DISCUSSION

Environmental Contexts

The three caves noted above occur in environments that differ in terms of elevation, the surrounding terrain, and the actual physical structure of the caves (Table 1). Jatun Uchco and Trigo Jirka are located within cliffsides, whereas Cueva Roselló is in a flat region. The opening to Trigo Jirka is broad and singular for this simple depression in the cliff, whereas Jatun Uchco has numerous deep and sometime steep passageways. "Boulder chokes" in Jatun Uchco block former, steep openings that may have served as natural traps.

One of the most conspicuous features regarding the remains recovered from Cueva Roselló was the non-stochastic representation of the elements of the skeleton. For example, of the hundreds of bones found, less than 10 were ribs or vertebrae. Distal limb elements far outnumbered proximal elements (e.g., whereas there were several metacarpals of †Onohippidium, no humerus of this animal was found) and forelimbs of vicuña far outnumbered hindlimbs (there were 30 adult Vicugna metacarpals, but only three complete metatarsals). We propose the provisional hypothesis that this skewed representation of skeletal elements in Cueva Roselló occurred via predation selection and concentration (e.g., by the Puma and/or Lycalopex) rather than hydrodynamic sorting. This predator selection hypothesis is weakly supported by tooth marks on some of the specimens, but these are few and generally ambiguous (e.g., the puncture in the ascending ramus of the Onohippidium [MUSM 613] jaw seen in Figure 8).

Elevation, terrain, and physical structures of the caves likely influenced the composition of the fauna preserved within. Some of these factors are considered in our discussions of the faunal components below.

Fauna

Xenarthra. Sloth remains were found at all three of these new cave localities (Table 1). †Diabolotherium nordenskioldi was the only sloth present at Jatun Uchco, but fragmentary remains of larger sloths (†Megatherium and a †mylodontid) were recovered at Cueva Roselló and Trigo Jirka. †Diabolotherium was present at all three localities, but abundant material of this peculiar beast was recovered at Jatun Uchco. Unfortunately, we were unable to obtain collagen from †Diabolotherium specimens of Jatun Uchco (or any other samples attempted from that locality), but organics were recovered from a Trigo Jirka specimen of †Diabolotherium.

Kraglievich (1926) originally placed the species now known as †Diabolotherium nordenskioldi in †Nothropus, but a recent analysis by Pujos et al. (2007) showed that it is quite distinct from †Nothropus species and thus assigned it a new generic name. Morphologically, †Diabolotherium is best described as a mosaic of primitive and derived features. Pujos et al. (2007) noted characteristics similar to Santacrucian forms (e.g., †Hapalops), †megatheriids, †megalonychids, and extant tree sloths (Bradypus and Choloepus). Based upon their phylogenetic analysis, they referred it to the †Megalonychidae.

Preliminary findings from the McMaster Ancient DNA Centre, based upon MUSM 1678 (Figure 4.3), suggest that †Diabolotherium is indeed a megalonychid. This Trigo Jirka specimen provided short nuclear amplifications for the PLCB4 gene (94 base pairs) and the CREM gene (120 base pairs). These showed no variation from the two genera of mid-Holocene Greater Antillean sloths under study, but differed from those genes of Choloepus, suggesting that †Diabolotherium was more closely related to the extinct, West Indian megalonychid taxa than to the extant, two-toed Choloepus. Repeat sampling is in process to rule out the possibility of contamination.

No cingulate xenarthrans were found in any of the three caves. Considering that a single armadillo specimen could have produced numerous osteoderms, in addition to the usual skeletal elements, we suspect that the absence of evidence, in this case, may actually represent evidence of absence.

Chiroptera. Bats were recovered only from Jatun Uchco. These were commonly encountered and included specimens of the vampire, Desmodus sp., a fruit bat, Anoura sp., and the nectarivorous, Platalina genovensium. A few bats currently reside in Jatun Uchco, so it is possible that many of the bones are Recent, but we note finding some bat specimens beneath bones of extinct taxa (e.g., †Smilodon), suggesting that some of the bats are Pleistocene in age.

Although Cueva Roselló yielded a large sample of mammals, no bats were recovered. When one considers the elevation of Cueva Roselló (3,875 m [12,713 ft]) and the fact that bat diversity declines as elevation increases and are scarce over 3,200 meters (Graham 1983), the absence of bats is not surprising. Bats were also absent at the lower elevation Trigo Jirka, however. This may be a reflection of the small sample of fossils recovered to date from Trigo Jirka, but we also failed to detect any evidence of living bats at that shallow cave.

Carnivora. Carnivorans were commonly encountered at Jatun Uchco and Cueva Roselló. Both localities yielded several specimens of Puma (c.f. P. concolor) and Andean foxes (Lycalopex sp.). Two felids discovered at Jatun Uchco are not known from Cueva Roselló. These included †Smilodon populator and a partial jaw of another large felid that is not referable to any taxon that we have compared it to (e.g., Puma, †Smilodon, jaguar [Panthera onca]), but may be an extinct species related to jaguar (Martin, personal commun., 2006).

Work in progress suggests that the Lycalopex specimens from Jatun Uchco and Cueva Roselló are close to, but morphometrically distinct from, the extant zorro Andino (Lycalopex culpaeus) and from the much larger extinct fox †Lycalopex peruanus from Casa del Diablo (southern Perú).

Both Jatun Uchco and Cueva Roselló had a single specimen of a "hognose" skunk, Conepatus sp. Otherwise, mustelids were not recorded at any of our localities.

Ungulates. Artiodactyls and perissodactyls were only recovered at Cueva Roselló. Their absence at Jatun Uchco and Trigo Jirka is no surprise when one considers the terrain surrounding the cave entrances there. Few ungulates, especially larger bodied forms, would have been capable of climbing such cliffs and the elevations of the entrances would have made it difficult for carnivores to carry prey items to such heights along steep rock surfaces. Cueva Roselló, however, is located on the Altiplano, the high plains between the eastern and western ranges of the Andes. Such relatively flat terrain would have allowed ungulates close proximity to the cave. As previously mentioned, we think that the ungulate remains were carried into the cave by predators, either puma, fox, or both.

The smallest of the ungulates was a diminutive deer, assignable to cf. Pudu sp. or possibly a very small species of Mazama. Fragmentary remains of a larger deer, apparently Hippocamelus, also were found.

The most common remains of any mammal at Cueva Roselló were those of vicuña (Vicugna sp.) (see Figure 7). Previously, it had been thought that vicuña did not migrate into high elevation regions of the Andes until less than 12,000 years ago (Wheeler et al. 1976; Hoffstetter 1986; Marín et al. 2007). However, our calibrated radiocarbon age of 22,220 ± 130 BP from bone sample of vicuña along with that of 23,340 ± 120 BP from a nearby (coeval or nearly so) †Onohippidum devillei specimen strongly suggest that vicuña migrated to the central Andes about 10,000 years earlier than previously thought.

In addition to providing organic molecules for radiocarbon dating (collagen), the †Onohippidum devillei specimens of Cueva Roselló also provided aDNA for a molecular phylogenetic study. Morphologically, the †Onohippidium of Cueva Roselló is indistinguishable from †O. devillei from Tarija, but it is distinct from the South American horses, †Hippidion principale, †H. saldiasi, and †Equus insulatus (see MacFadden 1997). How closely related †Onohippidium is to †Hippidion, and whether species of †Onohippidium should be included within the priority genus †Hippidion, is a matter of dispute (see MacFadden 1997; Alberdi and Prado 1998). Morphological studies have not resolved this debate, but a recently completed molecular study (Orlando et al. in press) is providing more data for the analyses and discussion.

High Altitude Caves as a Refugium for Ancient DNA

DNA decay. Heat, humidity, and intense solar radiation are physical features typically associated with low latitudes. These factors contribute to the degradation of DNA (Austin et al. 1997; Wayne et al. 1999). Thus, ancient DNA (aDNA) studies at tropical latitudes have been seriously constrained. In their recent survey regarding the survival of aDNA, Karanth and Yoder (in press) found a positive relationship between latitude and successful aDNA studies. The more removed the locality is from the equator the more likely aDNA could be recovered. Aside from one study of domestic dogs less than 1,500 years old (Leonard et al. 2002), no aDNA had been successfully recovered from bone specimens from localities less than 20° from the equator. The lowest latitude from which aDNA had been recovered from a Pleistocene taxon was 36° (Poulakakis et al. 2002). The results of the Karanth and Yoder study suggested that aDNA could not be obtained from samples collected between 18° North and 16° South. They did, however, provide a caveat for their somewhat pessimistic conclusion—high altitude caves.

Andean Caves: cool, dry, and shielding. Although heat, humidity, and intense solar radiation are associated with tropical latitudes, these physical factors may be limited by local conditions. For example, Karanth and Yoder (in press) noted that aDNA was successful recovered from a 3,300 ± 60 BP sea eagle of Hawaii at about 20° North latitude that had been recovered from a cave that was at an elevation of 1,463 m (Fleischer et al. 2000). They speculated that the relatively cool temperatures at the high altitude and the blockage of UV radiation by the eagle's presence within the cave facilitated the aDNA preservation.

At greater elevations, caves of the Andes provide opportunities for more extreme preservation. If indeed our radiocarbon age (over 29,000 BP) of the †Diabolotherium claw of Trigo Jirka is close to its actual age, then the preservation of its organic compounds (keratin, collagen, and DNA) from a specimen that old and from such a low latitude (10° South) is remarkable. Greater potential lies with the Cueva Roselló fossils. These are from a higher elevation, in a more arid region, and it includes a greater number of taxa. Extra protection against the elements may have been afforded by the fact that the cave had been naturally sealed until its accidental discovery during mining operations.