Dating the colonization of Sahul (Pleistocene Australia–New Guinea): a review of recent research

Journal of

Archaeological SCIENCE Journal of Archaeological Science 31 (2004) 835–853 http://www.elsevier.com/locate/jas

Dating the colonization of Sahul (Pleistocene Australia–New Guinea): a review of recent research J.F. O’Connell a*, J. Allen b a

b

Department of Anthropology, University of Utah, 270 South 1400 East, Salt Lake City, UT 84112, USA Archaeology Program, School of Historical and European Studies, La Trobe University, Bundoora, VIC 3086, Australia Received 15 May 2003; received in revised form 27 September 2003; accepted 9 November 2003

Abstract The date for the initial colonization of Sahul is a key benchmark in human history and the topic of a long-running debate. Most analysts favor either a 40,000 BP or 60,000 BP arrival time, though some have proposed a much earlier date. Here we review data from more than 30 archaeological sites with basal ages >20,000 years reported since 1993, giving special attention to five sites with purported ages >45,000 years. We conclude that while the continent was probably occupied by 42–45,000 BP, earlier arrival dates are not well-supported. This observation undercuts claims for modern human migrations out of Africa and beyond the Levant before 50,000 BP. It also has critical but not yet conclusive implications for arguments about a human role in the extinction of Sahul megafauna.  2003 Elsevier Ltd. All rights reserved. Keywords: Early colonization of Australia; Modern human origins; Dating in archaeology

1. Introduction Sahul is the continent formed when glacioeustatically lowered sea levels exposed dry land connections between Australia, New Guinea and Tasmania [7]. Despite decades of inquiry, the date for its initial human colonization remains a contentious issue. Some favor a 40–45,000 year (40–45 ka) BP figure [3,60]; others one R60 ka [107,131]; still others one >>100 ka [72,142]. Resolving the matter has important implications for a wide range of questions, among them the pattern and timing of early modern human population movements out of Africa [53,84] and the role of humans in Sahul megafaunal extinctions [94,116]. Here we briefly review the history of the problem, note some recent developments in radiocarbon chronometry that affect current debate, then consider in detail the key data, with emphasis on information from five localities published in the

* Corresponding author. Tel.: +1-801-581-3933; fax: +1-801-581-6252 E-mail addresses: [email protected] (J.F. O’Connell), [email protected] (J. Allen). 0305-4403/04/$ - see front matter  2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2003.11.005

last decade or so. We conclude that while an initial occupation date of 42–45 ka BP now seems wellsupported, earlier dates do not, largely for taphonomic reasons. We comment briefly on the implications of these findings for debates about early modern human population movements and their environmental impact. 2. Some history As recently as the early 1960s, humans were thought to have arrived in Sahul less than ten thousand years ago [96]. By the early 1980s, widespread application of the radiocarbon technique had pushed the minimum date to just under 40 ka [103,121]. While agreeing that this figure was the proper conservative estimate, the late Rhys Jones [77, p. 30] observed that it was “suspiciously close” to the limits of 14C chronometry, and suggested that an earlier date might well be documented by another technique. In the early 1990s, his prediction was arguably met by the report of luminescence dates of 50–60 ka from the Arnhem Land sites of Malakunanja and Nauwalabila [109,112] (see Fig. 1 for locations of all sites mentioned in text).

836

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Fig. 1. Sahul and adjacent parts of southeast Asia, showing sites and localities mentioned in text and tables. Shaded areas indicate land exposed by a 200 m fall in sea level.

In the decade since, debate about the initial occupation date has crystallized along two main lines: one underlining the technical limits of 14C dating and championing the virtues of luminescence [31,78,108], all in support of a minimum 60 ka date for Sahul colonization; the other contending that whatever the merits of luminescence as a technique, its application requires greater sensitivity to archaeological context than enthusiasts have so far displayed, and that the case for a date much greater than 40 ka is at best poorly warranted [2,97]. Quietly paralleling these developments for a time was the emergence of an argument for a much earlier date.

Writing in the mid-1970s, Kershaw [80] suggested that the occupation of Sahul might be marked by certain changes in fossil pollen records indicating the appearance of anthropogenic fire regimes, much like those known ethnographically [76]. Singh et al. [122] applied this line of reasoning to the analysis of a sediment core from Lake George, near Canberra, proposing that a decline in pollen from fire-sensitive taxa, coincident with a sharp rise in charcoal particle counts, indicated the arrival of humans in the surrounding countryside as early as 130 ka BP. Though archaeologists were reluctant to take this suggestion seriously [143, p. 42], palynologists and other Quaternarists continued to develop

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

it, ultimately pushing for colonization dates as early as 185 ka [72,81,142]. In the mid-1990s, this argument drew apparent support from reports on Jinmium, an archaeological site in northwestern Australia, dated by luminescence at 116–176 ka [54]. As it happened, the Jinmium dates were overturned almost as soon as they were published: reanalysis showed that the archaeological record there had accumulated within the last 20 ka [114]. Meanwhile, the proposed evidence of early anthropogenic fire was found to be attributable mainly to sharp changes in climate and related increases in fire frequency, not human activity [25,68,82]. This leaves us with the dispute between proponents of what are now called the “long” (R60 ka) and “short” (ca. 40 ka) chronologies [3]. Further discussion requires a brief review of recent developments in radiocarbon dating. 3. On

14

C chronometry

Critics of the use of radiocarbon in dating Sahul colonization have focused on two closely related issues: the technical limits of the method and the problem of sample contamination. It is generally agreed that conventional 14C analysis cannot produce reliable dates >40 ka. In principle, accelerator mass spectrometry (AMS) can yield dates >50 ka, but small amounts of postdepositional contamination can generate erroneously young age estimates in any sample, particularly those with real ages >30 ka. This problem has been alleviated (though not entirely eliminated [18]) by the development of improved pretreatment protocols, notably “acid base wet oxidation with stepped combustion” (ABOX-SC) [17]. Paired with AMS, this particular technique re-establishes the possibility of generating accurate sample age estimates >50 ka. As we will see, it has already had an important impact on Sahul chronology. A different but equally important problem has been brought into focus by continuing efforts to “calibrate” 14 C dates. It has long been recognized that age estimates produced by analysis of radiocarbon may differ significantly from real sample ages, primarily as a function of past changes in rates of atmospheric 14C production and exchange with the marine reservoir. Simply put, 14C years are not the same as solar years, and the relationship between the two varies through time. This relationship is well-controlled for the Holocene but remains highly problematic for earlier time periods [104,138]. Analyses of terrestrial and marine data sets by various combinations of radiocarbon, U-series, luminescence, paleomagnetic and stratigraphic techniques indicate very large fluctuations in atmospheric 14C content in the >30 ka time range, mainly as a function of changes in geomagnetic field strength and North Atlantic thermo-

837

haline circulation [14,37,83,86,139,146]. As a result, 14C age estimates on samples deposited prior to this date may fall short of real values by up to 7 ka. Complicating matters further, estimates of the timing and magnitude of these excursions vary significantly across data sets, making it impossible at present to produce a reliable calibration curve. For the moment, there is no easy way around this problem: 14C dating must remain an important source of information on Sahul chronology, yet in the absence of data from other sources the precise significance of any pre-30 ka radiocarbon date must also be considered unclear. Relying primarily on the summary in Van der Plicht [138], we assume for the sake of discussion that 14 C dates in the 30–42 ka range may be up to (but not necessarily as much as) 7 ka short of true calendar values. For 14C dates >42 ka, the offset may be somewhat less, possibly in the 3–4 ka range, perhaps even smaller (1–2 ka). All 14C dates reported here are uncalibrated and their “real” ages assessed on a case-by-case basis, often by comparison with associated luminescence dates, which are generally seen to be expressed in the rough equivalent of solar years [105]. 4. Data from the last decade In their comprehensive early 1990s review of the field, Smith and Sharp [124] listed 58 sites east of Wallacea with basal dates >20 ka, 27 with dates >30 ka. Collectively, they indicated human occupation in a variety of site types, dated by several radiometric techniques, located across a wide range of environments throughout Sahul and in island Melanesia nearby to the east. The oldest were Malakunanja at 55–60 ka, Nauwalabila at >53 ka, and Huon Peninsula at ca. 40 ka (all ages estimated on the basis of techniques other than radiocarbon). No other reported archaeological age determination exceeded 40 ka in its central tendency. Data published since 1993 fall into two categories: (1) those from newly discovered and/or reported sites, and (2) those from previously known sites that have been re-dated, often with different and/or refined radiometric techniques. Tables 1 and 2 show the oldest claimed ages for sites in each category. Table 1 lists 14 trans-Wallacean sites published since Smith and Sharp’s compilation: twelve from Sahul and two from island Melanesia. Four have produced dates >30 ka, but only one of these, a non-finite date, is older than 40 ka. While previous contamination-related criticisms of radiocarbon dates in the 30–40 ka range might equally well be leveled at Table 1 sites, these results reflect a pattern remarkably similar to that available a decade ago [2]. Table 2 lists 18 trans-Wallacean localities, several of which (including some also shown on Table 1) have been re-dated more than once. Seventeen are in Sahul; one is

838

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Table 1 Sites in Australia, New Guinea and island Melanesia with basal ages >20 ka, excavated or published since Smith and Sharp [124]. All radiocarbon dates are uncalibrated; all other dates are roughly equivalent to solar years.

consistent with the original estimates. Finally, at Devil’s Lair, OSL dates generally support the newer radiocarbon chronology, suggesting human occupation there before 43 ka, but not before 51 ka.

Site

Region

Reference Technique

4.2. Re-dating by AMS with ABOX-SC pretreatment

Drual FABM

W Victoria New Britain N Queensland Cape York Central Desert Aru Islands Cape York

[16] [132]

GRE 8 Hearth Cave Kulpi Mara Lemdubu Cave Mushroom Rock West New Guinea II Parmerpar Meethaner Riwi Sandy Creek 2 Serpents’ Glen Walkunder Arch Yombon

E Victoria SW Tasmania Kimberley Cape York Western Desert Cape York New Britain

14

Earliest date (ka)

C Fission track 14 C

22.140.16 >27, <40

[42] [130]

14

C AMS

21.50.25 29.510.23

[99] [95]

U-series TL

27.020.29 28.73.5

[101] [39]

14

C AMS

21.00.9 33.850.45

[8] [35] [98]

14

C AMS AMS

>40 24.60.22 23.550.14

[28]

AMS

29.70.5

[102]

14

35.570.48

[123]

C

37.112.95

Six of the eight sites in this group returned ages similar to or slightly younger than those reported earlier. Finite ages were produced for the initial occupations at Riwi and Carpenter’s Gap; dates for the first uses of Puritjarra, Cuddie Springs and Warreen were marginally reduced. The technique also managed to correct previous anomalies in the sequence for Puritjarra, suggesting that discrepancies were the product of contaminants that had escaped pre-treatment in earlier analyses [126]. At Nauwalabila, multiple analyses yielded dates so inconsistent with those produced by luminescence that they were rejected as contaminated [18] (see below for further discussion). At Devil’s Lair, the ABOX-SC technique proved remarkably successful in extending the previous 14C chronology. Five samples, arrayed in proper stratigraphic order, produced finite dates down to 48.13+2.59/
1.96 ka, plus three infinite readings from deeper in the profile (see also below). 4.3. Summary of recent re-dating

in island Melanesia. Only three have newly reported dates older than 50 ka: (1) Mungo, dated by Thorne et al. [131] at 62 ka; (2) Huon Peninsula at 47–61 ka; (3) Malakunanja at 56–61 ka, similar to the original estimate. Of the remaining 15 sites, only Devil’s Lair has a new date in excess of 45 ka. The other 14 are all pegged at <45 ka. Luminescence dating has been used at nine of these sites; ABOX-SC and AMS dating at eight. Four sites have been reassessed using both techniques. 4.1. Re-dating by luminescence Of the nine sites in this group, two—Malangangerr and Allen’s Cave—yielded results from levels below those containing datable carbon. Neither new date exceeds 40 ka in its central tendency. In two other sites, Parmerpar Meethaner and Puritjarra, newly reported luminescence dates have been rejected in favor of more coherent 14C chronologies (for details, see [3,126]). At Cuddie Springs and Ngarrabullgan, OSL dates are in essential agreement with newer radiocarbon chronologies. At Mungo, OSL dates were used to support the U-series age of 626 ka for the M-III burial [131]. They were later used by another dating team [24] to reduce the age of this interment to 402 ka, and the claimed presence of humans in the area to 46–50 ka. At Malakunanja, as indicated above, new dates are

Reanalysis of nine sites by luminescence suggests occupation at Devil’s Lair, Mungo and the Huon Peninsula at 45–50 ka, and at Malakunanja >50 ka. Data from the other five sites offer no support for ages greater than 40 ka. Of the eight sites re-dated by AMS with ABOX-SC pre-treatment, only Devil’s Lair produced a significantly earlier date, but again not older than 50 ka. Overall, re-dating has marginally extended the ages of a number of sites, but has mostly failed to support the long (R60 ka) chronology. The five sites where claims for human antiquity >45 ka persist are Huon Peninsula, Devil’s Lair, Lake Mungo, Nauwalabila, and Malakunanja. We now turn to a more detailed review of each. 5. Five key sites 5.1. Huon Peninsula The site is located at Bobongara (or Fortification) Point on the Huon Peninsula, northeast New Guinea [66]. It consists of a series of seven raised coral terraces, the uppermost of which is now about 400 m above sea level. Each terrace was formed when glacio-eustatic rises in sea level overtook the land, itself undergoing a process of continual tectonic uplift [30]. As reported by Groube et al. [66], the site was identified by the surface presence of flaked stone artifacts, including more than 100 waisted and grooved

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

839

Table 2 Sites in Australia, New Guinea and island Melanesia re-dated since Smith and Sharp [124] or not cited by them. Devil’s Lair date is from Layer 30; older dates from this site are discussed in text. For Malakunanja OSL dates, a=single-aliquot measure, b=single grain measure. As in Table 1, all radiocarbon dates are uncalibrated; all other dates are roughly equivalent to solar years. Site

Reference

Technique

Revised age (ka)

Previous age (ka)

Allen’s Cave Buang Merabak Carpenter’s Gap Cuddie Springs

[113] [87] [49] [116] [48] [134]

OSL AMS ABOX/AMS OSL ABOX/AMS ABOX/AMS, OSL, ESR, U-series AMS U-series TL OSL U-series, OSL, ESR OSL ABOX/AMS AMS, OSL ABOX/AMS ABOX/AMS OSL C14 C14, AMS, TL, OSL ABOX/AMS ABOX/AMS AMS ABOX/AMS

39.83.1 40.090.57 40.60.8 35.42.8 32.00.55 45.47+1.42,
1.21

>17–20 31.990.83 39.71.0 363 363 32.481.25

30.30.8 47–61 32.07.0 (a) 60.77.5, (b) 55.58.2 626 46–50 <16 34.92.1 35.46 +0.75
0.69 35.280.67 w44.2 29.80.72 34.61.6 31.140.47 41.31.0 21.80.4 33.175+0.67–0.62

26.010.41 w40 w23 6110 Various estimates 40–>50 626 >53.45.4, <60.36.7 >37.17 34.92.1 34.92.1 33.850.45 15.490.36 >20 34.61.6 >40 20.560.25 34.790.51

Devil’s Lair Fern Cave Huon Peninsula Malangangerr Malakunanja Mungo Nauwalabila Ngarrabullgan

Parmerpar Meethaner Pallawa Trounta (ACH/84/1) Puritjarra Riwi Wallen Wallen Creek Warreen

[41] [105] [107] [115] [131] [24] [18] [43] [41] [133] Unpublished [127] [125] [126] [8] [64] unpublished

axes, most >20 cm long and weighing >1 kg. Artifacts were found in creek beds cutting across terrace IIIa and the higher, older terrace IVb. A series of three tephras (volcanic ash deposits) was trapped in the depression behind terrace IIIa and on the lower front of the rise to terrace IVb. Neither tephras nor artifacts were found on the younger, lower terraces, beginning with IIIb, indicating that these terraces had yet to emerge from the sea when the artifacts and tephras were deposited. The ages for terraces IIIa and IIIb thus provide a general age range for the artifacts, assuming the reconstructed depositional sequence is accurate. Under these conditions, the artifacts are not younger than 44.50.7 ka (terrace IIIb), nor older than 61.40.6 ka (the oldest of several dates from terrace IIIa), as reported by Chappell and associates [30, Table 2; 32, Table 1]. (Note that these U-series dates have been generated more recently than those reported earlier by Groube et al. [66] and increase the terrace ages cited there by about 10%.) Excavations consisted of cleaning back a vertical section of creek bank, ca. 30 m long, and cutting a trench ca. 10 m long at right angles to the creek section [66, Table 2]. The three superimposed tephras thus exposed (T1 the youngest) sit above bedded tuff and reef limestone. Artifacts were recovered from four findspots within the tephras. Findspots 1–3 produced a single waisted axe each; findspot 4 a core and two flakes.

Findspots 1 and 2 are located at the intersection of T2 and T3; findspot 4 is within T2; findspot 3 is in the upper part of T1. TL dates were obtained by Groube et al. [66] for each of the three tephras (Table 3). There are uncertainties in the dose rate, discussed by Groube et al. [66, p. 454] and Roberts [105, pp. 867–868], mainly involving potassium and water content. Review of these considerations led Groube et al. to estimate a minimum age of ca. 40 ka for both the tephras and the artifacts they contain. Roberts [105, p. 868] subsequently recalculated a minimum age of ca. 47 ka based on different assumptions about water content. Groube et al. were cautious about the possibility of post-depositional disturbance at the site. The cleaned section is steeply sloping, and the excavators identified slope wash and re-deposition post-dating each tephra

Table 3 Huon Peninsula TL dates for tephras from Groube et al. [66]. Tephra Unit

Minimum age (ka)

Maximum age (ka)

T1 T2 T3

31.3 37.1 36.2

42.1 60.5 59.8

840

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Table 4 Stratigraphy and dating of the lowest artifact-bearing levels in Devil’s Lair. Depths are approximate and derived from the section drawing of the east face of Trench 9, the one used by Turney et al. [134] but better illustrated by Dortch [45, p. 26]. The radiocarbon date for Layer 30 Lower is listed only as Layer 30 by Turney et al. [134, Table 1] but is ascribed here to Layer 30 Lower on the given depth of this sample (365 cm). Two ESR samples were processed for both Layer 27 and Layer 39 and results presented using both early uptake and linear uptake models. In each case the younger ages were produced using the early uptake model. Layer

Depth (cm)

27

329–334

28

334–336

29 30 Upper 30 Lower 31 32 33 34–38 39

336–341 341–355

ABOX-SC (ka)

OSL (ka) ESR (ka) 423–524 443–594

41.46+1.4,
1.19

355–380

45.47+1.42,
1.21

380–390 390–395 395–418 418–458 458–530

46.73+2.19,
1.72

43.42.2 44.42.1

47.12.6

48.13+2.59,
1.96 51.12.6 647–8510 >52.0 757–889

fall as well as subsequent down-slope creep. They also allowed that the absence of tephras on younger terraces could reflect post-depositional erosion. That said, the axe at findspot 2 was found in several pieces, apparently buried “after breakage with little subsequent displacement” [66, p. 454]. Unless these artifacts were discarded on an ancient tephra and then buried by a similarly ancient but re-deposited tephra, the several lines of evidence reviewed here suggest that the Huon artifacts date to more than 44 ka. Establishing a maximum age less than the 52–61 ka limit imposed by the date for terrace IIIa is not possible at this time. 5.2. Devil’s Lair Devil’s Lair is a large limestone cave in the far southwest of Western Australia, excavated and reported by C. Dortch [45]. The most recent dating program reported by Turney et al. [134] has produced a coherent pattern of radiocarbon determinations, generally supported by related OSL dates. Table 4 lists the stratigraphic layers and dates relevant to this discussion. We include the electron spin resonance (ESR) dates although these are largely discounted in Turney and associates’ analysis. The lowest in situ feature is a set of four nested hearths, referred to as the “hearth complex in layers 27–30.” The radiocarbon age of the layer 28 hearth is

Table 5 Stone artifact counts from the lower levels of Devil’s Lair. Data from Dortch and Dortch [46, p. 31] and Turney et al. [134]. Layer

Artifacts

19–27 28 29 30 Upper 30 Lower 31 32 33 34 35 36 37 38

55 34 14 8 0 0 1 1 1 1 0 1 1

41.46+1.4/
1.19 ka, consistent (given the general comments on 14C chronometry above) with associated OSL dates of 41.2–45.6 ka and 42.3–46.5 ka. While there are uncertainties surrounding the precise stratigraphic relationships of these remains, and hence their relationships with the dates [3], it seems clear that they reflect a human presence at the site somewhere in the 41–46 ka range. Claims for artifacts deeper in the sequence are more problematic. Underlying layer 30 Upper is layer 30 Lower, a 25–30 cm thick fan of in-washed soils, devoid of artifacts and distinctly different from the sandy brown sediments above. At one standard deviation, it is dated by 14C at 44.26–46.89 ka. Further below, layers 31–38 reflect episodes of heavy erosion, marked by numerous channels and convoluted scouring features. The matrix contains large amounts of limestone rubble and boulders, weathered to sub-rounded and sub-angular forms, typical of cave entrance debris. Layers 28–30 Upper also display erosional features, but not to the extent shown in layers 31–38. Numbers of artifacts tallied for these layers have differed through time as analyses have continued. Dortch [45, p. 56] originally claimed fourteen, with another half-dozen “probables.” Turney et al. [134, p. 5] cite six, one each in layers 32–35, 37, and 38. Dortch (pers. comm.) now counts seven, including four calcrete and two quartz flakes plus two halves of a limestone concretion thought to have been used as a hammerstone. The lowest of these items was associated with layer 37. Table 5 shows the distribution in Turney and associates’ published tally, compared with those in overlying layers 19–30 Upper. Turney et al. [134, p. 11] and Dortch (pers. comm.) agree that the artifacts in layers 31–38 are probably not in primary depositional context. Dortch contends that they are unlikely to have filtered down from higher in the deposit, given the thickness of sterile layer 30 Lower

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

841

and the fact that upper level assemblages include proportionally more quartz and fewer calcrete items. Instead, he favors the notion that these remains were re-deposited from older sediments elsewhere in the cave or from area(s) outside a now-blocked cave entrance. The presence of charred bone in layers 31–38 is cited in support of this argument. Clearly, if these artifacts were washed in from elsewhere and are now overlain stratigraphically by layer 30 Lower, they must be older than 44–47 ka. Nevertheless, we remain skeptical about that inference for several reasons: • Uncertainty about the size and nature of the early assemblage. The stone tools represented are at best few in number, and charred bone is by Dortch’s [45, p. 18] own account “not unequivocally associated with human activities.” • The absence of evidence for older deposits elsewhere in the cave or outside the former entrance. Excavations in the latter area produced no date greater than 20 ka BP [45, p. 47]. • The observation, discussed more fully below for Nauwalabila, that artifacts can easily be displaced downward over the distances involved here, even through seemingly unbroken strata. Dortch’s suggestion that differences in raw material composition refute this possibility might be right, but the small size of the layer 31–38 assemblage makes it difficult to assess. While a human presence at Devil’s Lair somewhere in the 41–46 ka range is now indicated, the argument for an earlier occupation remains equivocal. 5.3. Lake Mungo Lake Mungo is one of thirteen inter-connected dry lake basins on the lower reaches of Willandra Billabong Creek, western New South Wales. It has been studied by geomorphologists and archaeologists since the late 1960s (see [24] and the many sources cited therein for review). Sedimentary units pertinent to this discussion fall into two groups: Lower Mungo (hereafter LM) deposits, defined in lake-bordering lunettes by a predominance of quartz sands, indicating relatively high lake levels; and Upper Mungo (UM) deposits, marked by a predominance of pelletal clay sands, indicating fluctuating, generally low lake levels (Fig. 2). At least six, possibly seven calcareous soils are identified in LM components of the Mungo lunette, each reflecting regional drying, absence of sediment supply and/or plant colonization. The upper boundary of the LM phase is defined either by a particularly well-developed soil, sometimes called the Lower Mungo Soil (hereafter LM soil), or by a sharp interface between LM quartz sands and UM pelletal clays, indicating that the LM soil has been removed by

Fig. 2. Simplified summary of mid-Upper Pleistocene stratigraphy at Joulni locality, Lake Mungo lunette, modified from Bowler et al. [24]. Note that the M-III burial and lowest artifacts were located several hundred meters apart.

erosion. Wu¨stenquartz (silt-sized quartz grains with red clay skins), an indicator of arid conditions west of the Willandra area, is present in small amounts intermittently through LM times, becoming more common in the UM phase. Chronometric control is provided primarily by radiocarbon and luminescence dates from several localities, mainly on the Mungo and nearby Arumpo lunettes (see [3] for a detailed summary). Comprehensive review of the radiocarbon record [58,59] indicates frequent uncertainty about the nature of materials being dated, the impact of sample contamination, and the effectiveness of pre-treatment. Some of the luminescence dates are equally open to question because of local differences in dose rates and degree of peri-depositional bleaching [21,62]. In addition, relationships between dates and key sedimentary features are sometimes uncertain [19,21]. Twenty-one luminescence and 14C dates for the UM phase range from 16–42 ka [3]. The oldest nine radiocarbon dates are >34 ka. At one standard deviation, eight bracket the period 34.5–35.4 ka (Fig. 3a). Two of three luminescence dates from the Joulni locality, at the southern end of the Mungo lunette, bracket the period

842

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Fig. 4. TL and OSL dates from three transects across Lower Mungo sand: (a) W-series dates reported by Bowler and Price [21]; (b) MI/MB dates on samples collected near the M-I burial location, reported by Bowler et al. [24]; (c) J3/M-III dates on samples collected near the M-III burial location reported by Bowler and Price [21] and Bowler et al. [24] (see Allen and O’Connell [3] for detailed review). Within each section (a–c), samples are arranged in descending stratigraphic order, uppermost at left.

Fig. 3. Selected 14C, TL and OSL dates from: (a) Upper Mungo clay/sand; (b) Lower Mungo soil (see Allen and O’Connell [3] for detailed review). Radiocarbon dates are uncalibrated.

36.9–39.5 ka. Given the adjustments indicated in the section on chronometry above, the 14C dates are probably consistent with the luminescence determinations. Bowler et al. [24] use these data to peg the beginning of the UM phase at ca. 38–40 ka. Eighteen radiocarbon and luminescence dates on the LM soil range from 16–45 ka [3]. Seven 14C dates are >34 ka (Fig. 3b). At one standard deviation, all bracket the period 35–37 ka. Rough calibration might put them as early as 43 ka. Five luminescence dates on the LM soil, all from Joulni, are in the 40–44 ka range. Bowler et

al. [24] appeal to both radiocarbon and luminescence dates in placing this component at ca. 40–42 ka. Thirty-two luminescence dates come from LM deposits below the LM soil at Joulni (Fig. 4a–c) [3]. Fourteen TL dates are in the 34–58 ka range; 18 OSL dates are in the 41–62 ka range. The dates show no clear relationship with depth, inviting skepticism about their precise validity and/or stratigraphic relationships. Archaeological materials from Willandra have not been published in much detail, but are reportedly present in some quantity (>>103 artifacts; >102 features) in the LM soil on the Mungo lunette and in the LM phase sands immediately underlying it [1,19,23,24,74,117,141]. Items most frequently mentioned include lithics, hearths, roasting pits, and small, shallow middens containing fish, shellfish and small mammal remains. Despite recent erosion and displacement, at least some of these materials must be at the point of initial deposition, implying ages equivalent to that of the LM soil or slightly older, roughly 40–43 ka.

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Similar remains are associated with the LM soil (but not the underlying sands) on the Arumpo lunette [21, pp. 133, 147]. Arguments for earlier dates on the archaeology appeal to two data sets. One is a collection of eleven silcrete flakes found well down in LM phase quartz sands at Joulni (Fig. 2) [24,120]. Stratigraphic position and roughly bracketing OSL dates of 50.12.4 ka and 49.72.7 ka (above the artifacts) and 47.92.4 ka and 45.72.3 ka (below) are read by Bowler et al. [24] and Shawcross [120] to indicate a 46–50 ka age for the assemblage; but we are skeptical, partly because of the small size of the assemblage, but mainly because of its problematic sedimentary context and the inversion of the bracketing dates. Bowler [19, p. 122] notes that LM quartz sands are typically marked by “steep avalanche bedding” and “cross sets,” indicating erosional and depositional processes likely to facilitate the movement of artifacts well below the point of initial deposition. Stratigraphic relationships may also have been disrupted by burrowing animals, which were once common in these deposits [55,141]. The lack of a strong age-depth relationship in any of the three luminescence-dated transects shown in Fig. 4 further underlines the need for caution here. Conjoining exercises involving these eleven flakes have not yet been attempted, but Shawcross [120, pp. 193–195] reports such efforts from higher in the Joulni section where the core and some flakes in one refitted set are separated vertically by up to 60 cm. In light of all this, the case for a pre-45 ka date for the lowest-lying eleven items must be considered unproven. The second argument for a pre-45 ka date involves the human burial called here M-III [131]. The skeleton was found at Joulni, partly exposed on an eroded surface in LM quartz sands. The stratigraphic origin for the pit in which it was placed could not be observed directly. In their first report, excavators Bowler and Thorne [22, p. 134] argued that the burial was in place before the LM soil was formed, partly because dark humic sands from the soil were absent from the grave fill. Compelling as that reasoning may have been at the time, it is undercut by more recent descriptions of the grave fill as containing Wu¨stenquartz, pelletal clay and (most notably) “dark, reworked soil” [24, p. 840], collectively implying an interment no earlier than the formation of the LM soil (i.e. 40–42 ka BP), possibly even post-dating it (see also [3,19, p. 150, Fig. 9; 20, Fig. 2]). Thorne et al. [131] have developed a very different chronology for the burial, based on eight U-series and ESR analyses of the skeleton itself, one U-series analysis of the attached calcite matrix, and two OSL determinations on LM sediments they claim are stratigraphically equivalent to the burial’s position (Table 6). Ages range from 50–103 ka, with the preferred estimate for the burial itself being 626 ka. Despite its inconsistency

843

Table 6 Dates on M-III burial from Thorne et al. [131]. All determinations except OSL pertain either to the skeleton itself or to immediately adjacent sediments. OSL dates relate to Lower Mungo sands collected several hundred meters away from that location. Sample

Age (ka)

Mass spectrometry Bone/M-III

96.82.1 58.31.2 50.70.9 54.50.7 8221 747 605 636 787 593 633

Sediment Gamma spectrometry ESR OSL

with the regional chronology, this estimate has been widely accepted by non-Australianists [89,128]. Bowler and others [20,60,62] have offered detailed and compelling critiques of Thorne and associates’ dating methods (see [67] for reply). The complexity and experimental nature of the U-series and ESR analyses, the assumptions required in controlling uranium uptake and loss, and (especially) the mismatch with the luminescence-based chronology for the LM sands outlined by Bowler et al. [24] all count against Thorne and associates’ position. Of the ten OSL dates in the array that transects the M-III burial site, only one is >52 ka (Fig. 4c). All ten apparently underlie the interment stratigraphically [21,24]. The luminescence dates reported by Thorne et al. still might be seen to support their argument, but the fact that the samples they analyzed were collected several hundred meters away from the M-III burial location greatly weakens that support. We conclude from all this that neither the M-III burial nor the local archaeological record provides solid evidence for a human presence at Lake Mungo before ca. 43 ka. 5.4. Nauwalabila This site is a rockshelter formed by an outlier of the Arnhem Land escarpment, about 200 km east of Darwin [79]. Artifacts are found through 2.5 m of sand deposits and into 30–40 cm of underlying sandstone rubble, the latter underlain by more sand and bedrock (Fig. 5). Fourteen radiocarbon and five luminescence dates cited in early reports indicated fairly steady accumulation from ca. 60 ka, with the lowest artifacts found in the rubble zone, bracketed between OSL dates of 535.4 ka and 60.36.7 ka [112]. A recent paper by Bird et al. [18] offers new data and arguments in support of this antiquity.

844

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

Fig. 5. Stratigraphic diagram for Nauwalabila, modified from O’Connell and Allen [97]. Vertical scale shows depth in centimeters from ground surface. TL and OSL dates from Roberts and Jones [107]. Black bands indicate depths from which dated sediment samples were drawn.

In a previous review [97], we suggested that the site dated no older than ca. 40 ka, and that the apparent association of artifacts, sandstone rubble and dated sands at the base of the deposit was secondary, the artifacts and rubble having settled into older sediments as a result of post-depositional termite activity in the overlying deposits. Both the process and its displacement effect on larger clasts are well-documented in similar situations elsewhere [27]. Bird and colleagues [18] attempted to determine whether this had occurred at Nauwalabila by examining clay particle size distribution down the profile, but the results were inconclusive. The significant point not addressed in their analysis is that large clasts were so well-represented at the base of the deposit that they were defined by the excavators as a separate stratigraphic unit—the rubble layer. Stone layers like this are elsewhere routinely taken as indicators of termite disturbance [27,144]. In the course of their analysis, Bird et al. also discovered small charcoal particles extending down to the base of the deposit, ca. 1 m below the lowest previously reported radiocarbon date. These particles were analyzed using ABOX-SC and other pre-treatment techniques and both AMS and conventional 14C dating. Results from samples recovered from 110 cm below

ground surface to the base of the deposit are shown in Table 7, together with OSL dates on some of the same strata previously reported by Roberts et al. [112, p. 579]. Clearly the 14C dates show no coherent relationship with depth, nor do they match well with the OSL dates. Instead, their distribution is consistent with the random downward displacement of small charcoal particles through bioturbation. Bird et al. reject this reading and offer a counter hypothesis based in part on the intra-site distribution of small iron oxide nodules called pisolites (sometimes pisoliths). The presence of these items below 110 cm is said to imply a fluctuating but generally high groundwater table, at or near the paleo-surface of the site, at or very near the Pleistocene–Holocene transition. Bird et al. [18, p. 1071] suggest higher rainfall, higher wet season runoff, higher river base levels and higher sea levels as possible explanations for this phenomenon. These processes are said be reflected in a gap in the calibrated radiocarbon dates between ca. 12.5 ka and 9.2 ka (Table 7). The purpose of this reconstruction is to support the view that a significant proportion of the dated charcoal samples below 110 cm have had their carbon composition altered by the proposed high water table. Above

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

845

Table 7 Nauwalabila 14C dates below 110 cm from Bird et al. [18, Table 1]. Samples subjected to ABOX-SC pre-treatment grouped separately. Samples pre-treated in other ways but designated by an ANUA code have been dated by AMS. All others are assumed to be conventional dates. Calibrated ages from Bird et al. [18, Table 1], except ANUA-6906 from p. 1066. Where more than one calibrated date is possible, only the one with the highest probability factor is listed. See Bird et al. [18] for additional details. OSL dates drawn from Roberts et al. [112, Table 2] citedthe total uncertainty listed therein. Depth below surface (cm)

Sample

104–110 107 111 115 118 123 127 139 151 151 151 152 164 164 164 170–175 176 176 176 176 180 189 189 236 236 236 228–240 241 241 256 293 293 293 293 285–301

OXODK172 ANU-8653 ANU-8654 ANU-8677 ANU-8676 ANU-8678 SUA-236 ANU-10929 ANUA-9513 ANUA-9905 ANUA-9906 ANU-10928 ANU-10927 ANUA-9514 ANUA-9907 OXODK166 ANU-3177 ANU-3182B ANUA-9512 ANUA-9908 SUA-237 ANUA-8131 ANUA-8223 ANUA-6906 ANUA-9902 ANUA-9909 OXODK168 ANUA-10317 ANUA-10318 ANUA-9903 ANUA-7618 ANUA-7619 ANUA-9904 ANUA-9912 OXODK169

ABOX-SC AMS (ka)

Other (ka)

14

C dates

14

C calibrated (BP)

OSL (ka) 13.50.9

8.180.07 10.530.24 9.810.12 11.090.24 10.920.11 13.190.18 12.330.12 18.330.28 13.290.18 13.890.34 19.990.36 12.000.25 17.120.30 16.450.32

9140–9028 12877–12105 11345–11081 13213–12892 13044–12871 16222–15500 14423–14102 22230–21314 16315–15624 17121–16204 24196–23138 14336–13500 20840–19919 20078–19139 30.02.4

22.840.52 12.000.60 8.750.1 14.60.34 19.970.365 5.880.15 7.050.14 27.350.44 12.730.14 6.960.14

26660–25420 14457–13405 9895–9599 17938–17022 24179–23122 6805–6498 7973–7724 31500–29700 15634–15148 7871–7673 53.45.4

8.210.12 9.1060.12 9.280.18 7.510.23 6.920.14 7.560.14 9.450.18

this level, hard black fragments of charcoal retain their internal structure, whereas below 110–130 cm and especially below 180 cm large charcoal fragments are sparse, often heavily coated with clays and iron oxides, and often soft and brown in appearance. The internal structures of these altered fragments are degraded and their carbon contents are mostly below 25%, compared with 36–47% for unaltered samples. Microbial activity is suggested as the vector of carbon replacement. Radiocarbon ages, especially below ca. 180 cm, are thus seen to reflect the age of this alteration, not the age of the sediments from which they were recovered. We agree that carbon samples from below 110 cm have been degraded, but for reasons developed at length elsewhere [3], we are much less certain that high ground

9298–9025 10426–10174 10688–10236 8483–8107 7862–7657 8454–8182 10878–10493 60.36.7

water levels and microbial activity are in any way implicated. For charcoal aged between 30 ka and 50 ka to yield apparent dates of 6–10 ka would require that the samples be composed almost entirely of carbon which is on average much younger than the proposed 9.2–12.9 ka period of high groundwater. Thus the questions remain: what happened to the original carbon in the charcoal? How has so much young carbon been incorporated in the samples in a form capable of resisting ABOX-SC pre-treatment, which is designed to remove humic acids and other soil organic compounds? Even more problematic for Bird and associates’ hypothesis, some of the charcoal from deeper levels of the site is neither degraded nor in situ. They report that small quantities of black and vitreous charcoal with

846

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

well-preserved woody structures are found throughout the deposit below 110 cm [18, pp. 1070–1071]. Regardless of pre-treatment, these samples also yield dates that are too young to represent the ages of the sediments in which they were found. Bird et al. [18, p. 1071] suggest that these un-degraded fragments “have most likely fallen down termite galleries from high in the sequence.” Though they do not say which of the deeper samples are degraded or intact, their own analysis thus indicates that bioturbation has affected all of the lower deposit to some unknown degree. It is not impossible that all or most of the samples, intact and degraded, have moved down the profile. The general similarity in ages of eleven of twelve charcoal samples below 180 cm supports this notion. The rubble layer at the base of the site and the distribution of relatively young but unaltered charcoal fragments below 110 cm collectively provide strong evidence for the vertical displacement of objects, including artifacts, through the lower two-thirds of the deposit. This process makes it difficult to assess the actual ages of those artifacts, particularly the ones found in the rubble layer, the area in which displacement indicators are most prominent. We therefore continue to be skeptical of the idea that Nauwalabila provides evidence of a human presence in Sahul >40 ka BP. 5.5. Malakunanja Malakunanja is a small rockshelter located about 45 km north of Nauwalabila [109]. It contains a sandy deposit >4 m deep, at the base of which is a sandstone rubble layer, similar to the one at Nauwalabila. Artifacts are found to a depth of 260 cm, bracketed by luminescence dates of 528 ka (above) and 55.58.2 ka, 60.77.5 ka and 6110 ka (below) [105,109,115]. Bird et al. [18, Table 1] have recently published several new ABOX-SC/AMS dates, one of which comes from the 254 cm level, in close proximity to the lowest artifacts. All available 14C and luminescence dates below 145 cm are shown in Table 8. Since—apart from the original brief announcement and several subsequent reports of dates—this site remains unpublished, little can be made of its claimed antiquity. Early comments by the excavators touted the site’s stratigraphic integrity [109–111]. More recently, Roberts [105, p. 856] has allowed the possibility of downward displacement of the lowest artifacts, but maintains the in situ status of those associated with “a small pit feature” at 232 cm, overlain by a TL date of 457 ka and OSL dates of 45.74.1 ka and 44.24.7 ka. While there is no published evidence allowing independent assessment of this feature, it is important to note that the underlying 14C date (10.810.20) is clearly too young. This sample may reflect vertical displacement of the kind and magnitude indicated at Nauwalabila,

Table 8 Malakunanja ABOX-SC/AMS and conventional radiocarbon dates from Bird et al. [18], Table 1 in comparison with TL and OSL dates from similar or nearby depths from Roberts et al. [109,112,115] citedthe total uncertainties provided there. For OSL dates, a=the single-aliquot measure, b=single grain measure. Depth below surface (cm) 146 149 149 152 192–216 202 232

250 254 260

ABOX-SC AMS (ka)

14

C (ka)

TL (ka)

OSL (ka)

13.390.40 10.330.15 13.050.21 152 244 18.040.30 457

(b) 45.74.1 (c) 44.24.7

528 10.810.20 6110

(b) 60.77.5 (c) 55.58.2

raising questions about the stratigraphic integrity of this part of the deposit, including the association of artifacts, dated sediments and the pit feature. If the artifacts and luminescence-dated sediments associated with the pit feature are in primary context, then humans were present at this site as early as 45 ka BP. Arguments for an earlier date, certainly one >50 ka, remain undocumented. 6. Discussion In our 1998 review, we found the evidence for a >40 ka Sahul arrival date uncompelling. Half a decade later, the situation is not much different. Improvements in 14C dating and the wider application of luminescence techniques have demonstrated a 40–45 ka human presence at eight sites—Allen’s Cave, Buang Merabak, Carpenter’s Gap, Devil’s Lair, GRE 8, Huon Peninsula, Mungo and Riwi. Malakunanja and Nauwalabila might ultimately make the cut as well, but not on currently available data: the stratigraphic connections between artifacts and dated materials are simply too tenuous. The case for a >45 ka figure is much less certain. The best candidate—Huon Peninsula—remains loosely dated; arguments for the other four (Devil’s Lair, Malakunanja, Mungo and Nauwalabila) are undercut by small numbers of artifacts and/or (again) uncertainties about the depositional relationships between dates and artifacts. There is still no firm evidence for a date >50 ka. This finding has implications for a wide range of questions in Upper Pleistocene paleoanthropology and ecology. Here we comment briefly on just two: the

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853 Table 9 Upper Pleistocene archaeological sites in southeast Asia with basal ages >20 ka. All dates are uncalibrated. Site

Region

Reference Technique

Earliest date (ka)

Golo Cave Kota Tampan Lang Rongrien Leang Burung Lene Hara Niah

Moluccas Peninsula Malaysia SW Thailand

[15] [147]

C14 C14

35.570.48 w31

[4,5]

C14

>43

S Sulawesi

[63]

C14

w31

East Timor Borneo

[100] [10]

C14 34.650.63 ABOX/AMS 420.67

timing of modern human migrations out of Africa and the role of humans in the extinction of Sahul megafauna. 6.1. Out of Africa Archaeological data have been used to support several scenarios for modern human movement beyond a probable African homeland, one pegged at about 45–50 ka BP [84], others at various dates >50 ka [53,85]. The earliest of the latter, set at about 90–120 ka, is grounded on archaeological and fossil data from the Levant [57,90,93,118,136,137] and appears to have been related to the spread of a suite of African biota into this region during a period of relatively warm inter-glacial conditions [129]. Modern humans were clearly part of this phenomenon, but there is no indication that they moved into other parts of southwest Asia or beyond at this time. In fact, the available evidence shows that they disappeared from the Levant with the return to cooler climatic conditions after 70 ka BP [6,13,57,90,119,129,135]. Another scenario commonly rehearsed posits an ex-African migration in the 60–70 ka range, possibly along the southern coast of Eurasia leading ultimately to southeast Asia and Sahul [56,85,140]. Elements of this model guide speculation about the genetic and morphological characteristics of migrant groups and their respective technological capabilities and economic orientations [38,47,128]. Support for the 60–70 ka date is drawn from two sources: certain features of Pleistocene Sahul lithic assemblages and the dates on the five problematic sites reviewed above, particularly Mungo, Malakunanja and Nauwalabila. As we have seen, the latter basis carries no weight: none of these sites indicates a definite human presence >45 ka BP. Data from adjacent areas of southeast Asia show a similar pattern: no evidence of occupation by modern humans >45 ka BP (Table 9). The lithic aspect of the argument is equally unpersuasive. Proponents characterize broad patterns in technological evolution in terms of a series of stages or

847

“modes” that define human adaptive capability, and so can be used to mark the advance and retreat of competing human or hominid populations [33,53]. Of special interest here are Modes 3–5: loosely, Middle Paleolithic, Upper Paleolithic and Mesolithic, respectively. Mode 4 industries are found across large areas of Eurasia and Africa after 45 ka BP but are absent in Pleistocene Sahul, where Mode 3 industries are the norm until mid-Holocene times. Since the latter generally pre-date Mode 4 industries elsewhere in the Old World, the inference is that they must have spread to Sahul before 45 ka BP. Dates for Malakunanja, Nauwalabila and Mungo are seen to set a minimum date for this spread: 60 ka BP [11,85,128]. This model is difficult to defend. As many have observed (and as its proponents [53] themselves concede), Modes 3 and 4 work best as descriptive devices in western Europe; elsewhere they display much less integrity [9,12,89]. Moreover, the order in which these and other modes appear is not unidirectional. The classic example is Howiesons Poort, a Mode 5 industry that surfaces briefly in southern Africa around 70 ka BP, succeeding a Mode 3 industry and later replaced by one [89]. An obvious inference is that these industries represent facultative responses to regional ecological (including social and demographic) circumstances, not phylogenetic markers, as advocates of the “mode” argument would have it. If so, they cannot be taken as a basis for speculation about the timing of Sahul colonization. While this does not necessarily imply that modern humans were restricted to Africa and (on occasion) immediately adjacent parts of southwest Asia before 50 ka BP, it does mean that archaeological data from Sahul do not support a significantly earlier excursion. 6.2. Megafaunal extinctions Pleistocene Sahul witnessed the disappearance of more than 20 genera of large-bodied fauna sometime in the middle-to-late Upper Pleistocene [51]. The possible role of humans in this process has been a focus of attention for some time [50,65,75,145]. Recent results from four projects offer potentially important purchase on the problem: • Miller et al. [94] report amino acid racemization, U-series, and AMS 14C dates on Genyornis eggshell and luminescence dates on associated sediments indicating that this large, now-extinct flightless bird disappeared from parts of the Eyre Basin (a large area of interior Australia) at 505 ka BP. • Roberts et al. [116] report 14C, luminescence and U-series dates on a taxonomically diverse sample of megafaunal remains and associated sediments continent-wide suggesting that many species disappeared within a very narrow time frame,

848

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

possibly around 46 ka BP, probably between 40–50 ka BP. • Cosgrove and Allen [40] report that the earliest sites in Tasmania, a series of rockshelters in the southwest dated by radiocarbon and luminescence to 34 ka BP, contained no extinct megafauna among >600,000 bones analyzed from eight locations. • Field and others [48,116] report 14C and luminescence dates of 29–36 ka BP on sediments containing the remains of six now-extinct taxa, including Genyornis, from Cuddie Springs (northern New South Wales). The Miller and Roberts data sets are seen by some to support the idea that all Sahul megafauna were driven to extinction by humans shortly after they arrived on the continent [44]. Flannery has in fact recently modified the 60 ka date he once favored for colonization [50] downward to 46 ka to better conform with the results of the Roberts et al. analysis [52, p. 21]. Given some flexibility in reading the data, our own assessment of the likely colonization date, 42–45 ka BP, fits comfortably with this scenario. The Tasmanian collection might be seen to do so as well, assuming it contains a comprehensive sample of the large-bodied taxa locally present at the time the sites from which it was drawn were formed. Neat as this formulation seems to be, we see several problems with accepting it [65,145]. Detailed review is beyond the scope of this paper, but three points are critical: • Most obvious are the Cuddie Springs data, which count against a rapid, continent-wide extinction model closely tied to the initial human colonization date. Though some have dismissed the association between megafauna and relatively late dates at this site on taphonomic grounds [61,106,116], there are good reasons to see it as reliable [36,48]. • Equally important is the question of the process [26]. The mechanisms most commonly nominated by proponents of human causation are direct predation (“overkill”) and habitat modification via anthropogenic change in the fire regime [50,73,94]. The overkill option is countered by the complete absence of any direct evidence of human predation on megafauna, and by the fact that Pleistocene human populations were almost certainly too low to have effected rapid extermination on a continental scale. The fire-related argument is challenged by the observation that the sharpest changes in Sahul vegetation evident over the last few hundred thousand years are climate- rather than human-driven, and that clearcut prehistoric human impacts unconnected with agriculture are difficult to identify in the Sahul pollen record [25,34,69,70,82]. • Since Tasmania was cut off from the mainland by relatively high sea levels until after 37 ka [71],

extinctions must have taken place there after this date if humans were involved. Their absence from the very large faunal collection noted above counts against this possibility. If they disappeared before 37 ka, then some factor other than human interference must be implicated. We conclude that, despite the apparent coalescence of certain provocative chronometric data, the case for rapid continent-wide extinction of megafauna as a result of human colonization has yet to be made. More complex models, involving detailed ecological analyses on a regional scale, will be required to resolve this issue satisfactorily. 7. Conclusion In the wake of reports of 50–60 ka luminescence dates for archaeological remains at Malakunanja, many welcomed the fall of the “radiocarbon barrier” and predicted an ensuing flood of similarly early dates continent-wide. Now, more than a decade later, aggressive application of improved dating techniques, including various forms of luminescence, ABOC-SC/ AMS 14C and U-series, has yet to produce the anticipated deluge. The outside date for human colonization has shifted from ca. 40 ka BP to ca. 45 ka BP, an adjustment that might have been expected given only what we now know about calibrating 14C dates >30 ka BP. No earlier dates have survived careful scrutiny unchallenged. The American case provides an obvious and instructive parallel [91]. There, an 11–12 ka BP colonization date was demonstrated almost as soon as the first relevant 14C analyses were reported. In the half century since, older dates have often been proposed, but in the end few have won broad acceptance. Those that have been affirmed have not changed the baseline date all that much [92]. The time required to refute the others, Martin’s [88] “shelf-life” for pre-Clovis sites, has averaged about ten years [91]. Uncertainties about dates and/or their association with evidence of a human presence have plagued each questionable case, and it has taken time to resolve them. The Sahul argument has so far developed along similar lines, prompting us to make an important point. Continuing improvements in dating technology have clearly been central to establishing the age of continental colonization. But contrary to recent statements by participants on one side of the debate [29,78,108], they are not the only consideration. As we hope our discussions of individual sites here and elsewhere [3,97] have shown, this problem is and always will be in large part about depositional context. In other words, this is an archaeological issue as much as it is a chronometric one. Characterizing a continuing concern with this aspect of

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

the problem as a reluctance to “[break] the moulds into which prehistory has been poured” [29, p. 80], or more pointedly as “carping” commentary [78, p. 53], is in the end simply dodging a fundamental element of the discussion. On current evidence, 45,000 calendar years is the best outside date for the colonization of Pleistocene Sahul. It requires more than the mere announcement of yet another old date, without adequate consideration of context, to change this conclusion.

Acknowledgements For information, assistance and comments we thank Shannon Arnold, Mike Barbetti, Graeme Barker, John Beaton, Peter Bellwood, Jim Bowler, John Chappell, Richard Cosgrove, Nic Conard, Bruno David, Charlie Dortch, Richard Fullagar, Jennifer Graves, Don Grayson, Rainer Grun, Simon Haberle, Jay Hall, Henry Harpending, Kristen Hawkes, Johan Kamminga, Peter Kershaw, Richard Klein, Matthew Leavesley, Dave Meltzer, Mike Morwood, Sue O’Connor, Bert Roberts, Michael Slack, Nigel Spooner, Matthew Spriggs, Marjorie Sullivan, Chris Turney and Peter Veth.

References [1] H. Allen, Where the crow flies backwards: man and land in the Darling Basin. Ph.D. thesis, Australian National University, Canberra, 1972. [2] J. Allen, S. Holdaway, The contamination of Pleistocene radiocarbon determinations in Australia, Antiquity 69 (1995) 101–112. [3] J. Allen, J.F. O’Connell, The long and the short of it: archaeological approaches to determining when humans first colonized Australia and New Guinea, Australian Archaeology 57 (2003) in press. [4] D.D. Anderson, Lang Rongrien Rockshelter: a Pleistocene/early Holocene Archaeological Site from Krabi, Southwestern Thailand, University Museum Monograph 71, The University Museum, University of Pennsylvania, Philadelphia, 1990. [5] D.D. Anderson, Cave archaeology in Southeast Asia, Geoarchaeology: An International Journal 12 (1997) 607–638. [6] B. Arensberg, O. Bar-Yosef, M. Chech, P. Goldberg, H. Laville, L. Meingnen, Y. Rak, E. Tchernov, A.-M. Tillier, B. Vandermeersch, Une sepulture neanderthalien dans la grotte de Kabara (Israel), Comptes Rendus de l’Academie des Sciences, Paris 300 (2) (1985) 227–230. [7] C. Ballard, Stimulating minds to fantasy? A critical etymology for Sahul, in: M.A. Smith, M. Spriggs, B. Fankhauser (Eds.), Sahul in Review: Pleistocene Archaeology in Australia, New Guinea and Island Melanesia, Occasional Papers in Prehistory 24, Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra, 1993, pp. 17–23. [8] J. Balme, Excavations revealing 40,000 years of occupation at Mimbi Caves, south central Kimberley, Western Australia, Australian Archaeology 51 (2000) 1–5. [9] L. Barham, Backed tools in Middle Pleistocene central Africa and their evolutionary significance, Journal of Human Evolution 43 (2002) 585–603.

849

[10] G. Barker, Prehistoric foragers and farmers in south-east Asia: renewed investigations at Niah Cave, Sarawak, Proceedings of the Prehistoric Society 68 (2002) 147–164. [11] O. Bar-Yosef, The Upper Paleolithic revolution, Annual Review of Anthropology 31 (2002) 363–393. [12] O. Bar-Yosef, S.L. Kuhn, The big deal about blades: Laminar technologies and human evolution, American Anthropologist 101 (1999) 322–338. [13] O. Bar-Yosef, B. Vandermeersch, B. Arensberg, A. Belfer-Cohen, P. Goldberg, H. Laville, L. Meignen, Y. Rak, J.D. Speth, E. Tchernov, A.-M. Tillier, S. Weiner, The excavations in Kebara Cave, Mt. Carmel, Current Anthropology 33 (1992) 497–550. [14] J.W. Beck, D.A. Richards, R.L. Edwards, B.W. Silverman, P.L. Smart, D.J. Donahue, S. Hererra-Osterheld, G.S. Burr, L. Calsyas, A.J.T. Jull, D. Biddulph, Extremely large variations of atmospheric 14C concentration during the last glacial period, Science 292 (2001) 2453–2458. [15] P. Bellwood, G. Nitihaminoto, G. Irwin, Gunadi, A. Waluyo, D. Tanudirjo, 35,000 years of prehistory in the northern Moluccas, in: G.-J. Bartstra (Ed.) Bird’s Head approaches: Irian Jaya Studies—A Programme for Interdisciplinary Research, Modern Quaternary Research in Southeast Asia 15 (1998) 233–275. [16] C. Bird, D. Frankel, N. van Waarden, New radiocarbon determinations from the Grampians-Gariwerd region, western Victoria, Archaeology in Oceania 33 (1998) 31–36. [17] M.I. Bird, L.K. Ayliffe, L.K. Fifield, C.S.M. Turney, R.G. Cresswell, T.T. Barrows, B. David, Radiocarbon dating of “old” charcoal using a wet oxidation, stepped-combustion procedure, Radiocarbon 41 (1999) 127–140. [18] M.I. Bird, C.S.M. Turney, L.K. Fifield, R. Jones, L.K. Ayliffe, A. Palmer, R.G. Cresswell, S. Robertson, Radiocarbon analysis of the early archaeological site of Nauwalabila 1, Arnhem Land, Australia: implications for sample suitability and stratigraphic integrity, Quaternary Science Reviews 21 (2002) 1061–1075. [19] J. Bowler, Willandra Lakes revisited: environmental framework for human occupation, Archaeology in Oceania 33 (1998) 120–155. [20] J. Bowler, J. Magee, Redating Australia’s oldest human remains: a skeptic’s view, Journal of Human Evolution 38 (2000) 719–726. [21] J. Bowler, D. Price, Luminescence dates and stratigraphic analyses at Lake Mungo: review and new perspectives, Archaeology in Oceania 33 (1998) 156–168. [22] J. Bowler, A.G. Thorne, Human remains from Lake Mungo: discovery and excavation of Lake Mungo III, in: R.L. Kirk, A.G. Thorne (Eds.), The Origin of the Australians, Australian Institute of Aboriginal Studies, Canberra, 1976, pp. 127–138. [23] J.M. Bowler, R. Jones, H. Allen, A. Thorne, Pleistocene human remains from Australia: a living site and human cremation from Lake Mungo, western New South Wales, World Archaeology 2 (1970) 39–60. [24] J.M. Bowler, H. Johnston, J.M. Olley, J.R. Prescott, R.G. Roberts, W. Shawcross, N.A. Spooner, New ages for human occupation and climatic change at Lake Mungo, Australia, Nature 421 (2003) 837–840. [25] D.M.J.S. Bowman, The impact of Aboriginal landscape burning on the Australian biota, New Phytologist 140 (1998) 385–410. [26] B. Brook, D.M.J.S. Bowman, Explaining the Pleistocene megafaunal extinctions: models, chronologies and assumptions, Proceedings of the National Academy of Sciences (USA) 99 (2002) 14624–14627. [27] D. Cahen, J. Moeyersons, Subsurface movements of stone artifacts and their implications for the prehistory of Central Africa, Nature 266 (1977) 812–815. [28] J. Campbell, N. Cole, E. Hatte, C. Tuniz, A. Watchman, Dating of rock surface accretions with Aboriginal paintings and engravings in north Queensland, in: S. Ulm, I. Lilley, A. Ross (Eds), Australian Archaeology ’95: Proceedings of the 1995 Australian

850

[29]

[30]

[31]

[32]

[33] [34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42] [43]

[44] [45] [46]

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853 Archaeological Association Annual Conference, Tempus 6, Anthropology Museum, Department of Anthropology and Sociology, University of Queensland, 1996, pp. 231–239. J. Chappell, Pleistocene seedbeds of western Pacific maritime cultures and the importance of chronology, in: S. O’Connor, P. Veth (Eds.), East of Wallace’s Line: Studies of Past and Present Maritime Cultures of the Indo-Pacific Region, Balkema, Rotterdam, 2000, pp. 77–98. J. Chappell, Sea level changes forced ice breakouts in the Last Glacial cycle: new results from coral terraces, Quaternary Science Reviews 21 (2002) 1229–1240. J. Chappell, J. Head, J. Magee, Beyond the radiocarbon limit in Australian archaeology and Quaternary research, Antiquity 70 (1996) 543–552. J. Chappell, A. Omura, T. East, M. McCulloch, J. Pandolfi, Y. Ota, B. Pillans, Reconciliation of late Quaternary sea levels derived from coral terraces at Huon Peninsula with deep sea oxygen isotope records, Earth and Planetary Science Letters 141 (1996) 227–236. J.D.G. Clark, World Prehistory: A New Outline, Cambridge University Press, Cambridge, 1977. R.L. Clark, Pollen and charcoal evidence for the effects of Aboriginal burning on the vegetation of Australia, Archaeology in Oceania 18 (1983) 32–37. N.A. Cole, A. Watchman, M.J. Morwood, Chronology of Laura rock art, in: M.J. Morwood, D.R. Hobbs (Eds.), Quinkan Prehistory: the Archaeology of Aboriginal Art in S.E. Cape York Peninsula, Australia, Tempus 3, Anthropology Museum, Department of Anthropology and Sociology, University of Queensland, 1995, pp. 147–160. J.B. Coltrain, J. Field, R. Cosgrove, J. O’Connell, Stable isotope and protein analyses of Genyornis remains from Cuddie Springs, Archaeology in Oceania (2004) in press. N.J. Conard, M. Bolus, Radiocarbon dating the appearance of modern humans and timing of cultural innovations in Europe: new results and new challenges, Journal of Human Evolution 44 (2003) 331–371. R. Cordeaux, M. Stoneking, South Asia, the Andamanese, and the genetic evidence for an “early” human dispersal out of Africa, American Journal of Human Genetics 72 (2003) 1586–1590. R. Cosgrove, Late Pleistocene behavioural variation and time trends: the case from Tasmania, Archaeology in Oceania 30 (1995) 83–104. R. Cosgrove, J. Allen, Prey choice and hunting strategies in the Late Pleistocene: Evidence from Southwest Tasmania, in: A. Anderson, I. Lilley, S. O’Connor (Eds.), Histories of Old Ages: Essays in honour of Rhys Jones, Research School of Pacific and Asian Studies, Australian National University, Canberra, 2001, pp. 397–430. B. David, Landscapes, Rock Art and the Dreaming: an Archaeology of Pre-understanding, Leicester University Press, 2002. B. David, L. Dagg, Two caves, Memoirs of the Queensland Museum 33 (1993) 143–162. B. David, R. Roberts, C. Tuniz, R. Jones, J. Head, New optical and radiocarbon dates from Ngarrabullgan Cave, a Pleistocene archaeological site in Australia: implications for the comparability of time clocks and for the human colonization of Australia, Antiquity 71 (1997) 183–188. J.M. Diamond, Australia’s last giants, Nature 411 (2001) 755–757. C. Dortch, Devil’s Lair, A Study in Prehistory, Western Australian Museum, Perth, 1984. C. Dortch, J. Dortch, Review of Devil’s Lair artefact classification and radiocarbon chronology, Australian Archaeology 43 (1996) 28–32.

[47] J. Erlandson, The archaeology of aquatic adaptations: paradigms for a new millennium, Journal of Archaeological Research 9 (2001) 287–350. [48] J. Field, J. Dodson, Late Pleistocene megafauna and archaeology from Cuddie Springs, southeastern Australia, Proceedings of the Prehistoric Society 65 (1999) 1–27. [49] L.K. Fifield, M.I. Bird, C.S.M. Turney, P.A. Hausladen, G.M. Santos, M.L. di Tada, Radiocarbon dating of the human occupation of Australia prior to 40 ka BP—successes and pitfalls, Radiocarbon 43 (2001) 1139–1145. [50] T. Flannery, The Future Eaters, Reed Books, Melbourne, 1994. [51] T. Flannery, R.G. Roberts, Late Quaternary extinctions in Australasia, in: R.D.E. MacPhee (Ed.), Extinctions in Near Time: Causes, Contexts and Consequences, Plenum, New York, 1999, pp. 239–256. [52] T. Flannery, P. Schouten, Lost and found. Australian Weekend Magazine, 22–23 September (2001) 18–22. [53] R. Foley, M.M. Lahr, Mode 3 technologies and the evolution of modern humans, Cambridge Archaeological Journal 7 (1997) 3–36. [54] R.L.K. Fullagar, D.M. Price, L.M. Head, Early human occupation of northern Australia: archaeology and thermoluminescene dating of Jinmium rockshelter, northern Australia, Antiquity 70 (1996) 751–773. [55] N. Fuller, Archaeological sites in the Garnpung-Gogolo and Garnpung-Leaghur interlake zone, BA Honors thesis, Department of Prehistory and Anthropology. The Faculties, Australian National University, 1986. [56] C. Gamble, Timewalkers: The Prehistory of Global Colonization, Penguin, Harmandsworth, 1993. [57] D.A.E. Garrod, D.M.A. Bate, The Stone Age of Mount Carmel I, Oxford University Press, Oxford, 1937. [58] R. Gillespie, Burnt and unburnt carbon: dating charcoal and bone from the Willandra Lakes, Australia, Radiocarbon 39 (1997) 239–250. [59] R. Gillespie, Alternative time scales: a critical review of Willandra Lakes dating, Archaeology in Oceania 33 (1998) 169–182. [60] R. Gillespie, Dating the first Australians, Radiocarbon 44 (2002) 455–472. [61] R. Gillespie, B. David, The importance, or impotence, of Cuddie Springs, Australasian Science 22 (9) (2001) 42–43. [62] R. Gillespie, R. Roberts, On the reliability of age estimates for human remains at Lake Mungo, Journal of Human Evolution 38 (2000) 727–732. [63] I. Glover, Leang Burung 2: an Upper Paleolithic rock shelter in south Sulawesi, Indonesia, Modern Quaternary Research in Southeast Asia 6 (1981) 1–38. [64] J.A.J. Gowlett, R.E.M. Hedges, I.A. Law, C. Perry, Radiocarbon dates from the Oxford AMS system: archaeometry datelist 5, Archaeometry 29 (1987) 125–155. [65] D.K. Grayson, The archaeological record of human impacts on animal populations, Journal of World Prehistory 15 (2001) 1–65. [66] L.M. Groube, J. Chappell, J. Muke, D. Price, A 40,000 year-old human occupation site at Huon Peninsula, Papua New Guinea, Nature 324 (1986) 453–455. [67] R. Grun, N. Spooner, A. Thorne, G. Mortimer, J.J. Simpson, M.T. McCullough, L. Taylor, D. Curnoe, Age of the Lake Mungo 3 skeleton: a reply to Bowler and Magee and to Gillespie and Roberts, Journal of Human Evolution 38 (2000) 733–741. [68] S.G. Haberle, M-P. Ledru, Correlations among charcoal records of fires from the past 16,000 years in Indonesia, Papua New Guinea, and Central and South America, Quaternary Research 55 (2001) 97–104. [69] S.G. Haberle, G.S. Hope, S. van der Kaars, Biomass burning in Indonesia and Papua New Guinea: Natural and human induced fire events in the fossil record, Paleogeography, Palaeoclimatology, Palaeoecology 171 (2001) 259–268.

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853 [70] L. Head, Both ends of the candle? Discerning human impact on the vegetation, Australian Archaeology 39 (1994) 82–86. [71] S. Holdaway, N. Porch, Cyclical pattern in the Pleistocene human occupation of southwest Tasmania, Archaeology in Oceania 30 (1995) 74–82. [72] W.D. Jackson, Palaeohistory of vegetation change: the last 2 million years, in: J.B. Reid, R.S. Hill, M.J. Brown, M.J. Hovenden (Eds). Vegetation of Tasmania, Flora of Australia, Supplementary Series 8, University of Tasmania, Forestry Tasmania, and Cooperative Research Centre for Sustainable Production Forestry, Hobart, 1999, pp. 64–88. [73] B. Johnson, G. Miller, M. Fogel, J. Magee, M. Gagan, A. Chirvas, 65,000 years of vegetation change in central Australia and the Australian summer monsoon, Science 284 (1999) 1150–1152. [74] H. Johnston, Pleistocene shell middens of the Willandra Lakes, in: M.A. Smith, M. Spriggs, B. Fankhauser (Eds). Sahul in Review: Pleistocene Archaeology in Australia, New Guinea and Island Melanesia, Occasional Papers in Prehistory 24, Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra, 1993, pp. 197–203. [75] R. Jones, The geographical background to the arrival of man in Australia and Tasmania, Archaeology and Physical Anthropology in Oceania 3 (1968) 186–215. [76] R. Jones, Fire stick farming, Australian Natural History 16 (1969) 224–228. [77] R. Jones, Ions and eons: some thoughts on archaeological science and scientific archaeology, in: W. Ambrose, P. Duerden (Eds.), Archaeometry: an Australasian Perspective, Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra, 1982, pp. 22–35. [78] R. Jones, Dating the human colonization of Australia: radiocarbon and luminescence revolutions, Proceedings of the British Academy 99 (1998) 37–65. [79] R. Jones, I. Johnson, Deaf Adder Gorge: Lindner Site, Nauwalabila I, in: R. Jones (Ed.) Archaeological Research in Kakadu National Park, Special Publication 13, Australian National Parks and Wildlife Service, Canberra, 1985, pp. 165–227. [80] A.P. Kershaw, A late Pleistocene and Holocene pollen diagram from Lynch’s Crater, north-eastern Queensland, Australia, New Phytologist 77 (1976) 469–498. [81] A.P. Kershaw, G.M. McKenzie, A. McMinn, A Quaternary vegetation history of northeastern Queensland from pollen analysis of ODP site 820, Proceedings of the Ocean Drilling Program, Scientific Results 133 (1993) 107–114. [82] A.P. Kershaw, J.S. Clark, A.M. Gill, D. D’Costa, A history of fire in Australia, in: R.A. Bradstock, J.E. Williams, M.A. Gill (Eds.), Flammable Australia: The Fire Regimes and Biodiversity of a Continent, Cambridge University Press, Cambridge, 2002, pp. 3–25. [83] H. Kitagawa, J. Van der Plicht, Atmospheric radiocarbon calibration beyond 11,900 BP from Lake Suigetsu laminated sediments, Radiocarbon 42 (2000) 369–380. [84] R.G. Klein, Why anatomically modern people did not disperse from Africa 100,000 years ago, in: T. Akazawa, K. Aoki, O. Bar-Yosef (Eds.), Neandertals and Modern Humans in Western Asia, Plenum, New York, 1998, pp. 509–521. [85] M.M. Lahr, R.A. Foley, Multiple dispersals and modern human origins, Evolutionary Anthropology 3 (1994) 48–60. [86] C. Laj, C. Kissel, A. Mazaud, E. Michel, R. Muscheler, J. Beer, Geomagnetic field intensity, North Atlantic Deep Water circulation and atmospheric delta C-14 during the last 50 kyr, Earth and Planetary Science Letters 200 (2002) 177–190. [87] M.G. Leavesley, M.I. Bird, L.K. Fifield, P.A. Hausladen, G.M. Santos, M.L. di Tada, Buang Merabak: early evidence for human

851

occupation in the Bismarck Archipelago, Papua New Guinea, Australian Archaeology 54 (2002) 55–57. [88] P.S. Martin, Clovisia the beautiful, Natural History 96 (10) (1987) 10–13. [89] S. McBrearty, A.S. Brooks, The revolution that wasn’t: a new interpretation of the origin of modern human behavior, Journal of Human Evolution 39 (2000) 453–563. [90] T.D. McCown, A. Keith, The Stone Age of Mount Carmel II, Oxford University Press, Oxford, 1939. [91] D. Meltzer, Clocking the first Americans, Annual Review of Anthropology 24 (1995) 21–45. [92] D.J. Meltzer, D.K. Grayson, G. Ardila, A.W. Barker, D.F. Dincauze, C.V. Haynes, F. Mena, L. Nunez, D.J. Stanford, On the Pleistocene antiquity of Monte Verde, southern Chile, American Antiquity 62 (1997) 659–663. [93] N. Mercier, H. Valladas, O. Bar-Yosef, B. Vandermeersch, C. Stringer, J.-L. Joron, Thermoluminescence date for the Mousterian burial site of es-Skhul, Mt. Carmel, Journal of Archaeological Science 20 (1993) 169–174. [94] G.H. Miller, J.W. Magee, B.J. Johnson, M.L. Fogel, N.A. Spooner, M.T. McCulloch, L. Ayliffe, Pleistocene extinction of Genyornis newtoni: Human impact on Australian megafauna, Science 283 (1999) 205–208. [95] M. Morwood, S. L’Oste-Brown, D. Price, Excavations at Mushroom Rock, in: M. Morwood, D. Hobbs (Eds). Quinkan Prehistory: The Archaeology of Aboriginal Art in S.E. Cape York Peninsula, Australia, Tempus 3, Anthropology Museum, Department of Anthropology and Sociology, University of Queensland, 1995, pp. 133–146. [96] D.J. Mulvaney, The Stone Age of Australia, Proceedings of the Prehistoric Society 27 (1961) 56–107. [97] J.F. O’Connell, J. Allen, When did humans first arrive in Greater Australia, and why is it important to know? Evolutionary Anthropology 6 (1998) 132–146. [98] S. O’Connor, P. Veth, C. Campbell, Serpent’s Glen rockshelter: report of the first Pleistocene-aged occupation sequence from the Western Desert, Australian Archaeology 46 (1998) 12–22. [99] S. O’Connor, K.P. Aplin, M. Spriggs, P. Veth, L.K. Ayliffe, From savannah to rainforest: changing environments and human occupation at Liang Lemdubu, Aru Islands, Maluku (Indonesia), in: P. Kershaw, B. David, N. Tapper, D. Penny, J. Brown (Eds). Bridging Wallace’s Line: The Environmental and Cultural History and Dynamics of the SE-Asian-Australian Region, Advances in Geoecology 34, Catena Verlag, Reiskirchen, 2002, pp. 279–306. [100] S. O’Connor, M. Spriggs, P. Veth, Excavation at Lene Hara Cave establishes occupation in East Timor at least 30,000– 35,000 years ago, Antiquity 76 (2002) 45–50. [101] P. Ossa, B. Marshall, C. Webb, New Guinea II cave: a Pleistocene site on the Snowy River, Victoria, Archaeology in Oceania 30 (1995) 22–35. [102] C. Pavlides, C. Gosden, 35,000-year-old sites in the rainforests of West New Britain, Papua New Guinea, Antiquity 68 (1994) 604–610. [103] R.H. Pearce, M. Barbetti, A 38,000-year-old archaeological site at Upper Swan, Western Australia, Archaeology in Oceania 16 (1981) 173–178. [104] D.A. Richards, J.W. Beck, Dramatic shifts in atmospheric radiocarbon during the Last Glacial period, Antiquity 73 (2001) 482–485. [105] R.G. Roberts, Luminescence dating in archaeology: from origins to optical, Radiation Measurements 27 (5/6) (1997) 819–892. [106] R.G. Roberts, The last Australian megafauna, Australasian Science 22 (9) (2001) 40–41. [107] R.G. Roberts, R. Jones, Luminescence dating of sediments: new light on the human colonization of Australia, Australian Aboriginal Studies 2 (1994) 2–17.

852

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853

[108] R.G. Roberts, R. Jones, Chronologies of carbon and silica: evidence concerning the dating of the earliest human presence in northern Australia, in: P.V. Tobias, M.A. Raath, J. Moggi-Cecchi, G.A. Doyle (Eds.), Humanity from African Naissance to Coming Millennia: Colloquia in Human Biology and Paleo-Anthropology, Florence University Press, Florence, 2000, pp. 243–252. [109] R.G. Roberts, R. Jones, M.A. Smith, Thermoluminescence dating of a 50,000 year old human occupation site in northern Australia, Nature 345 (1990) 153–156. [110] R.G. Roberts, R. Jones, M.A. Smith, Stratigraphy and statistics at Malakunanja II: reply to Hiscock, Archaeology in Oceania 25 (1990) 125–129. [111] R.G. Roberts, R. Jones, M.A. Smith, Early dates at Malakunanja II: a reply to Bowdler, Australian Archaeology 31 (1990) 94–97. [112] R.G. Roberts, R. Jones, N.A. Spooner, M.J. Head, A.S. Murray, M.A. Smith, The human colonization of Australia: optical dates of 53,000 and 60,000 bracket human arrival at Deaf Adder Gorge, Northern Territory, Quaternary Science Reviews 13 (1994) 575–583. [113] R.G. Roberts, N.A. Spooner, R. Jones, S. Cane, J.M. Olley, A.S. Murray, J. Head, Preliminary luminescence dating for archaeological sediments on the Nullarbor Plain, South Australia, Australian Archaeology 42 (1996) 7–16. [114] R.G. Roberts, M. Bird, J. Olley, R. Galbraith, E. Lawson, G. Laslett, H. Yoshida, R. Jones, R. Fullagar, G. Jacobsen, Q. Hua, Optical and radiocarbon dating at Jinmium rock shelter in northern Australia, Nature 395 (1998) 358–362. [115] R.G. Roberts, H. Yoshida, R. Gailbraith, G. Laslett, R. Jones, M. Smith, Single aloquot and single grain optical dating confirm thermoluminescence age estimates at Malakunanja II rock shelter in northern Australia, Ancient TL 16 (1998) 19–24. [116] R.G. Roberts, T.F. Flannery, L. Ayliffe, H. Yoshida, J.M. Olley, G.J. Prideaux, G.M. Laslett, A. Baynes, M.A. Smith, R. Jones, B.L. Smith, New ages for the last Australian megafauna: continent-wide extinction about 46,000 years ago, Science 292 (2001) 1888–1892. [117] J.G. Robinson, Lake Mungo: An analysis of the surface collection. MA thesis, Department of Prehistory and Anthropology, The Faculties, Australian National University, 1980. [118] H.P. Schwarcz, R. Grun, B. Vandermeersch, H. Valladas, E. Tchernov, ESR dates for the hominid burial site of Qafzeh in Israel, Journal of Human Evolution 17 (1988) 733–737. [119] H.P. Schwarcz, W.M. Buhay, R. Grun, H. Valladas, E. Tchernov, O. Bar-Yosef, B. Vandermeersch, ESR dating of the Neanderthal Site, Kebara Cave, Israel, Journal of Archaeological Science 16 (1989) 653–659. [120] W. Shawcross, Archaeological excavations at Mungo, Archaeology in Oceania 33 (1998) 183–200. [121] W. Shawcross, M. Kaye, Australian archaeology: implications of current interdisciplinary research, Interdisciplinary Science Reviews 5 (1980) 112–118. [122] G. Singh, A.P. Kershaw, R. Clark, Quaternary vegetation and fire history in Australia, in: A.M. Gill, R.H. Groves, I.R. Noble (Eds.), Fire and the Australian Biota, Australian Academy of Science, Canberra, 1981, pp. 23–54. [123] M. Slack, R. Fullagar, J. Field, A. Border, P. Tacon, M. Archer, Pleistocene human occupation in the Riversleigh World Heritage Area in northwest Queensland (in prep.). [124] M.A. Smith, N. Sharp, Pleistocene sites in Australia, New Guinea and island Melanesia: geographic and temporal structure of the archaeological record, in: M. Smith, M. Spriggs, B. Fankhauser (Eds). Sahul in Review: Pleistocene Archaeology in Australia, New Guinea and Island Melanesia, Occasional Papers in Prehistory 24, Department of Prehistory, Research School of

[125]

[126]

[127]

[128]

[129]

[130]

[131]

[132]

[133]

[134]

[135]

[136]

[137] [138]

[139]

[140] [141]

[142]

Pacific Studies, Australian National University, Canberra, 1993, pp. 37–59. M.A. Smith, J.R. Prescott, M.J. Head, Comparison of 14C and luminescence chronologies at Puritjarra rock shelter, central Australia, Quaternary Science Reviews 16 (1997) 299–320. M.A. Smith, M.J. Bird, C.S.M. Turney, L.K. Fifield, G.M. Santos, P.A. Hausladen, M.L. di Tada, New ABOX AMS-14C ages remove dating anomalies at Puritjarra rock shelter, Australian Archaeology 53 (2001) 45–47. N. Stern, J. Allen, Pallawa Trounta Shelter, in: J. Allen (Ed.), Report of the Southern Forests Archaeological Project, vol. 1: Site Descriptions, Stratigraphies and Chronologies, School of Archaeology, La Trobe University, 1996, pp. 169–193. C. Stringer, Modern human origins: progress and prospects, Philosophical Transactions of the Royal Society, series B 357 (2002) 563–579. E. Tchernov, Biochronology, paleoecology and dispersal events of hominids in the southern Levant, in: T. Akazawa, K. Aoki, T. Kimura (Eds.), The Evolution and Dispersal of Modern Humans in Asia, Hokusen-Sha, Tokyo, 1992, pp. 149–188. P.B. Thorley, Pleistocene settlement in the Australian arid zone: occupation of an inland riverine landscape in the central Australian ranges, Antiquity 72 (1998) 34–45. A.G. Thorne, R. Grun, G. Mortimer, N.A. Spooner, J.J. Simpson, M. McCulloch, L. Taylor, D. Curnoe, Australia’s oldest human remains: age of the Lake Mungo 3 skeleton, Journal of Human Evolution 36 (1999) 591–612. R. Torrence, Archaeological Fieldwork in West New Britain, PNG, June–July 2001, report prepared for the Papua New Guinea Museum and Art Gallery, the West New Britain Provincial Government and New Britain Palm Oil Ltd., Mahonia Na Dari, 2001. C.S.M. Turney, M.I. Bird, Determining the timing and pattern of human colonization in Australia: proposals for radiocarbon dating “early” sequences, Australian Archaeology 54 (2002) 1–5. C.S.M. Turney, M.I. Bird, L.K. Fifield, R.G. Roberts, M. Smith, C.E. Dortch, R. Grun, E. Lawson, L.K. Ayliffe, G.H. Miller, J. Dortch, R.G. Cresswell, Early human occupation at Devil’s Lair, southwestern Australia, 50,000 years ago, Quaternary Research 55 (2001) 3–13. H. Valladas, J.L Joron, G. Valladas, P. Arensberg, O. Bar-Yosef, A. Belfer-Cohen, P. Goldberg, H. Laville, L. Meignen, Y. Rak, E. Tchernov, A.-M. Tillier, B. Vandermeersch, Thermoluminescence dates for the Neanderthal burial site at Kebara (Mount Carmel, Israel), Nature 330 (1987) 159–160. H. Valladas, J.L. Reyss, J.L. Joron, G. Valladas, O. Bar-Yosef, B. Vandermeersch, Thermoluminescence dating of Mousterian ‘Cro-Magnon’ remains from Israel and the origin of modern man, Nature 331 (1988) 614–616. B. Vandermeersch, Les Hommes Fossiles de Qafzeh (Israel), Centre National de la Recherche Scientifique, Paris, 1981. J. Van der Plicht, Calibration of the C-14 time scale: towards a complete dating range, Geologie en Mijnbouw-Netherlands Journal of Geosciences 81 (2002) 85–96. A.H.L. Voelker, P.M. Grootes, M.J. Nadeau, M. Sarnthein, Radiocarbon levels in the Iceland Sea from 25–53 kyr and their link to the earth’s magnetic field intensity, Radiocarbon 42 (2000) 437–452. D.C. Wallace, M.T. Lott, “MITOMAP: A Human Mitochondrial Genome Database,” http://www.mitomap.org, 2003. K. Walshe, Taphonomy of Mungo B assemblage: indicators for subsistence and occupation of Lake Mungo, Archaeology in Oceania 33 (1998) 200–206. X. Wang, S. van der Kaars, A.P. Kershaw, M. Bird, F. Jansen, A record of fire, vegetation and climate through the last

J.F. O’Connell, J. Allen / Journal of Archaeological Science 31 (2004) 835–853 three glacial cycles from Lombok Ridge core G6-4, eastern Indian Ocean, Indonesia, Palaeogeography, Palaeoclimatology, Palaeoecology 147 (1998) 241–256. [143] J.P. White, J.F. O’Connell, A Prehistory of Australia, New Guinea and Sahul, Academic Press, Sydney, 1982. [144] M.A.J. Williams, Termites, soils and landscape equilibrium in the Northern Territory of Australia, in: J.L. Davies, M.A.J. Williams (Eds.), Landform Evolution in Australasia, Australian National University Press, Canberra, 1978, pp. 128–141. [145] S. Wroe, J. Field, R. Fullagar, L.S. Jermin, Megafaunal extinctions in the Late Quaternary and the global overkill hypothesis, Alcheringa (2003) in press.

853

[146] Y. Yokoyama, T.M. Esat, K. Lambeck, L.K. Fifield, Last ice age millennial timescale climate changes recorded in Huon Peninsula corals, Radiocarbon 42 (2000) 383–401. [147] M. Zuraina, H.D. Tjia, K. Tampan, Perak: The geological and archaeological evidence for a late Pleistocene site, Journal of the Malaysian Branch of the Royal Asiatic Society 61 (1988) 123–134.