, located ∼30 km to the southwest in the western reaches of the Río Teno. Preliminary assessments of the mammals recovered from these levels suggest they also correspond to the late Eocene Mustersan SALMA (Flynn et al., 2012).

SYSTEMATIC PALEONTOLOGY

FIG. 3.

Photo of the holotype of Chlorocyon phantasma (SGOPV 6200) showing orientation of dentaries as they were discovered. Scale = 5 cm.

FIG 4.

Right (A, C, E) and left (B, D, F) dentaries of the holotype of Chlorocyon phantasma (SGOPV 6200) in lingual view. A–B, Original specimen; C–D, replica of original specimen with the natural molds and areas of preserved bone and teeth painted light gray for enhanced contrast; E–F, “negative” cast made by infilling the natural molds preserved in the original specimen, with areas representing teeth and bone painted light gray. Scale = 1 cm.

FIG 5.

Holotype of Chlorocyon phantasma (SOGO-PV 6200): line drawings of “negative” casts made by infilling the natural molds of the original specimen (A, B) and the original specimen (C). A, right dentary in lingual view; B, left dentary in lingual view (with anterior region omitted); C, left dentary in labial view (reversed to match B). Areas where the border of the specimen is uncertain due to poor contrast between the cast and the matrix are denoted by dashed gray lines; areas where matrix has infilled the bone are denoted in gray. Scale = 1 cm.

TABLE 2.

Mesiodistal lengths (in mm) of the lower dentition of the holotype of Chlorocyon phantasma (SGOPV 6200). ^* = estimated measurement.

DISCUSSION

Chlorocyon cannot be referred to the Proborhyaenidae or Thylacosmilidae based on its closed-rooted canines, lack of a symphyseal flange, and presence of a metaconid on m2–4, among other features. Chlorocyon resembles borhyaenids in the presence of metaconids on m2–4, but it differs from all members of this group in lacking bulbous roots on its premolars, possessing diastemata among the premolars, and having a straight posteroventral margin of the dentary in lateral view. It is unlikely that Chlorocyon pertains to the Hathliacynidae, despite its relatively small size (see below). Although the early history of hathliacynids is poorly understood, all currently known hathliacynids lack metaconids on m2–4, have a hypoconulid on m4, and have a curved posteroventral margin of the dentary in lateral view. Hathliacynids other than Pseudonotictis pusillus and Borhyaenidium altiplanicus are also characterized by a p3 in which the anterior edge is less curved than the posterior one (as opposed to a more curved anterior edge, as in Chlorocyon). The only potential hathliacynid synapomorphy of Chlorocyon is a diastema between the lower canine and p1, a feature present in Acyon spp., Borhyaenidium spp., Cladosictis spp., Notogale mitis, and Sipalocyon gracilis, but absent in Perathereutes, Pseudonotictis, and Notocynus. A diastema between the lower canine and p1 is absent in other sparassodonts with gracile rostra, including Allqokirus australis, the Patene specimen from the Quebrada de Los Colorados Formation, Hondadelphys fieldsi, Stylocynus paranensis, Lycopsis viverensis, and possibly L. longirostrus (the holotype of this last taxon is of an immature individual and the anterior end of the dentary is not complete).

Among remaining sparassodonts, Chlorocyon most closely resembles the late Eocene basal borhyaenoid Plesiofelis schlosseri and the Oligocene borhyaenoid Pharsophorus spp. in having a p3 with an anterior edge that is more curved than the posterior edge, absence of the precingulid on p2 (the precingulid also is absent in a few other taxa such as Pseudothylacynus, but the condition is ambiguous in Pharsophorus spp.), presence of a metaconid on m2–4, and a straight posteroventral margin of the dentary in lateral view (shared with Pharsophorus lacerans and Plesiofelis but possibly absent in Pharsophorus tenax). However, Chlorocyon also differs from Plesiofelis and Pharsophorus in several respects. First, Chlorocyon is much smaller: both Plesiofelis and Pharsophorus are large-bodied taxa, the smallest species of which (Pharsophorus tenax) is estimated to have weighed about 21 kg (table 3). By contrast, Chlorocyon, with a molar row length ∼55%–65% that of Plesiofelis and Pharsophorus lacerans and ∼70% that of Pharsophorus tenax, is much smaller, with an estimated body mass of 5.9 kg (table 3), roughly the size of the hathliacynid Cladosictis patagonica (Ercoli and Prevosti, 2011). In addition, the dentary of Chlorocyon is much shallower—in both absolute and relative terms—than that of Plesiofelis or Pharsophorus (table 3). Chlorocyon further differs from these taxa in having diastemata among the lower premolars, and is distinct from Pharsophorus lacerans in having p1–3 that increase gradually in size, rather than p1 being significantly smaller than p2–3 (the condition is unknown in Plesiofelis and Pharsophorus tenax). Chlorocyon is comparable in size to the late Eocene Procladosictis anomala (see Croft et al., 2018), but the latter taxon is known only from a highly unusual upper dentition, thereby precluding direct comparison of dental morphology. Nonetheless, P. anomala is considered a hathliacynid or basal sparassodont rather than a borhyaenoid (Marshall, 1981; Forasiepi, 2009), thus distinguishing it from Chlorocyon, which is most likely a borhyaenoid. The talonids of Chlorocyon also are slightly shorter mesiodistally than would be expected for P. anomala, judging from the mesiodistal length of the protocones of the holotype of P. anomala (the talonid of m2 is ∼2.2 mm long in Chlorocyon, while the protocone of M2 is 3.2 mm long in P. anomala). In summary, Chlorocyon likely represents a borhyaenoid most closely related to Plesiofelis schlosseri and Pharsophorus spp., but differs sufficiently from these taxa (smaller size, shallower dentary, diastema among the premolars, premolar size differences, etc.) to warrant recognition as a new genus and species.

TABLE 3.

Lower molar row mesiodistal length (Lm1–4), estimated body mass, and dentary depth at m4 (Dm4) of Chlorocyon phantasma and species of Pharsophorus and Plesiofelis. All measurements in mm and all body mass estimates (in kg) were determined using m1–4 length and the regression equation of Myers (2001) for dasyuromorphians. Body mass and dentary depth for Chlorocyon was estimated using the left dentary, as the ventral border of the dentary and lengths of m1–4 are better preserved on this side. Dentary depth for Plesiofelis schlosseri was measured in such a way as to compensate for the likely inaccurate restoration of the jaw (see text). The body masses of Pharsophorus lacerans and Plesiofelis schlosseri represent extrapolations from the modern dasyuromorphian comparative dataset used by Myers (2001), but the estimated masses for SGOPV 6200 and Pharsophorus tenax are within the range of variation of the sample used to generate the equation and thus likely to be more accurate.

In addition to representing a new taxon, SGOPV 6200 is noteworthy for its late Eocene (Mustersan SALMA) age. Only a handful of Mustersan sparassodont specimens have been described previously, all from localities in Argentina: the holotype of Procladosictis anomala from Gran Barranca (Marshall, 1981); three specimens from the locality of Cerro del Humo (including the holotype of the borhyaenoid Plesiofelis schlosseri; Simpson, 1948; Marshall, 1978); four undescribed specimens from Gran Hondonada (Ruigómez, personal commun.) that have been referred to Procladosictis and Plesiofelis (Cladera et al., 2004: table 1); and an isolated lower molar of the proborhyaenid Callistoe sp. from Antofagasta de la Sierra (Goin et al., 1998; Powell et al., 2011).

The pre-late Oligocene record of Sparassodonta is remarkably poor, with most occurrences representing basal sparassodonts (e.g., Patene) or proborhyaenids, a grouping of exclusively Paleogene sparassodonts only distantly related to most Miocene forms (except possibly thylacosmilids; see Babot, 2005; Forasiepi et al., 2015; Engelman et al., 2017). If proborhyaenids are deeply nested within Borhyaenoidea, as is consistently recovered by many studies (Muizon, 1999; Babot et al., 2002; Forasiepi, 2009; Engelman and Croft, 2014; Forasiepi et al., 2015; Suarez et al., 2016; Muizon et al., 2018), then the major lineages of late Cenozoic sparassodonts (e.g., hathliacynids, borhyaenids) must have originated by the middle Eocene (Vacan subage of the Casamayoran SALMA), based on the earliest widely accepted occurrence of a proborhyaenid (Babot et al., 2002; Powell et al., 2011), or potentially even the early Eocene, based on possible proborhyaenid remains from the Las Violetas Fauna (Gelfo et al., 2010; Krause et al., 2017). An early or middle Eocene divergence of major sparassodont clades is supported by a recent report of a possible borhyaenid from the middle Eocene locality of La Barda, Argentina (Lorente et al., 2016). Chlorocyon indicates that the morphological diversity of Eocene sparassodonts was greater than previously thought, and that non-proborhyaenid borhyaenoids were far more diverse and morphologically disparate during the Eocene than currently reflected in the fossil record. This observation is compatible with the long ghost lineages inferred for many Neogene taxa and further highlights the need for additional sampling of Eocene localities from Chile and throughout South America. No less importantly, SGOPV 6200 provides a tantalizing preview of the paleontological novelty that will emerge from the Abanico Formation at Estero Los Helados as its fauna(s) become more completely studied.