The authors wish to thank Christian de Muizon (Muséum National d’Histoire Naturelle, Paris) for his critique revision of the manuscript and valuable suggestions; Jane Wheeler (Universidad Nacional Mayor de San Marcos, Lima) for her comments about the ontogeny of modern camelids; René Marocco (Institut de recherche pour le développement, Lima) and François Pujos (Instituto Francés de Estudios Andinos, Lima) for their helpful insights; Bruce MacFadden and Richard Hulbert of Florida Museum of Natural History for providing access to the vertebrate paleontology collection. We would like to thank A. Cisneros, J. Tejada and D. Omura. We also thank Jean Vacher (Director - Instituto Francés de Estudios Andinos, Lima) and Niels Valencia (Director - Museo de Historia Natural - UNMSM, Lima) for their continued support of this research.

Introduction

1The evolution of camelids took place principally in North America. Their migration to South America probably occurred during the late Pliocene (Hoffstetter, 1952). The earliest material was assigned tentatively to the Plio-Pleistocene of Argentina (Cabrera, 1935; Kraglievich, 1946; Marshall et al., 1984) and Uruguay (Mones, 1988).

2Descriptions of the genera Palaeolama GERVAIS, 1867, and Hemiauchenia GERVAIS & AMEGHINO, 1880, were based on material collected from the Argentine Pampas. Cabrera (1935) and Hoffstetter (1952) agreed in considering Hemiauchenia as synonymous with Palaeolama. Webb (1974) distinguished Hemiauchenia from Palaeolama as a separate genus and tried both diagnosis. Most recently, Guérin & Faure (1999) designated Hemiauchenia as a subgenus of Palaeolama.

3The best documented and oldest account of P. (Palaeolama) from South America comes from the Ensenadan of Tarija, Bolivia (Webb & Stehli, 1995). In contrast, P. (Hemiauchenia) was present in the upper Uquian, having immigrated during the second panamerican dispersal that occurred between 2,0 and 1,9 Ma (Marshall, 1985). In North America, P. (Hemiauchenia) is recognized from the Hemphillian (lower Pliocene), while P. (Palaeolama) is well represented in the Irvingtonian (lower Pleistocene) in shell banks from Leisey, Florida (Webb & Stehli, 1995).

4There are only two reliable reports of Palaeolama from the coast of Peru, both upper Pleistocene. The first report describes material from deposits of La Brea, near Talara (Lemon & Churcher, 1961), which was assigned by Churcher (1965) and Webb (1974) to P. aequatorialis HOFFSTETTER, 1952. The second refers to material identified by Hoffstetter collected from the Pampa de los Fósiles, La Libertad (Marshall et al., 1984).

5The objective of this paper is to describe material of Palaeolama (sensu Guérin & Faure, 1999) from the Plio-Pleistocene of Peru and discuss its biostratigraphic and biogeographic implications.

6The chronology of mammal ages for South America used in this study is that proposed by McKenna & Bell (1997).

7Institutional abbreviations:

8MUSM: Museo de Historia Natural – Universidad Nacional Mayor de San Marcos, Lima.

9UF: Florida Museum of Natural History – University of Florida, Florida.

1. Geology and age

10The camelid bones were found in an outcrop of conglomerate and bioclastic sandstone exposed in a roadcut along the Panamerican Highway on the northwestern side of Quebrada Infiernillo (SITE 596, Fig. 1). The lower part of the section (6,6 meters), which rests upon a seaward-tilting igneous platform; consists of a prograding sequence that represents shallow subtidal, lower shoreface, upper shoreface, and fluvial/alluvial environments characteristic of a high-energy beach fronting an attenuated flood-prone alluvial fan.

11The middle part of the section (7,0 meters) is a transgressive succession of laminated bioclastic gravel, lenticular bioclastic gravel, and several meters of massive, poorly sorted, coarse-grained bioclastic sandstone with dispersed, transported whole fossils of mollusks, barnacles, and the camelid bones. The massive sands may had been accumulated in a small embayment at the mouth of Quebrada Infiernillo that captured debris washed in from the Pacific Ocean and down from the Andean foothills.

12The upper part of the section includes a regressive sequence of cross-bedded bioclastic cobbly gravel and sandstone (2,2 m) overlain by a transgressive and regressive cycle (>3,2 meters) with similar deposits. A coquina that caps the second regressive sequence also constitutes the surface of the most elevated marine terrace, which stands 250 meters above sea level.

13The oldest mollusks from the Atiquipa outcrop, disarticulated valves of Choromytilus chorus (MOLINA, 1782) and Chlamys vidali (PHILIPPI, 1887), are wedged between boulders at the base of the section. C. vidali is found in lower and upper Pliocene strata of Chile (Herm, 1969), lower Pliocene shoreface deposits near Sacaco (Muizon & DeVries, 1985), and on the surface of the oldest marine terraces of northern and southern Peru (DeVries, 1986). Absent from the base of the Atiquipa section are mollusks that signify a lower Pliocene age, e.g., Acanthina ‘triangularis’ DEVRIES, 1986; Herminespina mirabilis (PHILIPPI, 1887); H. saskiae DEVRIES & VERMEIJ, 1997; Concholepas kieneri HUPE, 1859; and C. nodosa MORICKE, 1896 (Devries & Vermeij, 1997; Devries, 2000).

14The bone-bearing sandstones contain an assemblage of mollusks that includes Concholepas camerata DEVRIES, 2000, which indicates a latest Pliocene or early Pleistocene age. The coquina that caps the section at Atiquipa contains an assemblage of the extant mollusks Glycymeris ovata (BRODERIP, 1832), Eurhomalea lenticularis (SOWERBY, 1835), Mulina edulis (KING, 1831), and Oliva peruviana LAMARCK, 1811. Elsewhere along the coast between Chala and Camaná, these species occur in a comparable geological setting together with the extinct gastropod, Chorus grandis (PHILIPPI, 1887), suggesting a pre-Pleistocene age (Devries, 1997) (Fig. 2).

2. Systematic paleontology

15Order ARTIODACTYLA OWEN, 1848

16Family CAMELIDAE GRAY, 1821

17Subfamily CAMELINAE GRAY, 1821

18Tribe LAMINI WEBB, 1965

19Palaeolama GERVAIS, 1867

20Material – MUSM 51 - Departamento de Paleontología de Vertebrados, Museo de Historia Natural (UNMSM); associated distal portions of the left femur and right humerus without the proximal epiphysis. Collected by Thomas DeVries and Marcelo Stucchi.

21Locality – 15º48’42"S, 74º21’25"W; SITE 596, Atiquipa (Panamerican Highway, km. 596 south), 200-250 meters above sea level, Arequipa Department, Peru.

22Age – Late Pliocene to earliest Pleistocene (2,2 – 1,6 Ma).

2. 1. Description

23Humerus – In the humerus MUSM 51 (Fig. 3 A, B), the lateral epicondyle is more robust than the medial epicondyle, in contrast to that of Vicugna vicugna MOLINA, 1782, where the two are subequal. From a cranial perspective, the articular trochlea is compressed proximodistally. The interlabial longitudinal distance-distal articular transversal distance index in MUSM 51 is 39; in V. vicugna it is 48. The axial throat is deep. In relation to the medial lip, the lateral lip extends proximally more than in V. vicugna and P. weddellii. The epiphyses of the lateral and medial epicondyles are partially fused.

24The preserved part of the MUSM 51 humerus does not exhibit significant differences in shape or proportion from any of the Palaeolama species. Among the southamerican species, it is practically equal in size to the smallest forms of P. weddellii GERVAIS, 1855 of the Pleistocene of Ecuador (Hoffstetter, 1952) and the average of P. paradoxa (GERVAIS & AMEGHINO, 1880). P. major LIAS, 1872 has only been able to be compared in one of its dimensions. The total length of the MUSM 51 humerus is estimated to be 270-300 mm (Table 1).

Table 1 - Comparative dimensions of the humerus of Palaeolama sp. (MUSM 51). TL – total length; PTD – proximal transversal diameter; DTD – distal transversal diameter; DATD – distal articular transversal diameter; MTDD – maximum transversal diameter of the diaphysis (Dimensions in mm)

		TL-var	PDT-var	DTD-var	DATD-var	MTDD-var
Palaeolama sp.	MUSM 51	-	-	63,2	56	30,4
P. (Palaeolama) weddellii	Hoffstetter, 1956 Guerin & Faure, 1999	302-316- 278	81-92 83,5	65,2-79 68,5	54-64,5	- 32,5
P. (Hemiauchenia) paradoxa	Cabrera, 1935 Guerin & Faure, 1999	279-297 270	76-81 74	- 61,5	57-60 -	- 30
P. (Hemiauchenia) major	Winge, 1906	261	79	-	47-54,5	-
P. (Hemiauchenia) niedae	Guerin & Faure, 1999	325-348	79,5-93,5	70,5-82,5	-	36,5-43
P. (Hemiauchenia) Macrocephala Inglis, Florida	UF 179748 UF 45478 UF 179744 UF 45480 UF176912	329 326 - - -	~81 83 - - -	67 66 62,5 67 66	61 61 56 60 60	37 36 33 32 35
Palaeolama sp.** Leisey Shell Pits. FLorida	UF 65323 UF 66487 UF 66490 UF 80400 UF 80493 UF 83661 UF 85018	- 307 304 - - 301 306	- - 81 - - 84 83	68 65 71 65 67 64 67	60 61 60 55 57 61 61	34 36 36 31 30 33 36

** The specimens may belong to either P. (Palaeolama) mirifica or P. (Hemiauchenia) seymourensis.

25Femur – The distal epiphysis of the MUSM 51 femur is robust (Fig. 3 C). The femoral trochlea is symmetrical and perfectly parallel with the axial axis. Although the lateral condyle is only partially preserved, it can be seen to be larger than the medial condyle.

26As mentioned for the humerus, the form and proportions of the femur are similar to those of P. weddellii from Ecuador (Hoffstetter, 1952) and P. paradoxa (Cabrera, 1935). Its estimated maximum transverse diameter is slightly less than the minimum value for P. weddellii (Table 2). The material of P. major of Winge (1906) has proven to be that of a juvenile individual (Hoffstetter, 1952). As evidenced by the dimensions of a femur of P. major from Brazil (Guérin & Faure, 1999), the distal epiphysis does not have the same proportions (Fig. 3).

Table 2 - Comparative dimensions of the femur of Palaeolama sp. (MUSM 51). ML – medial length; LL – lateral length; DTD – distal transversal diameter; PTD – proximal transversal diameter; DAPD – distal antero-posterior diameter (Dimensions in mm)

		ML-var	LL-var	PTD-var	DTD-var	DAPD-var
Palaeolama sp.	MUSM 51	-	-	-	~80	84
P. (Palaeolama) weddellii	Hoffstetter, 1956	369-400,5	358-394	94,5-113	83-92,5	88,5
P. (Hemiauchenia) paradoxa	Cabrera, 1935	~370	-	90	79	-
P. (Hemiauchenia) major	Winge, 1906 Guérin & Faure, 1999	357 384*	- -	86-95 -	79 89	- 88
P. (Hemiauchenia) niedae	Guerin & Faure, 1999	443*	-	101-107	85-92	89,5-94
P. (Hemiauchenia) Macrocephala Inglis, Florida	UF 179749 UF 179750 UF 45279	- - -	- - -	- - -	93 83 83	98 89 87
Palaeolama sp.** Leisey Shell Pits. FLorida	UF 66500 UF 66602 UF 64331 UF 66601 UF 66603 UF 66604 UF 68884	377 - - 373 - - -	~375 - - 369 - - -	100 - - 101 - - -	82 82 ~81 85 ~86 82 86	87 83 90 89 95 86 87

* These measurements correspond to the maximum length of the bone.
** The specimens may belong to either P. (Palaeolama) mirifica or P. (Hemiauchenia) seymourensis

2. 2. Discussion

27Abundant material assigned to Palaeolama and “Auchenia” from Miramar, Argentina, is referred to the Chapadmalalian by Cabrera (1935) and Kraglievich (1946). Nevertheless, Marshall et al. (1983) did not include the family Camelidae among mammals of Holarctic origin assigned to the Chapadmalalian of Argentina. Later, Marshall et al. (1984) characterized the Uquian (lower Pleistocene) by the presence of Camelidae, among other families. The occurrence of Palaeolama (sensu Guérin & Faure, 1999) in upper Pliocene-lower Pleistocene deposits of coastal southern Peru confirms the presence of camelids, particularly of Palaeolama, in the Uquian, as was also noted by Mones (1988) based on evidence of associated fauna from Uruguay. Owing to these records, the Uquian Age (characterized by various first appearances of mammals in Argentina) can be extended, at least with respect to the Camelidae, to Peru and Uruguay. The Peruvian material has an age similar to that of the oldest occurrence of P. (Palaeolama) in North America, referred to the Irvingtonian of Leisey, Florida (Webb & Stehli, 1995). The existence of P. (Hemiauchenia) is pushed back to the Hemphillian (Webb & Stehli, 1995).

28P. (Palaeolama) is distinguished from P. (Hemiauchenia) principally by the proportion and size of its limb bones (Webb, 1974; Webb & Stehli, 1985; Guérin & Faure, 1999). Nevertheless, a morphological analysis of the fossil material does not reveal a distinction within the genus. All Palaeolama species show important intraespecific variation in size and also are indistinguishable in humerus and femur proportions (Table 1 and 2, Figs. 4 and 5). In localities like Leisey Shell Pits, Florida, where P. (Palaeolama) and P. (Hemiauchenia) are present, complete humerus and femur could not be assigned to infra-generic level. However, a morphological analysis of humerus and femur remains from Peru indicate that they are comparable in proportion and size with southamerican P. (Hemiauchenia) paradoxa and the smallest forms of P. (Palaeolama) weddellii. Additionally, this can be extended to the smallest specimens of northamerican P. (Palaeolama) mirifica, P. (Hemiauchenia) seymourensis and P. (Hemiauchenia) macrocephala. Considering that the peruvian material might be from a young adult (see below), the measurements could be slightly less than those for a completely mature individual (Figs. 4 and Fig. 5).

29Other points of distinction between P. (Palaeolama) and P. (Hemiauchenia) are habitat and diet. While P. (Palaeolama) was a browser with robust metapods that lived in montagne regions, P. (Hemiauchenia) had a mixed diet (browsing and grazing) and gracile metapods suitable to living on the open plains (Webb, 1974; Webb & Stehli, 1995). The fossil record of both subgenera, the rugged relief, and the desert character of the Peruvian coast since the middle Miocene (Sébrier et al., 1984; Alpers & Brimhall, 1988) suggest that the Lamini from Atiquipa was a P. (Palaeolama)-like camelid.

30MUSM 51 (portions of femur and humerus) was found associated in the same section of the outcrop. Preceding statement and their relative size suggest that both belong to a single individual. However, the bone fusion secuence does not correspond. In modern camelids, the fusion of the epiphyses of the distal epicondyles of the humerus takes place between an age of 12 and 18 months (Wheeler, 1999). The epiphyses are not completely fused in sample MUSM 51, which suggests it belongs to a young individual. The distal epiphysis of the femur of MUSM 51, however, is completely fused, contradicting the aforementioned suggestion. In modern camelids, the fusion of epiphyses in the femur happens at an age of 42 to 44 months, when the animal is a young adult (Wheeler, 1999). According to Wheeler (personal communication, 2000), exceptional cases do occur in modern camelids in which the fusion of the epiphyses of the distal epicondyles in the humerus never occurs due to certain environmental conditions. In light of these comparisons, we consider MUSM 51 could belong to an individual older than a young adult.

31The nearshore marine sediments which yielded MUSM 51 were deposited near the mouth of a canyon and a mountainous escarpment. We propose that the bones could have been transported by a river from the mountains to the coast. The fragmentary nature of the bones is consistent with high-energy transport. Thus, Palaeolama sp. might have inhabited both the Pacific coast and Andean mountains, as presently does the guanaco, Lama guanicoe (MULLER, 1776) according to Wheeler (1999). Ochsenius (1995) contends that the Lamini did not truly exploit Andean ecosystems prior to the Holocene. The occurrence of MUSM 51 in Plio-Pleistocene strata, as well as recent discoveries of Lama guanicoe in upper Pleistocene levels from the Peruvian coast (Salas & Stucchi, 2002) challenges that claim. We propose that the rarity of Pleistocene camelid material in the region is due to a scarcity of outcrop, rather than a failure of camelids to exploit pre-Holocene Andean environments.