U-series dating and taphonomy of Quaternary vertebrates from Brazilian caves

Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508 – 522 www.elsevier.com/locate/palaeo

U-series dating and taphonomy of Quaternary vertebrates from Brazilian caves Augusto S. Auler a,⁎, Luís B. Piló b , Peter L. Smart c , Xianfeng Wang d , Dirk Hoffmann c , David A. Richards c , R. Lawrence Edwards d , Walter A. Neves b , Hai Cheng d a

CPMTC – Instituto de Geociências, Universidade Federal de Minas Gerais, Av. Antonio Carlos 6627, Belo Horizonte, MG. 31270-901, Brazil b Laboratório de Estudos Evolutivos Humanos, Universidade de São Paulo, Rua do Matão 227, São Paulo, SP, Brazil c School of Geographical Sciences, University of Bristol, Bristol BS8 1SS, England d Department of Geology and Geophysics, University of Minnesota, MN 55455, USA Received 4 September 2004; received in revised form 26 February 2006; accepted 5 March 2006

Abstract The geochronology and taphonomy of internationally important fossil bearing cave deposits were studied, both in the semi-arid Northern Bahia area and the subtropical southeastern Lagoa Santa area of Brazil. Taphonomic analysis suggests that the processes responsible for bone accumulation in the Brazilian caves vary between sites, and taphonomic bias can therefore be significant in causing differences in faunal composition. In the Toca da Boa Vista caves the presence of single articulated skeletons, and the entrance-related distribution indicate that random penetration of animals is the main mechanism of fossil accumulation, a process that biases the assemblage to smaller species, and takes place over extended time periods. In nearby Toca dos Ossos cave transport by runoff in the cave river is predominant, and biases the fauna remains to larger more robust bones and species. Deposition probably also occurred only at times of enhanced runoff giving a more contemporaneous assemblage. Similar processes were responsible for emplacement of the copious fossil remains in the more humid Lagoa Santa area, where terrigenous fossil deposits are found intercalated by massive speleothem calcite layers. In this area runoff under a drier climate probably accounts for the sediment emplacement inside caves. In both areas the mode of emplacement implies bias in the fossil record, resulting in fossil assemblages that do not mirror surface faunas, limiting palaeoenvironmental reconstruction. Mass spectrometric U-series analysis of speleothem calcite overlaying fossil remains gives minimum ages for fossil deposition. These ages confirm the previous view that many of the deposits derive from the late glacial, but also show that much older material (some > 350,000 yr) is also present. The habitat requirements of critical fossil species such as bats and monkeys strongly suggest that they derive from much wetter periods when forest cover was present in the currently semi-arid Northern Bahia area. Taphonomy exerts a major control on the diversity and mode of emplacement of cave fossil deposits in eastern Brazil and thus detailed sedimentological and hydrological studies coupled with a sound geochronological approach are essential in quantifying the relative importance of each taphonomic processes before faunal and palaeoecological interpretations can be attempted. © 2006 Elsevier B.V. All rights reserved. Keywords: Late Quaternary; Speleothems; Taphonomy; Palaeontology; Palaeoecology

⁎ Corresponding author. Present address: Instituto do Carste, Rua Kepler, 385/04, Belo Horizonte, MG 30360-240, Brazil. Tel.: +55 31 9314 1570. E-mail address: [email protected] (A.S. Auler). 0031-0182/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.palaeo.2006.03.002

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

1. Introduction Since the pioneering studies by Lund (1840) in early 19th century, cave deposits have provided a remarkable number of well-preserved fossil remains, which have formed the basis for Pleistocene vertebrate palaeontological research in Brazil (Paula Couto, 1953). However, despite the significance of cave-derived fossil

509

remains, stratigraphic control has often been poor, and there has been limited research on both the geochronology and taphonomy of the deposits. The failure to establish a regional cave-derived chronology in Brazil has led researchers to seek a correlation with similar but much better chronologically constrained faunas in Argentina (Paula Couto, 1975), and a generalised “late Quaternary” age has been assigned for the Brazilian

Table 1 Reported ages (uncalibrated and calibrated radiocarbon, ESR and alpha spectrometric U-series) of fossil remains in Brazilian caves Cave site, locality

Method

Age (yr BP)

Age (cal yr BP)

Bahia Toca da Boa Vista a Gruta dos Brejões a

AMS 14C in bone AMS 14C in bone

20,060 ± 290 12,200 ± 120

>22,850 D 15,450 (41.2%) 14,550 D 14,450 (54.2%) 13,750

Moormops Coprolite of Nothrotherium

Piaui Toca do Garrincho b

14

10,020 ± 290

12,850 (95.4%) 10,650 MI

Toca do Serrote do Artur c

14

6890 ± 60 8490 ± 120

Hippidium, Palaeolama, Pampatherium, Toxodon, Catonyx Dasypus, Propaopus, Hoplophorus, Glyptodon, Conepatus, Panthera, Dicotyles, Tayassu, Palaeolama and Mazama

Minas Gerais Gruta do Baú, Lagoa Santa d Lapa Vermelha, Lagoa Santa e Lapa da Escrivânia 5, Lagoa Santa f Lapa da Escrivânia 5, Lagoa Santa f Lapa da Escrivânia 5, Lagoa Santa f Lapa da Escrivânia 5, Lagoa Santa f Lapa dos Tatus, Lagoa Santa f Lapa dos Tatus, Lagoa Santa f Gruta Cuvieri, Lagoa Santa f Lapa da Escrivânia 5, Lagoa Santa f Mato Grosso Santa Elina Rock Shelter g Sao Paulo Abismo Ponta de Flecha h

C in charcoal

C in charcoal

7840 (94.3%) 7610 9900 (95.4%) 9100

Calcite 230Th/234U 77,700 ± 6100

Correlation Fossil mammal

MI MI and C

MI

14

C in charcoal AMS 14C in bone AMS 14C in bone AMS 14C in bone AMS 14C in bone AMS 14C in bone AMS 14C in bone AMS 14C in bone AMS 14C in bone

9580 ± 200 16,900 ± 70 16,550 ± 60 16,250 ± 60 16,180 ± 70 14,030 ± 50 13,920 ± 50 9960 ± 40 9130 ± 150

11,550 (95.4%) 10,150 20,850 (95.4%) 19,450 20,450 (95.4%) 19,050 20,050 (95.4%) 18,750 19,950 (95.4%) 18,650 17,350 (95.4%) 16,350 17,200 (95.4%) 16,200 11,570 (95.4%) 11,220 10,700 (95.4%) 9750

14

10,120 ± 60

12,150 (94.2%) 11,300 C

C in charcoal

ESR in bone

6700 ± 1300 i 5000 ± 1600 j

C D D D D D D D D

D

Hoplophorus euphractus and Pampatherium humboldti Catonyx cuvieri Equus neogaeus Equinae (sp. indet.) Equus neogaeus Equus neogaeus Catonyx cuvieri Catonyx cuvieri Catonyx cuvieri Smilodon populator

Glossotherium aff. G. lettsomi

Toxodon platensis

Bracketed figures are probabilities that calibrated radiocarbon dates lie within the quoted range. D — Direct determination on fossil; C — Correlative age (charcoal in same stratigraphical layer); MI — Minimum age (dating of overlying material). Doubtful determinations such as radiocarbon ages on speleothem calcite and 230Th/234U analysis of fossil bone are not included. a Czaplewski and Cartelle (1998). b Guérin et al. (1996) and Peyre et al. (1998). c Faure et al. (1999). d Piló (1998). e Laming-Emperaire et al. (1975) and Prous (2002). f Neves and Piló (2003). g Vialou et al. (1995). h Baffa et al. (2000). i Dentine. j Enamel.

510

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

deposits as a whole. Cartelle (1995) interprets the vertebrate fossils in Brazilian caves as being chronologically synchronous. He suggests they were deposited at the transition between the Pleistocene and the Holocene. Doubts about the contemporaneity and age of the deposits have, however, been expressed by other workers (Paula Couto, 1975; Marshall et al., 1984), and are supported by the limited radiometric ages available, which range from the Holocene to the last glacial maximum (Table 1). Although caves in many distinct rock types are known to occur in Brazil, fossil remains have been reported only from carbonate (limestone or dolomite) caves (Fig. 1), the majority of which are developed in Precambrian carbonates (Auler and Farrant, 1996). The caves are however, very much younger geomorphological features. Given the limited thickness of carbonate rocks in Brazil (and thus the shallow nature of most caves), and considering current long-term denudation rates of 30 ± 10 m/Ma (Harman et al., 1998) in the cratonic low relief terrain in which most of the karst is located, it is unlikely that any presently existing cave

would survive in the landscape for more than a few million years before being bisected and removed by erosion. In fact, geomorphological studies in some of the major caves in the country have demonstrated that they are probably late Tertiary to Quaternary in age (Karmann, 1994; Auler et al., 2002). Given the age of the caves (and thus the age of the ubiquitous allogenic sedimentation inside them), there is potential for a considerable chronological range for the fossil deposits. It is generally recognised that consideration of the taphonomy of a deposit is an integral part of any modern palaeontological excavation, particularly for palaeoenvironmental interpretation. Surprisingly little is known about the processes of bone accumulation in caves (Andrews, 1990). Based on his extensive study of cave fossil deposits from the Lagoa Santa area, southeastern Brazil, Lund (1845a) believed that fossil bones could have been introduced into Brazilian caves in five different ways: (1) by predators who used the caves as dens; (2) by the fall of animals into vertical fissures or shafts; (3) by death of animals entering caves in search of water or temporary shelter; (4) by death of animals

Fig. 1. Distribution of carbonate karst regions in Brazil showing the location of the Northern Bahia and Lagoa Santa study sites. Areas with reported fossil finds in caves are indicated by a black dot. Triangles indicate small karst areas.

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

that lived in the caves; (5) by water transport during runoff into the caves. Similar processes are recognised by Simms (1994), although he adopts a different classification of cave deposits. Lyman (1994) suggests that four basic post-depositional taphonomic processes may subsequently affect animal bones: disarticulation, dispersal, fossilisation, and mechanical modification. Together these can produce changes in the relative proportion of bones present, alter the relative position of the bones, and affect bone integrity. Thus it is often possible from the sedimentology, distribution, degree of completeness and type of modification of the bones to identify the taphonomic processes. This is critical because the representativeness of any given fossil deposit in terms of the faunal community from which it derived will vary depending on the taphonomic history of both individual bones and the fossil assemblage as a whole. Understanding these effects is essential if reliable palaeoenvironmental information is to be obtained from fossil mammal assemblages in caves. This interpretation is however further complicated by the varied types of sedimentation that occur in caves (see Renault, 1968; Gillieson, 1996), which make cave stratigraphy highly complex, as demonstrated by detailed excavations in both Europe (Campy and Chaline, 1993) and Africa (Brain, 1995). In this paper we present the first systematic taphonomic and geochronological study of Brazilian fossil cave deposits, focusing on two areas of eastern Brazil, Northern Bahia and Lagoa Santa, both of which are well known for their remarkably rich fossil deposits. 2. Nature and taphonomy of the fossil deposits 2.1. Northern Bahia area The Northern Bahia area in northeastern Brazil (Fig. 1) is presently semi-arid with annual precipitation generally less than 500 mm and a predominantly xerophytic scrub, caatinga vegetation. In this area fossil remains from three separate types of cave sites which have very different taphonomies are considered, Toca da Boa Vista and associated caves (henceforth Toca da Boa Vista caves), Toca dos Ossos, and Toca das Onças. 2.1.1. Toca da Boa Vista caves Toca da Boa Vista and Toca da Barriguda, the two most extensive caves in the area (> 130 km in combined length), have yielded remarkably complete fossil vertebrate remains (Cartelle, 1995; Hartwig and Cartelle, 1996; Cartelle and Hartwig, 1996). These caves are very complex maze systems developed in Precam-

511

brian Una Group dolomites. They are largely decoupled from any present day route of water or sediment influx, the exception being the roof collapses, which currently provide the limited entrances to the system. Passage size is usually small (mostly less than 5 m high and wide), with very irregular cross-sections. Except for limited runoff in the immediate vicinity of entrances most cave passages appear to have been dry for extended periods of time, there is for instance no evidence of erosion of sediments nor of development of graded stream profiles along irregular passage floors. Dating of powdery bat guano has demonstrated that some passages have been dry since at least 44,000 yr. However, some lower level passages contain subaqueous speleothems related to previous high water table events, but at present access to water in the caves is limited to a few sites. The two caves and their taphonomic processes appear to be very similar, and the fossil remains from them are therefore discussed together. A list of fossils found in these caves has been provided by Cartelle (1995, 1999) and Czaplewski and Cartelle (1998) (Table 2). The bat fauna is particularly rich with 20 species recognised, none of which is extinct. In contrast 43% of the other 14 mammal species found in the Toca da Boa Vista caves are extinct. Of specific interest are two large fossil monkeys, and an extinct dog and bear. Also note the absence of any large species, extinct or modern. The fossil remains are usually found as complete and articulated skeletons lying on the cave floor. There is either a total absence or very limited clastic sediment cover, although speleothem calcite has precipitated over some of the bones. The high degree of articulation and the completeness of individual remains indicate both that post-depositional disturbance has been minimal at the majority of the sites, and that the animals entered the caves either whilst alive or very soon after death. Limited bone disturbance appears to occur mostly in sites located in the lower levels where fluctuation of former lakes could alter original bone position. In contrast, decomposition, and hence disarticulation, tends to be rapid in aquatic environments, and any water transport would have caused hydrodynamic sorting of the material, with smaller bones being dispersed (Behrensmeyer, 1991). Note that the bats are treated separately below because they inhabit caves as part of their life cycle, thus their mode of transport into the cave is different from that of the other mammals. We interpret the majority of fossil individuals found at Toca da Boa Vista caves as having entered the cave by their own means (autopod mechanism of Vrba (1976) or

512

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

Table 2 List of vertebrates found in Toca dos Ossos (TO) and the Toca da Boa Vista caves (TBV) Species Bats Mormoops megalophylla Pteronotus parnellii Pteronotus davyi Micronycteris Tonatia bidens Mimon bennettii Phyllostomus discolor Chrotopterus auritus Lonchophylla mordax Carollia perspicillata Platyrrhinus Artibeus jamaicensis Diphylla ecaudata Desmodus rotundus Desmodus draculae Natalus Furipterus horrens Myotis Eptesicus fuscus Tadarida brasiliensis Mammals Alouatta sp. ⁎Protopithecus brasiliensis ⁎Eremotherium laurillardi ⁎Glossotherium aff. G. lettsomi ⁎Catonyx cuvieri ⁎Ocnotherium giganteum ⁎Mylodonopsis ibseni ⁎Nothrotherium maquinense Myrmecophaga tridactyla Tamandua tetradactyla Tolypeutes tricinctus ⁎Glyptodon clavipes ⁎Pampatherium humboldtii Coendou sp. Hydrochoerus hydrochaeris ⁎Neochoerus sulcidens Myocastor coypus ⁎Protocyon troglodytes ⁎Smilodon populator Puma concolor ⁎Arctotherium brasiliense ⁎Toxodon platensis

Animal type

Body mass

Bat

Cave

TO

S S

TO

Giant ground sloth Ground sloth

L

TO

L

TO

Ground sloth Ground sloth

M L

Ground sloth Ground sloth

L M

TBV

TO TO

Anteater

S

TBV

TO

Anteater Armadillo Giant armadillo Giant armadillo

S S L M

Porcupine Capybara

S S

Giant capybara Rodent Dog Sabre tooth tiger Puma Small bear

M S S L M M

Hippopotamuslike animal

L

TBV

TO TO

TO TO TO TO TBV

TO TO TO TO

TBV TBV TBV TBV

Body mass

⁎Trigonodops lopesi

Hippopotamuslike animal Mastodon Horse Horse Wild pig Llama Llama Deer Deer

L L L L S M L M S

Cave TO

TBV TBV TBV

TBV

TO TO TO TO TO TO TO

Approximate body mass: S (small): 0–50 kg; M (medium): 30– 200 kg; L (large): >100 kg. ⁎Indicates an extinct species. Based on list of species from Cartelle (1995, 1999) and Czaplewski and Cartelle (1998).

Monkey Large monkey

TBV

Animal type

⁎Haplomastodon waringi ⁎Equus neogaeus ⁎Hippidion principale Tayassu tajacu Lama guanicoe ⁎Palaeolama major Odocoileus virginianus Mazama gouazoubira

TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV TBV

Table 2 (continued) Species

TO

TO

proprio motu of Cartelle (1995)) in search of a cool atmosphere or water. They subsequently died in the cave. This suggestion is supported by the observation that finds are to some extent associated with passages that had in the past contained water pools, but the majority of remains are of individuals distributed at random within the vicinity of entrances (Fig. 2). In contrast, pitfalls down vertical passages inside the cave or from the surface, would result in fewer sites with much more concentrated bone accumulation. The relatively small size of all fossils found at the Toca da Boa Vista caves, and the absence of larger animals commonly found in Pleistocene fossil deposits elsewhere in Brazil (Paula Couto, 1979) also gives support to the autopod mechanism of fossil emplacement, since runoff would undoubtedly have carried larger bones into the cave. It may be that larger animals were unable to penetrate far underground in the relatively small passages. According to Sutcliffe (1985) a great variety of species will go underground of their own free will. Recently, a small group of Conepatus (Mustelidae) has been observed living in a passage over 200 m from the entrance of Toca da Barriguda. In fact, a number of recent non-fossilised dead individuals of other species including anteaters and opossums have also been found in dry passages in Toca da Boa Vista caves. These animals were clearly not transported by water as the passages are very dry, and show no effects of runoff, other post-depositional disturbance or sediment cover. They are at present mummified due to the cave dry atmosphere, and their bones have not been dispersed. This strongly suggests that the Toca da Boa Vista fossil deposits are still actively accumulating and may span a considerable time period.

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

513

Fig. 2. Plan map showing location of fossil deposits at Toca da Barriguda (A) and Toca da Boa Vista (B). Dots indicate vertebrate fossil accumulation. Shaded areas indicate fossil bat guano.

2.1.2. Bat remains in Toca da Boa Vista caves Toca da Boa Vista and Toca da Barriguda contain very extensive deposits of bat bones and guano, suggesting either a very large bat population in the past, a long history of bat occupation, or both. Fossil guano occurs as a powdery deposit up to 1 m thick covering the floor of several passages. No layering is visible in most of the deposits. Bat bones were not observed in direct association with the guano, probably because they would have been consumed by the guanophile fauna or guano associated chemical weathering processes. However, separate accumulations of thousands of intermixed bat bones are common. These are sometimes covered by later speleothem. These deposits are generally close to the major deposits of bat guano, and both are close to present day or past cave entrances (Fig. 2). Lesser deposits or single articulated skeletons have also been observed. Czaplewski and Cartelle (1998) identified at least twenty different species in these deposits. The fossil deposit they studied was associated with cave rafts in a passage that had been flooded in the past, leading these authors to suppose that the flooding event could have trapped the colony, starving them to death. Although this hypothesis could explain this particular deposit, the majority of bat bone

accumulations are found in passages well above the water table that have not experienced water flow for a significant period of time. Bat bone deposits, and guano, are likely to represent in situ accumulations, with minimal transport after deposition. 2.1.3. Toca dos Ossos An interesting contrast is provided by fossil mammals found at Toca dos Ossos, a long and largely unmapped cave developed in the much younger (possibly late Tertiary) Caatinga limestone near the town of Ourolândia, about 50 km away from the Toca da Boa Vista caves, but under the same regional climate and vegetation. The cave comprises a major stream passage with an adjacent floodwater maze of spongework pattern. Passage size varies between the main 10 m high and wide stream passage and spongework conduits too small to enter. A dry upper level is present in part of the cave. At present, active water flow occurs in this cave only during high intensity rainfall events, when a high-energy stream enters the cave through an upstream sink. A total of 19 species are recorded from Toca dos Ossos (Table 2), 63% of which are extinct species. Surprisingly only 5 species are known from both Toca

514

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

da Boa Vista caves and Toca dos Ossos, despite their being in the same region. Only one species of bat is reported from Toca dos Ossos, but with respect to surface species, Toca dos Ossos is richer than the Toca da Boa Vista caves with armadillos, capybaras, hippolike mammals and mastodon only known from Toca dos Ossos. Many of the finds unique to Toca dos Ossos are large species with a body weight in excess of 100 kg. In fact abundant remains of this Pleistocene megafauna were found, Cartelle (1992) reporting that the remains of at least 36 Eremotherium (giant ground sloth) were present in the cave. The lack of complete articulated skeletons, the association of the fossil remains with recently deposited fluvial sediments in the flood prone lower passages, and the lack of concentration of deposits at specific pitfalls all confirm that in Toca dos Ossos, the fossils were undoubtedly emplaced by runoff from the stream sink. Hydrodynamic sorting either during transport into the cave or due to later flooding events may be responsible for removal of the smaller bones, the heavier bones remaining as a lag deposit. Alternatively, the larger size of the passage and transport capacity of the stream may explain the presence of fossils of larger mammals than at Toca da Boa Vista caves. Given this process of emplacement, it is possible that there is some palaeoclimate control of the fossil deposits, they being derived when precipitation was relatively high. Thus unlike the Toca da Boa Vista cave faunas, they may have been emplaced in one (or more) wet intervals rather than over an extended time period. Thus, both in terms of their age and the surface fauna represented, the Toca da Boa Vista and Toca dos Ossos faunas might be expected to be different. Direct intercomparison of the faunas with associated palaeoenvironmental interpretation is therefore inadvisable. 2.1.4. Toca das Onças Toca das Onças is a small cave near the village of Caatinga do Moura, located c. 100 km to the south of the Toca da Boa Vista. The cave comprises a single dry chamber that can only be entered through vertical or steeply inclined entrances. Much of the chamber is naturally illuminated, although the distal portions lie in twilight. This cave was excavated in 1979 and 1980, but unfortunately no stratigraphical or taphonomical data were recorded. Toca das Onças is a remarkably rich palaeontological site and has yielded two complete skeletons of the giant ground sloth Eremotherium laurillardi and fragments belonging to at least 13 other individuals, together with bones from other smaller species (Cartelle and Bohorquez, 1982). These

authors have hypothesised that the inner chamber could have contained water in the past, leading many animals to climb down into the cave. We believe, however, given the near vertical nature of the entrance, that the animals are more likely to have fallen accidentally into the cave. Some post mortem reworking of bones inside the cave also occurred, yielding non-articulated and fragmentary bone remains, possibly by runoff from the entrances or by the activity of smaller carnivores who were capable of negotiating the steep entrance climbs in search of carrion. 2.2. Lagoa Santa area The Lagoa Santa area (Fig. 1) is the birthplace of Brazilian palaeontology, and research in the local caves has been carried out since the early 19th century (Lund, 1840; Liais, 1872). Over 500 caves are known in the Precambrian limestone of the Bambuí Group. The caves are usually short and dry, with meandering passages. Cave entrances are usually at the base of cliffs or dolines, and are subject to seasonal water inflow due to runoff. Extensive sedimentation occurs in the caves, with thick clastic deposits interbedded with speleothem layers. At some sites, later removal of the clastic sediment has left residual speleothem layers attached to the walls above the cave floor. Climate in the area is at present subtropical, with annual precipitation around 1400 mm. Precipitation is highly seasonal with over 80% of the rainfall occurring in a few months during the wet summer/autumn season. The long history of palaeontological excavations of the Lagoa Santa caves has not been recently summarised. Partial lists of fossils found at these caves can be found in a series of memoirs by Winge (1888–1915), Walter (1948), Paula Couto (1970), and especially Cartelle (1995, 1999), but they tend to include only determinations based on material excavated by Lund, with emphasis on extinct species. Cartelle (1999) lists 121 species from Minas Gerais state (mostly from Lagoa Santa), 47 of which are extinct, compared to only 61 species (24 extinct) from all the sites in Bahia (including Pesqueira near the Atlantic coast, a site not discussed here). The Lagoa Santa list includes many of the large species found in Toca dos Ossos. Of the 22 extinct ‘autochthonous’ species listed by Cartelle (1999, Table 4.4), 15 are present at Lagoa Santa, and 13 in the Northern Bahia caves. The intensive excavation that has taken place in the Lagoa Santa caves by saltpetre miners and early palaeontologists makes the recognition of the original stratigraphy difficult. Our observations at a limited number of well-preserved sites suggest that

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

515

(speleothem precipitation, sediment input and sediment removal) are commonly observed to occur repeatedly in many of the Lagoa Santa caves, resulting in a complex stratigraphy that can vary from cave to cave and even within the same cave. As originally suggested by Lund (1845a), the majority of the fossiliferous sediments from the Lagoa Santa area were undoubtedly brought into the caves by runoff, either from the doline slopes associated with cave entrances, or in streams from the swallets. Concentration of bones in the terrigenous matrix varies greatly, being regulated by the rate of bone inflow relative to that of sediment (Brain, 1981). Multiple hydrological events (causing either erosion or deposition of sediment) are likely to cause sediment reworking, a fact noticed since early times (Lund, 1845a). Bone accumulation at the base of internal pits can be a result of random penetration and fall of animals, although runoff also appears to play a role in carrying bones into funnel shaped depressions. The number and diversity of speleothem layers present in these caves suggest the occurrence of more than one episode of speleothem precipitation. Multiple episodes of sediment input and removal are also suggested by lack of stratigraphic correlation between deposits in different caves, and by

fossils are usually found intermixed in coarse cave breccias with a clayey matrix. There are very big differences in the content of bone from place to place along the cave passages and no clear concentration near entrances. Small bones and small bone fragments usually predominate. At a few sites, such as Gruta Cuvieri, bone accumulation at the base of vertical pits has occurred, and here articulated skeletons have been reported. A conspicuous feature of the Lagoa Santa deposits is the presence of layers of speleothem calcite overlying (and sometimes underlying) the deposits. The deposition of speleothem and the emplacement of the clastic sediment are associated with different hydrological conditions, the former requiring a wet and wellvegetated environment at the surface in order to prompt infiltration to the cave, and the latter probably indicating drier conditions with limited vegetation cover favouring runoff, soil erosion and sediment transport during occasional intense rainfall events. Similar cave stratigraphies have been described in Africa (Brook et al., 1997) and Australia (Moriarty et al., 2000). Furthermore, the common presence of “hanging” speleothem layers implies that sediment removal has also taken place, presumably due to a change to a more runoff dominated hydrological regime. These three processes

Table 3 Uranium series and radiocarbon ages of fossil remains and guano from Toca da Boa Vista (TBV), Toca da Barriguda (TBR) and Toca das Onças (ONC) Sample

238

U (ppb)

δ234U (measured)

[230Th/238U] (activity)

[230Th/232Th] (ppm)

Age (yr) (uncorrected)

Age (yr) (corrected)

δ234U (initial)

TBR-21A TBR-21B TBR-22A TBR-22B TBR-23 TBR-19A TBR-19B TBR-25A TBR-25B TBR-26 TBR-20 TBR-24A TBR-24B TBV-70 TBV-69A TBV-69B ONC-01 ONC-02 ONC-03

32.2 30.0 32.9 33.0 37.4 31.6 20.2 21.0 21.0 34.0 46.1 77.4 84.9 4123 42 42 108.3 108.4 94.5

544.1 508.1 637.5 649.6 718.4 371.1 360.4 373.5 381.0 885.9 572.8 498.3 540.8 888.0 669.0 675.1 1660.0 1862.0 1882.0

0.21511 0.20214 0.23026 0.22964 0.24948 1.38561 1.00659 1.28121 1.28814 0.24656 1.16873 1.11284 1.11142 1.2777 0.2370 0.23710 0.3659 0.5669 0.4858

81.0 113.5 451.8 389.2 157.5 448.8 202.1 323.5 324.8 50.6 199.0 75.0 132.4 6709 97 88.3 70.6 12.8 20.2

16,241 ± 273 15,592 ± 249 16,389 ± 284 16,213 ± 151 16,948 ± 204 294,064 ± 10,053 134,383 ± 2,565 224,167 ± 11,066 223,548 ± 6,884 15,138 ± 220 131,600 ± 1,685 132,658 ± 1,870 125,389 ± 1,451 109,786 ± 584 16,555 ± 178 16,499 ± 188 15,900 ± 200 23,500 ± 200 19,800 ± 300

15,425 ± 491 15,031 ± 375 16,242 ± 293 16,044 ± 172 16,513 ± 298 293,259 ± 9,994 132,803 ± 2,646 223,043 ± 10,911 222,431 ± 6,840 13,913 ± 654 130,046 ± 1,836 128,433 ± 2,814 123,084 ± 1,838 109,744.3 ± 584 15,865 ± 389 15,741 ± 424 15,000 ± 500 16,100 ± 3900 15,800 ± 2000

568.4 530.2 667.5 679.7 752.7 850.0 524.6 701.6 714.4 921.4 827.2 716.4 765.8 1211 700 705.8 1699 2316 2104

Sample

Lab code

Radiocarbon age (yr BP)

Calibrated age (yr BP)

TBV-1 GUA-1

Beta-103616 Beta-183567

16,700 ± 100 >44,150

20,559–19,275 –

Analytical errors are ± 2σ of the mean.

516

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

the presence of more than one sedimentary sequence intercalated by distinct speleothem layers. The assumption by Cartelle (1992, 1995) that the faunal remains are all essentially contemporary is therefore unlikely to be correct. 3. U-series methods A number of attempts to directly date fossil bone by the U-series method have been reported. Despite some apparently successful results (Rae and Ivanovich, 1986), post mortem uranium uptake causes inhomogeneous uranium distribution in the bone, precluding reliable dating (Schwarcz and Blackwell, 1992; Ayliffe and Veeh, 1988). More recently Pike et al. (2002) have suggested that more reliable age estimates on bone may be obtained by modelling the profile of U uptake within the bone. In this study U-series ages for fossil emplacement within the caves were obtained by dating of calcite speleothem associated with the fossil bones. For speleothem overlying the bone, a minimum age was obtained, whilst for speleothem underlying the bone, the age is a maximum. Isotopic analysis was by thermal ionisation (Edwards et al., 1986) and inductively coupled plasma mass spectroscopic (Shen et al., 2002) techniques. A review of the U-series method as applied specifically to cave deposits can be found in Richards and Dorale (2003). Where possible duplicate analyses were obtained to confirm analytical reliability, and lack of disturbance of the closed isotopic system within the calcite. We corrected for initial 230Th using an initial 230 Th/232 Th atomic ration of 4.4 × 10− 6 ± 2.2 × 10− 6. This is the value for a material at secular equilibrium

with a bulk earth 232Th/238U value of 3.8. The error was arbitrarily assumed to be 50%. Most of the samples had low 232Th/238 U ratios and high measured 230Th/232 Th ratios. For these samples, the correction for initial 230 Th was minimal and any error in our assumptions regarding initial 230Th/232 Th is small. However, a few of the samples had relatively low measured 230Th/232 Th ratios. These samples are in italics in Tables 3 and 4. For these samples, we recommend considering the uncorrected age to be a maximum age for the calcite. In addition, two samples of fossil guano were directly dated by the conventional Beta counting 14C method. 4. Results 4.1. Northern Bahia area 4.1.1. Toca da Boa Vista caves U-series analyses were performed on samples from several sites both in Toca da Boa Vista and Toca da Barriguda (Table 3). Two sites in Toca da Barriguda (Fig. 2) yielded ages for fossil skeletons. At site 1, three skeletons occur a short distance apart in the same passage. A complete and articulated skeleton of the extinct small ground sloth Nothrotherium maquinense was found lying inside a rimstone dam. The skeleton was covered by a thin veneer of speleothem calcite that was collected and separated in two sub-samples (TBR21A and TBR-21B). The analyses yielded concordant minimum ages around 15,000 yr that post-date animal emplacement. About 15 m from the Nothrotherium skeleton, unarticulated remains of the fossil (but still living) deer Mazama were found inside a depression in

Table 4 Uranium series ages for fossil remains from the Lagoa Santa area Sample

238

U (ppb)

δ234U (measured)

[230Th/238U] (activity)

[230Th/232Th] (ppm)

Age (yr) (uncorrected)

Age (yr) (corrected)

δ234U (initial)

CAIAC-01 CAIAC-02 CUV-01 BAU-01 BAU-02 ZMUC-14 ZMUC-69 ZMUC-130 ZMUC-460 ZMUC-2322 ZMUC-316A-1 ZMUC-316A-2 ZMUC-316B-1 ZMUC-316B-2

208.7 171.3 105.0 126.2 131.1 367.5 407.4 140.7 209.8 161.1 168.0 145.6 153.4 141.6

45.3 106.5 50.8 413.2 416.1 166.4 23.9 226.7 113.9 232.5 −11.4 − 8.9 − 10.0 −11.7

1.0066 1.0414 0.2776 0.8486 0.8902 0.8170 0.1250 0.1834 0.1499 0.1709 0.9316 0.9511 0.9518 0.9494

24.9 163.0 20.6 10.1 9.0 151.6 15.2 18.6 27.4 48.2 65.7 78.3 87.9 80.6

326,000 ± 12,200 269,500 ± 6500 33,400 ± 700 94,563 ± 3198 101,437 ± 3785 126,048 ± 1269 14,215 ± 244 17,591 ± 296 15,748 ± 212 16,229 ± 169 318,609 ± 17,285 358,943 ± 23,008 365,376 ± 24,555 365,211 ± 24,220

323,300 ± 14,900 251,600 ± 32,000 27,100 ± 3400 64,218 ± 16,534 64,850 ± 20,200 123,934 ± 1635 10,276 ± 2001 13,687 ± 1990 13,372 ± 1212 14,849 ± 711 311,369 ± 16,488 352,789 ± 21,830 359,890 ± 23,373 359,195 ± 22,967

54.0 46.5 130.7 495.4 499.8 236.2 24.6 235.7 118.3 242.5 − 27.6 − 24.1 − 27.6 − 32.4

Sampling codes: Gruta da Caianga (CAIAC), Gruta Cuvieri (CUV), Gruta do Baú (BAU), ZMUC (Zoological Museum of the University of Copenhagen — Lund breccia collection). Analytical errors are ±2σ of the mean.

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

the floor of the cave passage. The fossil was covered by speleothem calcite, and a sample of it was separated into two sub-samples (TBR-22A and TBR-22B) prior to analysis. Both ages agree within errors and correspond to a minimum age around 16,000 yr for the emplacement of the Mazama bones. A few metres away from the Mazama remains, a few scattered long bones lie within another depression. A sample of calcite overlying bone (TBR-23) yielded a minimum age of 16,500 yr for bone emplacement. Unfortunately, no taxonomic identification of the bones has been possible. Another skeleton of N. maquinense in Toca da Barriguda (site 2) was heavily embedded in speleothem calcite but had been removed by palaeontologists prior to our sampling. Fragments of scattered bone and calcite were found at the site as discarded debris from the excavation. The cleanest samples containing both bone and speleothem calcite were collected. Since it was not possible to assess the original depositional context, there is some uncertainty in the relative age of the bone fragments and the calcite. Six analyses were performed. TBR-19A and TBR-19B are sub-samples from the same calcite fragment, which contained a bone attached to its base. We do not know the source of the discrepancy in ages between these sub-samples, although it is possible that there are multiple growth layers on the sample. We note that the date for TBR-19B is represented in other samples from the same cave. Four further samples were analysed. TBR-25A and TBR-25B are identical sub-samples from another calcite fragment attached to bone. The two ages are concordant and indicate a minimum emplacement age around 223,000 yr. Another fragment of calcite (TBR-26) yielded a very much younger age (c. 14,000 yr) and may represent speleothem that covered the bones, the age of sample TBR-25 possibly pre- rather than post-dating deposition of the fossil. TBR-20 is a somewhat porous calcite that represents material underlying the fossil bones, and has an age of 130,000 yr. There are considerable difficulties in assessing the relationship between the several bone fragments and separated pieces of calcite. It is clear that more than one generation of speleothem (or perhaps even bone) existed at this site. The age of c. 14,000 (TBR-26) probably represents a minimum age for the entire deposit, but it seems likely that most of the bones are older than c. 223,000 yr. Elsewhere in Toca da Barriguda, a semi-articulated skeleton of Puma concolor, the living South American puma, was found embedded in a thin layer of speleothem calcite. Sub-samples from the layer were separated for analysis (TBR-24A and TBR-24B). The

517

resulting ages do not overlap within errors (Table 3). Although the cleanest sub-sample, the age of TBR-24B does not correlate with episodic growth periods recorded from a much larger U-series dataset from the same cave (Wang et al., 2004). Thus TBR-24A probably represents the best analysis and yields a minimum age of 128,000 yr. An additional age was determined on a stalagmite sampled from Toca da Boa Vista (TBV-70) which contained a broken fragment of bone embedded near its base. A sub-sample from immediately above the bone was dated at 109,744 ± 584 yr, a minimum age for the bone emplacement. Unfortunately, the very fragmentary nature of the bone does not allow identification of the fossil (C. Cartelle, pers. comm.). The results from the Toca da Boa Vista caves thus suggest that fossil species may date from as early as 223,000 yr, while still extant species (Mazama) were present prior to 128,000 yr. In both cases, the most recent ages are about 15,000 yr. 4.1.2. Toca da Boa Vista caves—bat remains Four dates were obtained for the ubiquitous bat remains from Toca da Boa Vista (Table 3). A clean layer of speleothem calcite overlies a one-metre thick layer of clastic sediment intermixed with abundant bat bones. Duplicate ages (TBV-69A and TBV-69B) were in close agreement, and yield a minimum age of around 15,800 yr. Two further samples were radiocarbon dated. Sample TBV-1 is of fossil guano and yielded a calibrated age between 20,559 and 19,275 cal yr BP. These results are somewhat younger than a calibrated AMS radiocarbon age reported by Czaplewski and Cartelle (1998) for the humeri of a fossil bat of 24,656– 22,824 cal yr BP. Another area of bat guano lies close to a former entrance of the cave, now blocked by sediment and speleothems. The upper portion of a 1 m thick unit of powdery guano was sampled (GUA-1) and gave an age beyond the radiocarbon limit. The results strongly suggest that there was more than one period of bat occupation in Toca da Boa Vista. 4.1.3. Other sites The lack of suitable speleothem calcite deposits prevented our dating of bones in Toca dos Ossos. However, U-series analyses were performed on samples from Toca da Onça. Three speleothem samples were obtained from above (ONC-01), between (ONC-02), and below (ONC-03) large bones of Eremotherium. All three samples have a high content of detrital 230Th, but the ages agree within errors and indicate that these bones were emplaced around 15,000 yr.

518

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

4.2. Lagoa Santa area Fourteen ages were obtained from Lagoa Santa cave fossil deposits (Table 4). The palaeontological site at Gruta Cuvieri is among the few from the Lagoa Santa area that has not been emplaced by runoff. The horizontal passage from the entrance gives place to three vertical pits that have worked as natural traps. In one of the pits an articulated skeleton of the medium sized ground sloth Catonyx cuvieri was found below clastic sediment and a speleothem layer. A sample of this layer (CUV-01) has relatively high 232Th concentration and has a corrected age of 27,100 ± 3400 yr. The speleothem layer has been naturally truncated as it shows signs of dissolution and later sediment emplacement on its basal portion. The Catonyx skeleton, however, clearly represents an individual that fell in an upright position into the pit. CUV-01 age probably represents a minimum age for the fossil skeleton. At another vertical pit in the same cave, Neves and Piló (2003) have reported a calibrated AMS radiocarbon age of 11,570–11,250 cal yr BP on bone collagen of another C. cuvieri skeleton. At Gruta da Caianga, a series of speleothem layers overlie a fossiliferous breccia with a silty clay matrix cemented by calcite. Fossil bones comprise medium sized fragments that become more frequent towards the base of the deposit. Among the fragments, bones belonging to the still living wild pig Tayassu can be distinguished. Two samples of speleothem calcite (CAIAC-01 and CAIAC-02) were obtained. They both overlie what is probably the same fossiliferous breccia, although there is no continuity with the bone bearing sediments. Both results (323,300 ± 14,900 yr and 251,600 ± 32,000 yr) could represent minimum ages for this deposit. Gruta do Baú contains fossiliferous breccia with fragmentary remains of extinct giant armadillos Hoplophorus euphractus and Pampatherium humboldti (Abuhid and Ferreira, 1991). A layer of speleothem, previously U-series dated by Piló (1998) at 77,700 ± 6100 yr, overlies the bone bearing breccia. Two further samples were collected from the same layer (BAU-01 and BAU-02), both with high 232Th content. The results, despite errors associated with correction for initial 230 Th agree with the previously reported age. Nine additional samples were obtained from Lund's breccia collection, housed at the Zoological Museum of the University of Copenhagen (ZMUC). They comprise speleothem calcite with fossil bone fragments attached on their underside. All results are thus thought to represent minimum ages for the deposits. Site locations

and bone attribution come from Lund (1845b), although no detailed stratigraphic information is provided. ZMUC-69 from Lapa do Marinho is a detritally contaminated sample that lies over the skull of the living deer Mazama. It has yielded a minimum corrected age at 10,276 ± 2001 yr. ZMUC-130 also overlies Mazama bones from an unknown site at Gruta do Baú. This sample also has relatively high 232 Th concentration and its corrected age (13,687 ± 1990 yr) represents a minimum age for these bones. A further minimum age on Mazama is from calcite on a fragment of jaw from Lapa do Taquaral (ZMUC-460), which gave a corrected age of 13,372 ± 1212 yr. Turning now to extinct species, a sample from Gruta do Baú has yielded a corrected age of 14,849 ± 711 yr for speleothem calcite (ZMUC-2332) overlying the skull of H. euphractus. It is not clear whether this site excavated by Lund has any stratigraphical relation to the fossiliferous site studied by Piló (1998) and sampled in our work (above). Sample ZMUC-14 is of speleothem calcite that overlies the vertebra of C. cuvieri from an unspecified location, and gives an age of 123,934 ± 1635 yr. The remaining sample comes from a cave near Cochos Lake. Four subsamples of the same speleothem layer (ZMUC-316), which overlies an atlas bone of the extinct llama Palaeolama were dated. Three of the analyses yielded concordant minimum ages (around 350,000 yr) that demonstrate the considerable antiquity of these animals in subtropical Brazil. 5. Discussion We have demonstrated above that the taphonomic processes giving rise to the fossil mammal remains in the Toca da Boa Vista caves, the river cave Toca dos Ossos, and Toca das Onças deposits are very different. The distribution, size and articulated nature of the remains from the Toca da Boa Vista caves indicate that animals entered the cave individually in search of water or shelter. The disarticulated and broken nature of the bones, bias to larger species and localisation predominantly in the active river passages indicate that the Toca dos Ossos fossils were emplaced by the river running through the cave. We have less information on the taphomy of the Toca das Onças deposits, but it is clear that this site is a classic pitfall trap. Thus differences in the faunal records from the 3 Northern Bahia sites, which are less than 100 km apart in the same climatic zone, could be due to their different taphonomy, rather than due to differences in the ecosystem above the caves, the interpretation advanced by Cartelle (1999). The fossil remains in the Lagoa

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

Santa area caves include both deposits emplaced by runoff and pitfalls. In fact because the various cave deposits at Lagoa Santa have been extensively excavated over an extended period, the species list is much greater than that for the much more sparse Toca da Boa Vista caves fauna (121 species compared to 14 at Toca da Boa Vista, whereas the number of bat species are comparable 22 at Lagoa Santa compared to 20), a second potential source of bias in any palaeoenvironmental interpretation. The taphonomy of the deposits also has implications for the age and contemporaneity of the faunas. Remains that are emplaced by surface runoff are dependent on specific climatic conditions, and may thus be episodic in nature. Maximum runoff and sediment yield is produced under dry climate conditions subject to occasional intense precipitation events (Langbein and Schumm, 1958; Knox, 1972). Under these conditions the lack of vegetation leaves the soil unprotected and a considerable flux of sediment can enter the caves during runoff events. Speleothem calcite deposition, on the other hand, denotes a more equable but moist climate when vegetation cover is more continuous, and rainfall infiltrates the soil and percolates into the cave. The deposits of the Lagoa Santa caves thus suggest at least two very different palaeoclimates associated with the fossil accumulation and later calcite deposition. Some delay between animal death and transport of bone into the cave by runoff may occur and the rate of bone accumulation may be very slow, with the deposits spanning several thousand years (Graham, 1993). Such time-averaging effects (Martin, 1999) will cause some mixing and prevent precise bracketing of the age of fossil deposits, but nevertheless the inclusion of several species together in a deposit strongly suggests that they were contemporaneous. However, given the cyclic nature of Pleistocene climates, deposits derived from several runoff episodes and therefore of different antiquity may be present in the same cave. This is most clearly seen in the Lagoa Santa caves, where the bone bearing sediments are intercalated between speleothem calcite indicating periods of very different climate. In contrast the pitfall and autopod processes may continue over extended periods of time, and thus the contemporaneity of either individuals or species cannot be assumed. This may explain why Cartelle (1999, pp. 41) found that the faunal associations from NE Brazil “represent a very diversified set of habitats… that is difficult to analyse”. They may simply be noncontemporaneous species. In the case of pitfalls the stratigraphy of the deposits may permit temporal correlation and ordering of the deposits, but in the case

519

of Toca das Onças no record of the site stratigraphy is available. Published results from radiocarbon dating of fossil bones from the Lagoa Santa caves (Table 1) appear to support the arguments above, ages ranging from 9130 ± 150 to 16,900 ± 70 yr BP, and suggest emplacement over a quite limited time period at the termination of the last glacial. Similar 14C ages from Mato Grosso, Piauí and Bahia support this contention, although some of these dates result from archaeological excavations, which in South America comprise necessarily late Pleistocene–Holocene sites. Cartelle (1992, 1995, 1999) concluded that the Brazilian bone deposits were derived from the Pleistocene–Holocene boundary, a period of marked climate change. Our results suggest that this interpretation is not actually correct. The uranium series ages which we have obtained are not directly on bone, but on speleothem calcite associated with it. The majority of our samples are from speleothem which overlies the bone and thus postdates its emplacement. They thus provide minimum ages for the remains. The youngest ages we have obtained are comparable to those from 14C dating (Table 1) from Northern Bahia and Lagoa Santa areas (10,276 ± 2001 yr from over the skull of the living deer Mazama in Lapa do Marinho, Lagoa Santa area) and there is a clustering of ages between 15,000 and 16,000 yr with the oldest of this group being 16,242 ± 293 yr. The earliest radiocarbon dates are from somewhat prior to that (3 ages in the range 19,000– 20,000 cal yr BP at Lagoa Santa, and one > 22,850 cal yr BP at Toca da Boa Vista). However, many of the fossil bones dated come from much earlier (9 out of 19 specimens sampled). This is somewhat surprising as no infinite radiocarbon ages have previously been reported, even for samples also drawn from the Lund breccia collection (Neves and Piló, 2003). This may be due to the lack of collagen in the older samples which precludes their dating by the 14C method. Interestingly, the bat guano dated in this study did retain sufficient carbon to yield an infinite 14C age (sample GUA-1 from Toca da Boa Vista > 44,150 yr BP). This indicates that it is not only the surface dwelling mammals which are represented in the earlier fossil deposits. The oldest sample dated (360,000 ± 23,000 yr) is of speleothem overlying the bones of the extinct llama Palaeolama from Lagoa Santa, and indicates the considerable age range of the deposits present in Brazilian caves. Also of note is that this age range is also represented in nonextinct species, such as modern wild pig (Tayassu tajacu, > 251,600 ± 32,000 yr), also from the Lagoa Santa area. Similarly in the Northern Bahia area,

520

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

extinct Nothrotherium (ground sloth) was present prior to 223,000 ± 11,000 yr, and the still living P. concolor prior to 128,000 ± 2,800 yr. The great range of ages reported here for fossil mammals from NE Brazil call into question the previous assertions of Cartelle (1999, pp. 39) that “the fauna in question is absolutely synchronous”, and “that the palaeontological sites mentioned in this studied region (the Cerrado and Caatinga of Brazil) can be considered synchronous”. Given the limitations on any palaeoclimate interpretation because of differences in site taphonomy, fossil-based palaeoclimate interpretation is more likely to produce reliable results if it focuses on the presumed living mode and probable palaeohabitat of specific species, although this approach is not without its own risks (Shipman, 1981). For instance, for fluvial-derived bone concentrations, interpretation depends on the scale of the catchment basin, which can represent more than one habitat, limiting palaeoenvironmental reconstruction (Andrews, 1990). There is considerable evidence from the fossil remains that conditions were much wetter in NE Brazil at the time of accumulation. Czaplewski and Cartelle (1998) point out that 2 of the bat species present in the fossil remains in Toca da Boa Vista, Mimon bennettii and Furipterus horrens are currently limited to the wetter Mata Atlantica and Amazonia biomes (Fonseca et al., 1996), while Eptesicus fuscus is currently limited to the Amazonia biome. The two Mormoopidae species which numerically dominate the bat remains (Mormoops megalophylla and Pteronotus parnellii) are respectively absent and uncommon in the cattinga and cerrado biomes of NE Brazil at present, requiring hot and humid (> 90% relative humidity) caves for roosting (Bonaccorso et al., 1992). Similar conclusions can be reached from the mammal fauna. The large extinct monkey Protopithecus brasiliensis found in Toca da Boa Vista was a brachiator and had a prehensile tail, and thus was adapted to life in a closed forest environment (but see Heymann, 1998) but other extinct sloths such as E. laurillardi are considered to be generalists, and could have survived in a wide range of environments. Of the currently extant species represented in the Toca da Boa Vista caves and Toca dos Ossos faunas, 70% are species that may live in the forest, but do not depend on it (Cartelle, 1999). Two are also species that may have been displaced from their current distribution by colder conditions (Lama guanicoe and the large rodent Myocastor coypus). The porcupine Coendou prehensilis is currently restricted to forested environments, although it is not certain that this is the species represented in Toca da Boa Vista.

The overall evidence is thus that wetter and more forested conditions existed in the Northern Bahia area at the time the fossil remains were deposited, probably related to the wet events determined by Wang et al. (2004) and Auler et al. (2004). 6. Conclusions We have shown that because of the different taphonomic processes occurring, palaeoenvironmental interpretation of faunas from different cave sites should only be undertaken with considerable caution. At Toca da Boa Vista caves the fossil deposits were emplaced by random penetration of animals into the caves, a process that could occur over an extended time scale, and which gave a bias to smaller mammals. In contrast for most of the Lagoa Santa caves and Toca dos Ossos, the faunal remains were emplaced by runoff, and whilst individuals are more probably contemporaneous, there may well be a bias to preservation of larger species with more robust bones. Our uranium series dating shows rather than accumulating in a single period at the end of the Pleistocene and start of the Holocene, the fossil remains in Brazilian caves span much of the upper Pleistocene, some being emplaced prior to 350,000 yr. We have emphasised that palaeoenvironmental interpretation of the deposits is best based on the known habitat requirements of extant species, or fossils which have particular adaptations to a particular environment, such as a prehensile tail. Sedimentological and chronological analysis of the fossil bearing deposits at Brazilian caves indicate that significant variation in type and intensity of accumulation processes can occur at the scale of karst areas, caves, and individual cave passages, resulting in highly complex fossil assemblages that do not allow immediate palaeoecological and faunal interpretation without a solid understanding of taphonomic processes. Faunal reconstruction should be based on a composite assemblage of several fossil sites produced by diverse taphonomic processes. Acknowledgements We acknowledge CNPq grants 200711/95-4, 540064/01-7 and 840059/95-4, FAPESP grants 99/ 00670-7 and 00/14917-3 and NSF grants 971203, 9809459 and 01116395. We thank the speleological assistance of Grupo Bambuí de Pesquisas Espeleológicas. Cave sampling was performed with the kind permission of IBAMA/CECAV.

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

References Abuhid, V.S., Ferreira, M.A.C., 1991. Novas descobertas paleontológicas na Gruta do Baú. Proceedings 3rd Congress ABEQUA. ABEQUA, Belo Horizonte, p. 27. Andrews, P., 1990. Owls, Caves and Fossils. The University of Chicago Press, Chicago. Auler, A.S., Farrant, A.R., 1996. A brief introduction to karst and caves in Brazil. Proceedings of the University of Bristol Spelaeological Society, vol. 20, pp. 187–200. Auler, A.S., Smart, P.L., Tarling, D.H., Farrant, A.R., 2002. Fluvial incision rates derived from magnetostratigraphy of cave sediments in the cratonic area of eastern Brazil. Zeitschrift fur Geomorphologie 46, 391–403. Auler, A.S., Wang, X., Edwards, R.L., Cheng, H., Cristalli, P.S., Smart, P.L., Richards, D.A., 2004. Quaternary ecological and geomorphic changes associated with rainfall events in presently semi-arid northeastern Brazil. Journal of Quaternary Science 19, 693–701. Ayliffe, L.K., Veeh, H.H., 1988. Uranium-series dating of speleothems and bones from Victoria Cave, Naracoorte, South Australia. Chemical Geology 72, 211–234. Baffa, O., Brunetti, A., Karmann, I., Dias Neto, C.M., 2000. ESR dating of a toxodon tooth from a Brazilian karstic cave. Applied Radiation and Isotopes 52, 1345–1349. Behrensmeyer, A.K., 1991. Terrestrial vertebrate accumulations. In: Allison, P.A., Briggs, D.E.G. (Eds.), Taphonomy: Releasing the Data Locked in the Fossil Record. Plenum Press, New York, pp. 291–335. Bonaccorso, F.J., Arends, A., Genoud, M., Cantoni, D., Morton, T., 1992. Thermal ecology of moustached and ghost-faced bats (Mormoopidae) in Venezuela. Journal of Mammalogy 73, 365–378. Brain, C.K., 1981. The Hunters or the Hunted? An Introduction to African Cave Taphonomy. The University of Chicago Press, Chicago. Brain, C.K., 1995. Understanding the stratigraphic complexity of South African australopithecine cave deposits: the contribution of John T. Robinson. South African Journal of Science 91, 435–437. Brook, G.A., Cowart, J.B., Brandt, S.A., Scott, L., 1997. Quaternary climatic change in southern and eastern Africa during the last 300 ka: the evidence from caves in Somalia and the Transvaal region of South Africa. Zeitschrift fur Geomorphologie Suppl. Bd. 108, 15–48. Campy, M., Chaline, J., 1993. Missing records and depositional breaks in French late Pleistocene cave sediments. Quaternary Research 40, 318–331. Cartelle, C., 1992. Edentata e megamamíferos herbívoros extintos da Toca dos Ossos (Ourolândia, BA, Brasil). Doctoral Thesis, Universidade Federal de Minas Gerais, Brazil. Cartelle, C., 1995. A fauna local de mamíferos pleistocênicos da Toca da Boa Vista (Laje dos Negros, BA). Thesis, Universidade Federal de Minas Gerais, Brazil. Cartelle, C., 1999. Pleistocene mammals of the cerrado and caatinga of Brazil. In: Eisenberg, J.B., Redford, K.H. (Eds.), Mammals of the Neotropics. The Central Neotropics. University of Chicago Press, Chicago, pp. 27–46. Cartelle, C., Bohorquez, G.A., 1982. Eremotherium laurillardi Lund, 1842: Parte I. Determinação específica e dimorfismo sexual. Iheringia 7, 45–63. Cartelle, C., Hartwig, W.C., 1996. A new extinct primate among the Pleistocene megafauna of Bahia, Brazil. Proceedings of the

521

National Academy of Sciences of the United States of America 93, 6405–6409. Czaplewski, N.J., Cartelle, C., 1998. Pleistocene bats from cave deposits in Bahia, Brazil. Journal of Mammalogy 79, 784–803. Edwards, R.L., Chen, J.H., Wasserburg, G.J., 1986. 238U–234U– 230 Th–232Th systematics and the precise measurement of time over the past 500,000 years. Earth Planetary Science Letters 81, 175–192. Faure, M., Guérin, C., Parenti, F., 1999. The Holocene megafauna from the Toca do Serrote do Artur (São Raimundo Nonato archaeological area, Piauí, Brazil). Comptes Rendus de l'Academie Sciences Paris 329, 443–448. Fonseca, G.A.B., Herrmann, G., Leite, Y.L.R., Mittermeier, R.A., Rylands, A.B., Patton, J.L., 1996. Lista Anotada dos Mamíferos do Brasil. Occasional Papers in Conservation Biology 4, 1–38. Gillieson, D., 1996. Caves. Processes, Development, Management. Blackwell, Oxford. Guérin, C., Curvello, M.A., Faure, M., Hugueney, M., MourierChauviré, C., 1996. The Pleistocene fauna of Piaui (northeastern Brazil): palaeoecological and biochronological implications. Fumdhamentos 1, 56–103. Graham, R.W., 1993. Processes of time-averaging in the terrestrial vertebrate record. In: Kidwell, S.M., Behrensmeyer, A.K. (Eds.), Taphonomic Approaches to Time Resolution in Fossil Assemblages. Paleontological Society Short Courses in Paleontology, vol. 6, pp. 102–124. Harman, R., Gallagher, K., Brown, R., Raza, A., Bizzi, L., 1998. Accelerated denudation and tectonic/geomorphic reactivation of the cratons of northeastern Brazil during the late Cretaceous. Journal of Geophysical Research 103, 27091–27105. Hartwig, W.C., Cartelle, C., 1996. A complete skeleton of the giant South American primate Protopithecus. Nature 381, 307–311. Heymann, E.W., 1998. Giant fossil New World primates: arboreal or terrestrial? Journal of Human Evolution 34, 99–101. Karmann, I., 1994. Evolução e dinâmica atual do sistema cárstico do Alto Vale do Rio Ribeira de Iguape, sudeste do Estado de São Paulo. Doctoral Thesis, Universidade de São Paulo, Brazil. Knox, J.C., 1972. Valley alluviation in southwestern Wisconsin. Annals of the Association of American Geographers 62, 401–410. Laming-Emperaire, A., Prous, A., Morais, A.V., Beltrão, M., 1975. Grottes et Abris de la Région de Lagoa Santa, Minas Gerais, Brésil. Cahiers d'Archeologie d'Amerique du Sud 1. Paris. 185 pp. Langbein, W.B., Schumm, S.A., 1958. Yield of sediment in relation to mean annual precipitation. American Geophysical Union Transactions 39, 1076–1084. Liais, E., 1872. Climats, Geologie, Faune et Geographie Botanique du Brésil. Garnier Freres, Paris. Lund, P.W., 1840. View of the fauna of Brazil previous to the last geological revolution. Magazine of Natural History 4, 1–8, 49–57, 105–112, 153–161, 207–213, 251–259, 307–317, 373–389. Lund, P.W., 1845a. Notice sur des ossements humains fossiles, trouvés dans une caverne du Brésil. Mémoires de la Societé Royale des Antiquaires du Nord 49–77. Lund, P.W., 1845b. Katalog over Professor Dr. P.W. Lund Palaeontologiske Samling. 2. Afdeling Breccier. Zoological Museum of the University of Copenhagen Archives. Unpublished manuscript. Lyman, R.L., 1994. Vertebrate Taphonomy. Cambridge University Press, Cambridge. Marshall, L.G., Berta, A., Hoffstetter, R., Pascual, R., Reig, O.A., Bombim, M., Mones, A., 1984. Mammals and stratigraphy: geochronology of the continental mammal-bearing Quaternary of

522

A.S. Auler et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 240 (2006) 508–522

South America. Paleovertebrata, Memóire Extraordinaire, Montpellier 1–76. Martin, R.E., 1999. Taphonomy. A Process Approach. Cambridge University Press, Cambridge. Moriarty, K.C., McCulloch, M.T., Wells, R.T., McDowell, M.C., 2000. Mid-Pleistocene cave fills, megafaunal remains and climate change at Naracoorte, South Australia: towards a predictive model using U–Th dating of speleothems. Palaeogeography, Palaeoclimatology, Palaeoecology 159, 113–143. Neves, W.A., Piló, L.B., 2003. Solving Lund's dilemma: new AMS dates confirm that humans and megafauna coexisted at Lagoa Santa. Current Research in the Pleistocene 20, 57–60. Paula Couto, C., 1953. Paleontologia Brasileira. Mamíferos. Instituto Nacional do Livro, Rio de Janeiro. Paula Couto, C., 1970. Paleontologia da região de Lagoa Santa, Minas Gerais, Brasil. Boletim do Museu de História Natural 1, 1–21. Paula Couto, C., 1975. Mamíferos fósseis do Quaternário do sudeste brasileiro. Boletim Paranaense de Geociências 33, 89–132. Paula Couto, C., 1979. Tratado de Paleomastozoologia. Academia Brasileira de Ciências, Rio de Janeiro. Peyre, E., Guérin, C., Guidon, N., Coppens, Y., 1998. Human Pleistocene remains from Garrincho Cave, Piauí, Brazil. Comptes Rendus de l'Academie des Sciences Paris 327, 335–360. Piló, L.B., 1998. Morfologia cárstica e materiais constituintes: Dinâmica e evolução da depressão poligonal Macacos-Baú–Carste de Lagoa Santa, MG. Doctoral Thesis, Universidade de São Paulo, Brazil. Prous, A., 2002. O homem pré-histórico e a megamastofauna pleistocênica no Brasil. Uma revisão bibliográfica. O Carste 14, 52–59. Pike, A.W.G., Hedges, R.E.M., Van Calsteren, P., 2002. U-series dating of bone using the diffusion–adsorption model. Geochimica et Cosmochmica Acta 66, 4273–4286. Rae, A.M., Ivanovich, M., 1986. Successful application of uranium series dating of fossil bone. Applied Geochemistry 1, 419–426. Renault, P., 1968. Contribution a l'etude des actions mécaniques et sédimentologiques dans la spéleogenèse. Annales de Spéléologie 23, 529–596.

Richards, D.A., Dorale, J.A., 2003. Uranium series chronology and environmental applications of speleothems. In: Bourdon, B., Henderson, G.M., Lundstrom, C.C., Turner, S.P. (Eds.), Uranium Series Geochemistry. Reviews in Mineralogy and Geochemistry, vol. 52, pp. 407–460. Schwarcz, H.P., Blackwell, B.A., 1992. Archaeological applications. In: Ivanovich, M., Harmon, R.S. (Eds.), Uranium Series Disequilibrium. Clarendon Press, Oxford, pp. 513–552. Shen, C.C., Edwards, R.L., Cheng, H., Dorale, J.A., Thomas, R.B., Moran, S.B., Weinstein, S.E., Edmonds, H.N., 2002. Uranium and thorium isotopic and concentration measurements by magnetic sector inductively coupled plasma mass spectrometry. Chemical Geology 185, 165–178. Shipman, P., 1981. Life History of a Fossil. An Introduction to Taphonomy and Paleoecology. Harvard University Press, Cambridge. Simms, M.J., 1994. Emplacement and preservation of vertebrates in caves and fissures. Zoological Journal of the Linnean Society 112, 261–283. Sutcliffe, A.J., 1985. On the Track of Ice Age Mammals. Harvard University Press, Cambridge. Vialou, A.V., Aubry, T., Benabdelhadi, M., Cartelle, C., Figuti, L., Fontugne, M., Solari, M.E., Vialou, D., 1995. Découverte de Mylodontinae dans un habitat préhistorique daté du Mato Grosso (Brésil): l'abri rupestre de Santa Elina. Comptes Rendus de l'Academie Sciences Paris 320, 655–661. Vrba, E.S., 1976. The significance of bovid remains as indicators of environment and predation patterns. In: Behrensmeyer, A.K., Hill, A.P. (Eds.), Fossils in the Making. University of Chicago Press, Chicago, pp. 247–271. Walter, H.V., 1948. The Pre-history of the Lagoa Santa Region (Minas Gerais). Tipografia Brasil, Belo Horizonte. Wang, X., Auler, A.S., Edwards, R.L., Cheng, H., Cristalli, P.S., Smart, P.L., Richards, D.A., Shen, C.C., 2004. Wet periods in northeastern Brazil over the past 210 kyr linked to distant climate anomalies. Nature 432, 740–743. Winge, H., 1888–1915. E Museo Lundi. 5 volumes. Carlsberg, Copenhagen.