Neogene ‘Horn Sharks’ Heterodontus (Chondrichthyes: Elasmobranchii) from the Southeastern Pacific and Their Paleoenvironmental Significance

Abstract. Horn sharks (Elasmobranchii: Heterodontus Blainville) correspond to a genus of chondrichthyan fishes, mostly distributed in warmtemperate to tropical regions of the Pacific and Indian Oceans. The fossil record shows that, in contrast to its current distribution, horn sharks were widely distributed both in the eastern Pacific and western Atlantic during the Neogene, being subsequently extirpated from some of these areas. In this contribution, we describe new Heterodontus teeth from three Pliocene localities in the Coquimbo Region, in north-central Chile, and make an extensive revision of the fossil record of the genus in the Americas, in order to specify the timing of their extirpation in the southeastern Pacific and discuss the possible causes of this event. The new specimens described herein belong to a species with a Heterodontus francisci type dentition. Our analysis suggest that the removal of horn sharks occurred in the context of a general faunal turnover in the transition from Pliocene to Pleistocene, and that it was probably controlled by an interplay between the oceanographic, tectono-eustatic and ecological changes occurred in the region at that time.

. Distribution of Heterodontus in the Americas, from Neogene to Recent. Fossil records have been compiled from specialized literature (references in Table 1), while current occurrence records have been taken from the UNESCO's Ocean Biogeographic Information System database (OBIS, 2018). Note the existence of a large extirpation area, with no horn sharks occurrences at Recent times. and Western Asia, northern Africa, South America, North America and Australia (Christiansen and Bonde, 2002;Fitzgerald, 2004;Kennedy et al., 2008;Cappetta, 2012). In the Americas, their fossil record ranges from the Cretaceous to the Pleistocene (e.g., Cappetta, 2012;Carrillo-Briceño et al., 2018: tables S3-S4). Their current distribution, however, is very limited (Fig. 1). The Neogene fossil record of the Americas, for instance, indicates a wider distribution that included also the Western Atlantic coast, as far as Argentina (e.g., Ameghino, 1906;Cione, 1978;Case, 1980;Laurito, 1999;Cione et al., 2000Cione et al., , 2005Cione et al., , 2011Aguilera, 2010;Carrillo-Briceño et al., 2015. Likewise, for the eastern Pacific coast, the fossil record suggests that Heterodontus reached latitudes far as south as 34°S during the Neogene, thus inhabiting waters off the coast of north and central Chile (e.g., Walsh, 2001;Suárez et al., , 2006Carrillo-Briceño et al., 2013;Suárez, 2015), where today it is absent.
This region is nowadays affected by the upwelling of cold waters, brought from the sub-Antarctic region by the Humboldt Current (Marchant et al., 2007;Montecino and Lange, 2009), which creates an environment that contrasts with the ecological preferences of horn sharks. Their ancient presence at mid-latitudes agrees well with the warmer water conditions suggested for this region during Miocene and Pliocene times (e.g., Ibaraki, 1997;Tsuchi, 2002;Dekens et al., 2007;Nielsen and Glodny, 2009;Le Roux, 2012). Therefore, the subsequent extirpation of Heterodontus from mid-latitudes in the southeastern Pacific is probably the consequence of deep paleoenvironmental changes and seems to be related to the oceanographic, tectonic and ecological changes that took placed in that region during Plio-Pleistocene times. The reasons that explain the decline of horn sharks in the southeastern Pacific are possibly different from those for which they were extirpated from the western Atlantic, since these events occurred at different times and in dissimilar oceanographic and tectonic settings.
In any case, to identify the controls that may have caused the extirpation of Heterodontus, it will be necessary to pinpoint the timing of its decline and disappearance in the fossil record.
In this contribution, we report and describe an abundant occurrence of horn shark's fossil teeth collected from new localities in the Coquimbo Region (north-central Chile). We also analyze the Heterodontus fossil record in the Americas, reviewing the geographic and chronostratigraphic distribu-tion of the source localities in order to discuss the paleobiogeographic evolution of the genus and the time and mode in which the extirpation of Heterodontus may have occurred.

MATERIALS AND METHODS
The specimens herein described were collected through sediment sampling at three fossiliferous sites-called 'Lomas del Sauce' (LdS), 'Los Clarines' (LCIV), and 'Quebrada Camarones' (QCT)-and subsequent sieving using meshes with 6, 3, and 2 mm openings. This allowed generating three rich collections of anterior and lateral teeth (89 from LdS, 46 from LCIV and 17 from QCT), most of which are complete and well-preserved. The specimens are housed at the Sala de Colecciones Biológicas of the Universidad Católica del Norte, Chile (SCBUCN).
Measurements and tooth terminology used in the text are illustrated in Figure 2. Taxonomic determination of the material were achieved by following the illustrated descriptions provided by Taylor (1972), Reif (1976) Froese and Pauly, 2018;OBIS, 2018). Sea surface temperatures (SST) in areas of occurrence of these taxa have been obtained by registering the mapping parameters provided by Kaschner et al. (2016) for each species; in particular, the information associated with the cells used for creating environmental envelope

GEOLOGICAL SETTING
The collected material comes from three sites of the Neogene marine sediments of the Coquimbo Formation, in north-central Chile, and it is the first record of Heterodontus for this geological unit (Fig. 3). Both LdS and LCIV sites are placed in the city of Coquimbo, in front of La Herradura bay; while QCT site is located in the Tongoy bay area (Fig. 4).  (Long, 1993a;Suárez y Marquardt, 2003;Le Roux et al., 2006;Staig et al., 2015;Suárez, 2015). Sea surface temperatures (SST), are illustrated for the regions of occurrences of each taxon, based on the current records provided by Kaschner et al. (2016) in "Aquamaps" (datasets in Supplementary Online Material 1). For each taxon, it is shown: the temperature range for all the occurrences (thin bars), for the records located between the 10th and 90th percentiles (medium bars), and for those located between the 20th and 80th percentiles (thick bars). Mean temperature for each species is also shown (vertical white line inside the thick bars). For taxa with no SST data available, a dashed line is used.
Locally, some facies of sandstones and coquinaceous conglomerates of Pleistocene age appear, whose fauna and taphonomic and sedimentological characteristics are remarkably different. They are regression sediments that are distributed in some high areas of the terrace, resting on the Pliocene series in erosional disconformity (Herm, 1969;Paskoff, 1970). Based on their litho-and biofacies, LdS can be entirely assigned to the Pliocene series, whereas in LCIV both Pliocene and Pleistocene series crop out (Fig. 5). At LdS, the Pliocene age is supported by the association of the gas-  (DeVries, 1997;DeVries and Vermeij, 1997;Guzmán et al., 2000;Nielsen, 2013). The upper member of LCIV is a succession of sand, limestones and coquinaceous layers that overlies the Pliocene sediments with an erosional disconformity. Their mollusk fossil content includes the bivalves Argopecten purpuratus, Cyclocardia compressa, Mesodesma donacium, and Mulinia edulis, as well as the gastropods Incatella cingulata, and Concholepas concholepas. This favors a Pleistocene age for the upper member, since all these species are forms which appeared or acquired massive development in the region during that epoch (Herm, 1969;DeVries, 1995;Guzmán et al., 2000;Tsuchi, 2002).

Quebrada Camarones, Tongoy
In the Tongoy area (Fig. 4), the marine sediments appear as a succession of muds, sands, coquinas and gravel infilling an extensive paleobay surrounded by topographic highs to the west and to the east (Le Roux et al., 2006). The stratigraphic interval encompassed by the whole set is broad: while the basal layers date back to the early Miocene, the younger strata were deposited during the Pleistocene (Paskoff, 1970;Martínez-Pardo, 1979;Martínez-Pardo and Caro, 1980;Le Roux et al., 2006). As in Coquimbo, they form marine terraces that have been affected by fluvial erosion, thus resulting in different ravines where the stratigraphy is exposed. One of them is Quebrada Camarones, next to the town of Tongoy. QCT site (30° 19' 14" S; 71° 26' 32" W; Fig.   4) is located on the north bank, where a 20 m-thick sedimentary succession crops out on a slope (Fig. 6). The strata where Heterodontus materials come from, overlies coquinaceous layers that contain the gastropods Concholepas kieneri and Concholepas nodosa, and underlies calcareous coquinaceous beds with the gastropods Chorus grandis, Chorus giganteus, and C. doliaris. Both assemblages support a Pliocene age for the source beds (DeVries, 1995(DeVries, , 1997Nielsen, 2013).
?Heterodontus francisci (Girard, 1854)   and usually higher than wide. They range from 2.4 to 3.6 mm in height, and from 1.9 to 3.6 mm in width. Their crown is tricuspid, with a triangular and erect main cusp and two lateral cusplets, which are joined to the base of the main cusp and converge slightly towards it. The height of the main cusp is significantly greater than that of the lateral cusplets.
The crown expands over the root both labially and lingually, forming a wide apron (Fig. 2). On the lingual face the apron rests over the root, reaching the root edge. On the labial face, it extends beyond the root with a rectilinear basal outline.
In profile view, the lingual face of the crown is strongly convex, whereas the labial face shows a more rectilinear outline. The root is low and narrower than the crown. It is 'V'shaped in basal view, showing two well-developed branches, which join towards the lingual side (Fig. 7). It also differentiates a lingual protuberance, which is partially covered by the apron of the crown. In many specimens this protuberance is pierced by a foramen, in the middle of the lingual root face. This medio-lingual foramen seems to be connected by a duct with a central foramen, which is located in the middle of the basal face.
Lateral teeth (Fig. 7.4-5). The lateral teeth are larger than the anterior ones and, in contrast to the latter, they are wider than high. They range from 1.4 to 4 mm in height, and from 3.6 to 11.6 mm in width. They are molariform-like teeth, which show a mesio-distally elongated and labio-lingually compressed shape. The crown extends beyond the root at all points in occlusal view, and it is slightly curved, showing an oval to sigmoidal shape. In lingual and profile view, it is cambered to nearly triangular, because of the presence of a keel-like longitudinal crest on its occlusal face The crown is strongly textured by folding of the enameloid.
The longitudinal crest, mesio-distally oriented, runs through the whole occlusal face in a roughly medial position, following the outline of the lingual edge of the crown. From this crest arise numerous short ridges, which intertwine and extend towards both the lingual and labial edges of the crown, thus defining an ornamentation of enameloid ridges and alveoli. Towards its lingual edge, the crown also bears a medio-lingual articular facet (Figs. 2, 7), which is elongated in the mesio-distal direction.
The root is low and narrower than the crown and has a flat to slightly concave basal face. It bears a lingual protuberance pierced by a foramen, which appears to be connected with other two foramina, which are sometimes observed in the labial face of the root.

Heterodonty. Morphological transition between anterior and
lateral teeth is continuous, and can be clearly appreciated in both basal and lingual views. This progressive change is partially illustrated in Figure 7.
In lingual view, while the symphyseal tooth is symmetrical, in more distal anterior teeth the main cusp of the crown is bent distally (Fig. 7.1-3). In the latter, there is also a difference in the size of the lateral cusplets, being smaller the distal one, towards which the main cusp bends. For the first lateral tooth, both the main cusp and the mesial lateral cusplet have significantly reduced their heights, while the distal lateral cusplet is barely noticeable. So, at this point, the tooth is no longer tricuspid but it rather has a keel-like shape ( Fig. 7.4). For lateral teeth in more distal positions, the keel acquires their characteristic triangular to cambered outline ( Fig. 7.5).
In basal view, while the root of symphyseal teeth is symmetrical, with two wings of equal size arranged in 'V' form ( Fig. 7.1), in the more distal anterior teeth the distal wing becomes smaller relative to the mesial wing ( Fig. 7.2-3).
For the first lateral teeth, the distal wing has become much reduced, and the rest of the base has broadened into a more rectangular form (Fig. 7.4). This trend continues in more distal lateral teeth, where the outline of the root gets to resemble that of the crown (Fig. 7.5).

Taphonomic comments on Heterodontus teeth
In contrast to all the teeth found in Pliocene outcrops from Tongoy and Coquimbo, many of the teeth found in the Pleistocene of LCIV exhibit intense abrasion, suggesting that they have undergone a significant reworking. Since the Pleistocene layers rest over fossiliferous Pliocene strata in erosional disconformity, these teeth could well have been Taxonomic inferences Taylor (1972), Reif (1976) and Herman et al. (1993) have studied the dentition of current horn sharks, noticing some variances among species. According to Reif (1976) there are two types of dentition in the living species of Heterodontus: a) the 'francisci-type' of dentition, in which the molariformlike teeth are slender and have a strong longitudinal crest, as in H. francisci (e.g., Reif, 1976: fig. 2;Herman et al., 1993: pl. 23-30) or H. quoyi (e.g., Reif, 1976: fig. 4;Herman et al., 1993: pl. 31-38); and b) the 'portusjacksoni-type' of dentition, with very broad molariform-like teeth that no longer looks like a keel, as in the case of Heterodontus portusjacksoni (Meyer, 1793) (e.g., Reif, 1976: fig. 5;Herman et al., 1993: pl. 3-22). These two types are equivalent to the 'carinate molar' and the 'rounded molar' ones previously proposed by Taylor (1972). Although useful for morphological distinctions, it has not been proven that dentition types of Taylor (1972) and Reif (1976) have a systematic significance, so that the groups derived from these distinctions may not represent monophyletic clades within the genus.
In the case of the lateral teeth from Coquimbo and Tongoy, most of them are slender (Fig. 8) and show a well-developed longitudinal crest, as in teeth of the francisci-type.
They usually show a rather blunt keel. Some authors (e.g., Reif, 1976) have proposed the strength of the keel as an additional criterion that would help to distinguish among different species with francisci-type dentition. This criterion, however, cannot be evaluated straightforwardly on isolated teeth, since the shape and prominence of the keel can vary widely depending on the ontogenetic stage of the individual and the functional position of the tooth in the jaw (e.g., Herman et al., 1993: pl. 23-38).
Regarding the anterior teeth, they are usually tricuspid in adults; except for Heterodontus zebra (Gray, 1831), whose anterior teeth can have 5 cusps; and H. portusjacksoni, whose anterior teeth can be both unicuspid and tricuspid (Taylor, 1972;Reif, 1976;Herman et al., 1993). The anterior teeth described herein are all tricuspid. The lateral cusplets are poorly developed in some specimens, a feature that is observed in H. francisci and differentiates it from H. quoyi (Hermann et al., 1993), but not from other species such as H. mexicanus or even H. japonicus (e.g., Reif, 1976: fig. 25).
In summary, all the traits observed in both lateral and anterior teeth agree well with those observed in Heterodontus francisci, so their attribution to this species is possible. However, we have used an open nomenclature (e.g., Matthews, 1973;Bengtson, 1988;Sigovini et al., 2016), since it is not possible to discard the presence of some other species with francisci-type dentition among these specimens. According to Reif (1976), there are five species showing such a dentition: the three 'American' species (H. francisci, H. quoyi, and H. mexicanus), currently distributed along the eastern Pacific; and two other species, Heterodontus galeatus (Günther, 1870) and Heterodontus ramalheira (Smith, 1949), currently confined to the southwestern Pacific and the Indian Ocean, respectively (Compagno et al., 2005). In addition to the current species, there are two Neogene extinct species that have been described in the Americas: Heterodontus janefirdae Case, 1980 . Relationship between breadth and length of lateral teeth from Coquimbo and Tongoy (regression equation, breadth= 0.34* length + 0.13, r 2 = 0.76). These measurements exclude fragmented teeth. Grow behavior of main crushing teeth of the Portusjacksonitype (breadth= 0.51*length + 0.13, r 2 = 0.96) and the Francisci-type (breadth= 0.30*length + 0.10, r 2 = 0.89) is also shown, as reported by Reif (1976). Lines show the fit of linear regression. Coquimbo and Tongoy teeth show a better correlation with H. francisci than with H. portusjacksoni; however, an analysis of co-variance revealed that still exist significant differences with both of them (intercepts, P < 0.0001 in both cases). specimens reported for the Neogene basins of Western Atlantic (Fig. 1), could support the taxonomic validity of both species, or by the contrary, a relationship with the extant species on the Pacific side.

Late Cenozoic paleobiogeography of Heterodontus in eastern Pacific
The fossil record of Heterodontus in the Americas shows that horn sharks were widespread in the eastern Pacific and western Atlantic during the Neogene ( Fig. 1; Tab. 1). This Neogene distribution, in both the eastern Pacific and western Atlantic, contrasts markedly with the current distribution of the genus in the Americas, now restricted to the Pacific coast from California to Ecuador and Peru (Compagno et al., 2005;Kaschner et al., 2016;Froese and Pauly, 2018;OBIS, 2018). In the southeastern Pacific region, one of the most striking differences between fossil and current records is observed: there is a large region of more than 2,500 km of coastline (between 10°S and 34°S) where no horn sharks have been observed at Recent times (Compagno et al., 2005;Lamilla and Bustamante, 2005;Kaschner et al., 2016;Froese and Pauly, 2018;OBIS, 2018); nevertheless, there have been several paleontological finds of Heterodontus in Neogene sediments of the same area ( Fig. 1; Tab. 1). So far, the genus has already been reported from the early Miocene of Navidad (Suárez and Encinas, 2002;Suárez et al., 2006), the middle to late Miocene of Mejillones  and Caldera (Walsh, 2001;Suárez et al., 2004;Gutstein et al., 2008;Villafaña, 2015), the late Miocene to early Pliocene of Pisco (Muizon and DeVries, 1985;Kindlimann, 1990), the Pliocene of Tongoy and Coquimbo (this work), and the late Pliocene of Horcón (Carrillo-Briceño et al., 2013). These records suggest that horn sharks inhabited the region during most of the Neogene and that, at some stage after; they have been extirpated from that area (Fig. 1).
For the Pleistocene, teeth of Heterodontus have been found in California (Kanakoff, 1956;Fitch, 1966Fitch, , 1968Long, 1993b) and Ecuador (pers. obs. JDCB; ongoing research). In all the extirpation area, there have not been records of the genus from Pleistocene strata so far. So it seems like the last occurrence of Heterodontus in this region is around the Pliocene/Pleistocene boundary. In any event, research on Pleistocene fossil fishes from the southeastern Pacific is still scarce, so that the absence of horn sharks teeth in the Pleistocene could well be a sampling effect, produced by a reduced sample size in the fossil record. Systematic sampling through complete sections, at different Pliocene and Pleistocene locations along the Chilean and Peruvian coast, could contribute to addressing the information gaps still existing in the fish fossil record, thereby improving our understanding of the Heterodontus extirpation pattern.

Environmental controls on Heterodontus extirpation
The extirpation of horn sharks from the southeastern Pacific occurred in the context of a general faunal turnover that has been widely recognized in the transition from Pliocene to Pleistocene in the region (e.g., Philippi, 1887;Möricke, 1896;Herm, 1969;Rivadeneira and Marquet, 2007;Kiel and Nielsen, 2010;Valenzuela-Toro et al., 2013;Villafaña and Rivadeneira, 2014;Rivadeneira and Nielsen, 2017).
Oceanographic, tectono-eustatic and ecological factors have been mentioned as mutually non-exclusive drivers of this turnover: Oceanographic changes. Temperature of sea water is known to affect physiological processes in ectothermic organisms, which may influence their movement and distribution patterns (e.g., Fry and Hart, 1948;Brett, 1971;Di Santo and Bennett, 2011;Johansen and Jones, 2011;Luongo and Lowe, 2018). In the case of horn sharks living in temperate regions, temperature seems to be a major limiting factor: in northeastern Pacific, Heterodontus francisci occurs off central California only in warmer-than-usual years, being otherwise restricted to the southern coast of California; while in southwestern Pacific, H. portusjacksoni is known to conduct long migrations of up to 800 km from the southeastern coast of Australia to Tasmania in summer, and back to the north in winter (Compagno, 2001).
Most of the records of horn sharks worldwide have occurred in regions with SST between 17.9°C and 23.9°C; while for the eastern Pacific species only, they are concentrated between 19°C and 24.2°C Froese and Pauly, 2018).
The subsequent collapse of sea water temperature occurred at times of major global climatic and oceanographic changes (Zachos et al., 2001;Ravelo et al., 2004;Wara et al., 2005;Lawrence et al., 2006), being coeval with a major expansion of upwelling cells in the region (Ibaraki, 1997) and with the onset of modern oceanic conditions during late Pliocene (Le Roux, 2012). This cooling is thought to have exercised a control on the decrease in the diversity of mollusks (Herm, 1969;Covacevich and Frassinetti, 1990;De-Vries, 2001;Rivadeneira and Marquet, 2007) and vertebrates (Cione et al., 2007;Villafaña andRivadeneira, 2014, 2018;Amson et al., 2015) in the region.

Tectonic activity and sea-level changes.
Many regions of the central Andes have experienced significant tectonic activity since late Pliocene (González et al., 2003;Le Roux et al., 2005, 2006Clift and Hartley, 2007). In north-central Chile, facies changes observed in marine sediments reflects that the coastal area began to rise rapidly from 2.6 Ma, leading to the emergence of the platform during the Pleistocene (Le Roux et al., 2005Roux et al., , 2006Roux et al., , 2016 habitats for short periods of time (Luongo and Lowe, 2018).
It appears thus that Heterodontus is able to live on those environments, either making incursions into higher latitudes or into cooler waters brought to low latitudes by the effect of currents. This capacity of horn sharks to perform in cooler waters would explain the fact that, despite their preference for tropical to subtropical environments, some of them have been found in areas with SST even lower than 17° C (Fig. 9).
Then it seems that cooling by itself would not account for the extirpation of Heterodontus, so favoring the idea that tectono-eustatic and ecological drivers must have played an important role as well. Additionally, changes on other oceanographic variables, such as environmental oxygenation or salinity, could have also affect in the paleobiogeographic distribution of the genus. To evaluate this, further information on how these ecological variables evolved from Pliocene to Pleistocene in the region is needed.

CONCLUSIONS
The fossil record of Heterodontus in Chile and Peru (between 10°S and 34°S) suggests that horn sharks inhabited the southeastern Pacific during most of the Neogene, and that they were extirpated from the region around the Pliocene/Pleistocene boundary. Oceanographic changes, such as sea water cooling; reduction of habitats, by tectonic activity and sea-level changes; and ecological feedbacks,