Aridity and settlement in northwest Australia.

Veth, Peter

An element in the changing pattern of Australian archaeology has been the filling-in of great blanks on the archoeological map, once survey and excavation has begun to explore them. The dry lands of the great central and western deserts of Australia, a hard place for humans to this day, have in the lost couple of decades come to find a large place in the transitional story.

The arid northwest

This paper focusses on the northwest portion of Australia incorporating the Gascoyne, Pilbara, Western Desert and Kimberley regions (Figure 1). This huge area has changed over the last 20 years from the archaeological `blackhole' reviewed by Dortch (1977), to one providing some of the oldest occupation sequences for both the continent and its offshore islands (O'Connor in press, pers. comm.; Veth 1994). These results are not unexpected given the region's propinquity to proposed entry routes for colonizers out of island southeast Asia and into Sahul (Birdsell 1977).

This huge region has many diverse landform, vegetation and climate types (Brown 1987; Harrison 1993; O'Connor 1993; Veth 1993. The Gascoyne and Western Desert, generally semi-arid to arid, have highly weathered and often subdued landscapes, with locally prominent ranges. In the major difference between the two regions, the Gascoyne has numerous and sometimes major ephemeral water-courses that harbour important concentrations of food and water resources, while the Western Desert is characterized by uncoordinated drainage, and resource distribution is more scattered. The Pilbara region, also with an arid to semi-arid climate, contains a major upland system (e.g. the Hamersley and Chichester ranges) which gives rise to substantial drainage courses, such as the Fortescue and de Grey. These discharge over extensive coastal plains to enter the Indian Ocean along an otherwise featureless arid coastline. Vegetation of these arid areas is predominantly tree/shrub steppe and hummock grassland.

In contrast to these regions, which experience as little as 250 mm of rain per annum, the Kimberley is characterized by substantially higher rainfall (up to 1400 mm) with a markedly seasonal distribution. The lowest rainfall occurs in the southwest Kimberley, which is generally featureless and flat; it increases substantially as the numerous and sometimes massive ranges to the north and east are encountered. Drainage is well co-ordinated, and in many cases, the discharge volumes of rivers are very high (Lau et al. 1987). Vegetation is much more dense and includes extensive woodlands, major mangrove forests and even patches of rainforest.

An intriguing pattern in the prehistoric record of these varied regions is the evidence for a changing nature of occupation in stratified sites spanning the Pleistocen - Holocene transition. Varied interpretations of these changes have given rise to a number of demographic models (e.g. Hiscock 1988; Morse 1993a; O'Connor 1990; O'Connor et al. 1993; Smith 1988; Veth 1993).

Climates and environments

These varied models see major changes in climate and sea-levels associated with the Last Glacial Maximum as restructuring resources significant to human ecology. Changes in demography and settlement behaviour ensued, with implications for prehistoric exchange systems, local social organization and dietary composition. The period 15,000-7000 b.p. is the time of greatest change in certain archaeological signatures (after Gould 1980), as witnessed in marked fluctuations in regional cultural discard rates, long-distance exchange systems and inferred alterations to resource catchments.

While palaeoclimatic reconstructions for northwest Australia are sketchy, it is reasonable to propose that the arid conditions experienced today were exacerbated during the Last Glacial Maximum with climatic amelioration possibly occurring as late as the early Holocene. Areas almost certainly affected include the Gascoyne, Pilbara and Western Desert. For the period under review, portions of the interior probably experienced lowered temperatures and precipitation, and increased evaporation and seasonality, with resulting reduction in plant cover. This reactivated old dune fields and generated new ones.

Climatic amelioration may have begun by 12,000 b.p., although some data indicate that the last major arid phase may have continued until as late as 7000 b.p. (A. Chappel pers. comm.; Bowler & Wasson 1984; Harrison 1993; Jennings 1975; Jones & Bowler 1983; O'Connor et al. 1993; Pederson 1983; Wasson 1984; Wyrwoll 1979). The results of general circulation models (glacial mode) suggest that the central Kimberley region may have been less affected during this period (Harrison 1993; Hubbard n.d.).

The extensive areas covered by the northern half of the Western Desert, the Gascoyne and the Pilbara regions fall within the arid zone, as defined by Mabbutt (1971; see also Harrison 1993). Importantly, the coastal margin north from Shark Bay to south of Broome comprises the longest arid marine zone in the continent; where the desert meets the sea (O'Connor & Veth 1993) in one of the largest arid coastal zones in the world (cf. Stafford-Smith & Morton 1990). Greater Australia, as one of the most arid landmasses colonized by modern humans, has likely always had an extensive arid coast-line (Veth in press a). Early colonizers were almost certainly coastally adapted and employed more than casual maritime skills (Gosden 1993; & this volume). Therefore, the archaeological record of these arid Pleistocen - Holocene coast-lines is of immense interest, particularly for the dynamics of arid coastal and hinterland resource exploitation and settlement in the face of major changes in climate and sea-level.

The timing for, and the nature of, permanent occupation of the desert lowlands of the Gibson and Great Sandy Deserts is also of critical interest. Most reviewers agree that desert lowlands would have represented a marginal habitat during the period of increased aridity associated with the Last Glacial Maximum, with resulting changes in settlement patterns, and even regional abandonment occurring in favour of better-watered `refuge' areas, such as montane uplands (Ross & Wasson 1992; Smith 1988; Veth 1993). In one current view, desert lowlands were immediately (re)-colonized from refuges, following climatic amelioration (Smith 1993; Smith et al. 1991); according to another model - the so-called `clinal model' - the lowlands would have been (re)colonized gradually (after Veth in press b; 1993; Hiscock 1994).

Zones of refuge are still generally defined as areas providing reliable networks of (more) permanent waters. The suggestion that the now-submerged Pleistocene coast-line served as a major fall-back area (cf. Morse 1994; Veth 1993) gains support from recent research in northwest Australia (see below). Other regions which may have constituted fall-back areas during the Glacial Maximum include the Pilbara and the Kimberley uplands (Figure 1).

Inland occupation: the Gascoyne, the

Pilbara and the Western Desert

Sites dating to over 20,000 years ago in the arid uplands of the Pilbara have been located at an increasing rate since the late 1970s (Brown 1987; Hughes 1992; Murphy 1994). Table 1 lists available dates. The currently oldest published date of 26,300[+ or -]500 b.p. for Newman Rockshelter, on charcoal associated with artefacts, comes from at least 70 cm above the base of the deposit, which has yet to be excavated. As the reported dates for Pilbara Upland rockshelter sites in Table 1 illustrate, none fall between 17,900 b.p. and 9870o b.p. Despite this 800-year gap, various authors have interpreted the unequivocal presence of artefacts, and often organics, in all intervening units bracketed by these dates as representing intermittent occupation (e.g. Brown 1987; Smith & Sharp 1993; Veth 1993).

[TABULAR DATA OMITTED]

Brown (1987) makes a distinction between core and marginal areas in the Pilbara uplands. The Newman Rockshelters (P2055.2 & P0187) lying on major drainage lines are seen to be in more optimal locations than, for example, the Packsaddle Ridge rock-shelters (FIGURE 1). In the wider regional context, Brown (1987: 55) sees the Pilbara Uplands as an optimal zone: `the Hamersley Plateau, being a mountain and piedmont desert landscape ... represents one of the least harsh of the desert physiographic types for human settlement, since water availability is comparatively reliable'.

Importantly, Brown predicts that human adaptation to this environmental zone would be a prerequisite for the permanent colonization of sandy and stony deserts. I have also previously made this point (Veth 1987: 109):

I would suggest an early- to mid-Holocene expansion of humans from the well-defined gorges of the Pilbara into the hummock grasslands of the Sandy Deserts.... The sandy and stony deserts lying in between and lacking co-ordinated drainage, may have been more difficult to utilize during the last major arid phase.

The concept of gradual and permanent colonization of marginal lands after the last major arid phase, accompanied by technological and socio-economic transformations aimed at reducing the effects of environmental stochasticity, finds wide support (e.g. Hiscock in press; Ross & Wasson 1992; Veth 1993).

Two major programmes of regional survey and excavation have been conducted within the northern half of the Western Desert in the Gibson and Great Sandy deserts, respectively (Gould 1977; Veth 1993). The type-site for the conservative Australian desert culture', Puntutjarpa Rockshelter, provides lithic and faunal assemblages and living surfaces which were argued by Gould (1977; 1980) to represent remarkable continuities for a 10,000-year span of occupation. Re-analysis of the dating for the introduction of hafted (adze) technology to the mid Holocene suggests, however, that changes in both types and proportions of extractive tools have occurred at Puntutjarpa and other arid-zone sites of similar antiquity (e.g. Brown 1987; Hiscock & Veth 1991).

Also challenging the notion of an unchanging record is evidence from the massive dunefields of the Great and Little Sandy deserts (and from the central Australian Ranges) for increases in the rate of cultural discard in sites during the late Holocene (Veth 1989; Smith 1988). Dates for sites excavated within the northern half of the Western Desert are listed in TABLE 1. All of these desert lowland sites post-date the time of presumed climatic amelioration, providing limited support for the clinal model of lowland occupation.

Coastal occupation

Coastal sequences of the same order of antiquity as sites located within the Pilbara uplands were less forthcoming, although now there is a range of early evidence for human occupation of arid coast-lines from Shark Bay through to the southwest Kimberley. The most detailed economic evidence comes from the Pilbara coast-line, where relict Pleistocene land surfaces have been discovered proximal to the sea (see Bowdler 1992; Morse 1993a; 1993c; 1994; O'connor 1990; Veth 1994). Pertinent dates for the Gascoyne and Pilbara coast-line are presented in TABLE 1.

On the Cape Range Peninsula, Morse (1994) has excavated three rock-shelters which, she argues, together provide clear evidence for occupation between c. 34,000 and c. 17,500 b.p. and then from c. 12,000 b.p. to the late Holocene. Other open middens associated with palaeo-lagoons and relict shore-lines provide additional dates in the period between c. 20,000 and c. 5500 b.p. (Kendrick S, Morse 1982; 1993a). Mandu Mandu Rockshelter has a hiatus in occupation/deposition between c. 12,000 and c. 5500 b.p. Unfortunately the dating of critical deposits at Pilgonaman Creek Rockshelter between c. 17,500 and c. 12,000 b.p. is problematic due to a number of dating inversions. Although Morse (1994: 210) considers Pilgonaman Rockshelter likely to have been visited on an occasional basis during this period, support for this interpretation is minimal.

Marine fauna is always present in these deposits. Morse (1993b) has reasoned that as sea-level fell towards the height of the Last Glacial Maximum and the Pleistocene coast retreated up to 12 km west of the sites, the frequency of visits probably decreased. The Holocene records for Mandu Mandu, Pilgonaman and Yardie Well rock-sheiters demonstrate the continuation of a broadly-based marine diet. This is best explained by the relative propinquity of the sites to rich marine zones, because of the unusually steep declination of the continental shelf at this point.

There is good archaeological evidence for the exchange of cultural items from the interior portions of the Pilbara to the Pleistocene coast-line at Cape Range, but only before the height of the Last Glacial Maximum and following climatic amelioration at c. 10,000 b.p. Morse (1993b) notes ochres in deposits from Mandu Mandu until approximately 20,000 b.p., their absence during the Glacial Maximum from Pilgonaman, and then their re-appearance by 10,000 b.p. at Pilgonaman, Yardie Well and later at Mandu Mandu. The likely source is from the Chichester Ranges in the Pilbara uplands. The absence of ochres during the Glacial Maximum is seen to be the result of the breakdown of long-distance networks (see also Veth 1993).

The Cape Range sites demonstrate exploitation of the Pleistocene-Holocene coast-line, however with a considerably lower intensity of occupation between c. 17,500 and c. 10,000 b.p, than before or after this period (see artefact discard rates presented in O'connor et al. 1993). This pattern can be interpreted as groups persisting on the terminal Pleistocene coastline, as a fall-back zone, and decreasing their systematic use of the present hinterland due to the combined effects of increasing aridity and distance (see also Morse 1993b: 163).

The Silver Dollar Site from Shark Bay provides evidence for occupation of the arid interior during the Last Glacial Maximum (see TABLE 1). At this time the coast would have been located at least 100 km to the west. The presence of baler shell (Melo sp.) indicates some form of contact with the coast. However occupation ceases by c. 18,500 b.p. and is only registered again at this site between c. 7500 and c. 6500 b.p. (Bowdler 1990; 1992). The later upper unit of this site and other open surface middens in the region contain shellfish which have been taken from mangrove habitats. There is currently, therefore, no evidence for occupation during the Pleistocene - Holocene transition period under review.

Occupation on present@day islands

Evidence for early use of the drowned Northwest Shelf has come from excavations on the Montebello Islands (Veth 1994, now located c. 100 km offshore. Three rock-shelter deposits provide direct evidence for the exploitation of a wide range of terrestrial and marine faunae, with a basal occupation of c. 27,000 b.p. and for subsequent occupation between c. 10,000 and c. 7500 b.p., after which the islands are assumed to have been abandoned. The radiocarbon determinations are presented in TABLE 1.

[TABULAR DATA OMITTED]

These excavations indicate that the Montebello Islands were first used when they were part of the mainland, before 27,000 years ago (FIGURE 2). In Noala 2 Cave, apart from marine valves (Polymesoda coaxans) and fish vertebrae, the fauna during the pre-Glacial Maximum phase is largely terrestrial. Only sparse cultural remains separate the early date and a radiocarbon determination of 10,030[plus or minus]200 b.p. from the adjacent excavation unit above, while the next firmly dated occupations are from the adjacent Noala 1 Cave site and the near-by Hayne's Cave 2, 3 and 4 excavations (FIGURE 2).

Occupation dates for these sites all fall between c. 10,000 and c. 7500 b.p., the period when the coast lies within 4 km of the caves and sea-level rises to approximately - 10m. At c. 10,000 b.p. the islands comprised the most northerly portion of a large peninsula. From c. 8500 to 7500 b.p. the Montebello Islands would have been joined to Barrow Island and other smaller islands to form a large land-mass separated from the mainland by the Mary Anne Passage (FIGURE 2). It is likely that Aboriginal groups would have travelled across this 5 km sea-gap. While this large island appears to have been used for a millennium after its initial separation from the mainland, all current evidence from its present-day component islands points to final abandonment soon after 7500 b.p.

During this latest phase of occupation, the abundant shell remains indicate a strong reliance on mangrove species. Other marine fauna include a range of fish such as wrasse, parrot-fish and sea bream, and also crustaceans and turtles. The terrestrial fauna is highly diverse, with a bias towards medium-sized game such as bettongs, bandicoots and wallabies. Significantly, the diversity and richness of medium-sized mammals is higher than on the adjacent Cape Range Peninsula (Baynes SE Jones 1992). These species are likely to have been part of a now submerged plains or extended dune-fields fauna (Veth 1994).

The Pleistocene-Holocene transitionary assemblages from the Montebello Islands provide good evidence for access to a range of both terrestrial and marine habitats. Woodroffe's (1990) arguments that the rising sea-level may not necessarily have been a destructive phenomenon, and hence caused little stress to coastal occupants, is supported by faunal data from the Montebello Islands. The existence of rocky foreshores is indicated by chiton and limpet in the middens and of extensive coral flats by large parrot-fish. Intertidal mud-flats have always been available, as species such as Polymesoda sp. occur continuously from the earliest levels.

Finally, with reference to mangrove development, the first regional evidence comes from pollen samples from three cores located to the north of the Sahul shelf (van de Kaars 1991) where a relative date of 12,000 b.p. was obtained. Firm evidence for mangrove communities comes from the Montebello Islands with the continuous presence of the gastropods Terebralia sp. and Telescopium sp., mud crab (Scylla serrata) and lobster, all dating to between c. 10,000 and c. 7500 b.p. (Pulsford 1994; Veth 1994). These species were probably collected from the protected landward side of the Montebello-Barrow combined island. There is, therefore, no sound reason to believe that the arid Pleistocene coast-line of northwest Australia was necessarily poor in food resources.

Table 2 presents a summary of all economic fauna identified from Hayne's Cave Square 4 and Noala Caves 1 and 2. Line breaks in this figure mark assumed transitions in the location of the sites from the mainland, to the Barrow-Montebello large island phase and, finally, to their current configuration as small island outliers.

[TABULAR DATA OMITTED]

Before final insulation, the assemblages contain a wide range of terrestrial fauna including wallaby, bettong, possum, bandicoot, bilby, snake and lizard. There is also a marine component in the oldest Noala Cave 2 unit, where Polymesoda sp. and fish vertebrae are dated to before c. 27,000 b.p., when the Pleistocene coast was nearer the site. A wider range of marine fauna then appears in the near-by sites between c. 10,000 and c. 8500 b.p., including Terebralia palustris, Tridacna sp., Melo amphora and crustaceans.

The relative contribution of marine resources then increases markedly after c. 8500 b.p., during the large island phase, without the loss of terrestrial faunal species. Marine fauna include a wide range of mangrove and rocky-substrate molluscs, crustaceans, fish, turtles and sea-urchins. In the last phase of island occupation the terrestrial base is severely depleted (to the right of the second line in Table 2), a marked change best explained by the drowning of coastal plains which supported these fauna.

Although these sites require further excavation and dating, a familiar pattern emerges. The volumes of cultural material and assumed rates of cultural deposition are low, if indeed present at all, during the Last Glacial Maximum (Veth et al. in prep.). This is again likely to reflect a combination of intensified aridity and increased distance from the Pleistocene coast during the period of significantly lower sea-levels.

Analysis of aeolian-derived clays from solid samples taken in Hayne's Cave indicates the gradual but consistent decrease of initially dominant smectite clays with a reciprocal increase of kaolinite (Veth et al. in prep.). This suggests a change in climate from very arid, to less arid with warmer and wetter conditions. The dates for these samples suggest that arid conditions may have persisted until the early Holocene, after which the densest deposits in the Montebello sites occur.

There are no further inferred occupation sequences or indeed radiocarbon determinations for the entire arid coast-line during the period 15,000 to 7000 b.p. At least another 28 rock-shelters and open middens have been excavated and/or sampled with a total of 69 dates obtained (Table 1). Sites are located on the Burrup Peninsula, the Pilbara coast on either side of the Burrup Peninsula and in the Shark Bay area (Bowdler 1990; 1992; Bradshaw pers. comm.; Lorblanchet 1977; Veitch & Warren 1992; Veth & O'Brien 1986; Vinnicombe 1987). Most of these sites are associated with post-transgression marine catchments with the earliest, Site P2772 on the Burrup Peninsula, dating from 6740 [+ or -] 130 b.p.

A broad regional trend is seen in the early dominance of mangrove gastropods such as Terebralia palustris and Turbo cinereus, changing to the ubiquitous Anadara granosa by c. 3600 b.p. The type-sequence from Skew Valley (Lorblanchet 1977) is replicated in a number of other sites in the region. The inferred mid- to late-Holocene retraction of mangrove communities along extensive areas of the arid western coast-line seems to be a common phenomenon which follows Woodruffe's (1990) `big swamp' phase.

Occupation in the Kimberley region

For the present analysis, the Kimberley region can be divided into a southwest region which (after Jones & Bowler 1980:10) likely experienced arid conditions during the transitional period, and the large remaining areas incorporating the uplands of the Mitchell Plateau and east Kimberley.

Southwest Kimberley

Regional survey and excavation by O'Connor (1990; in press) on the southwest Kimberley coastal margin has located the Widgingarri and Koolan 2 rock-shelter sites dated to c. 28,000 b.p. and containing significant deposits below the dated levels (see Table 1).

When first occupied, the Koolan Shelter 2 was located near the Pleistocene coast-line, while the Widgingarri shelters were situated at least 50 km inland. O'Connor (190:344) shows that the Pleistocene levels at the sites reflect comparatively intensive occupation, seen in high sedimentation and artefact discard rates; there is evidence for the early emergence of distinctive regional economies and long-distance movement of materials.

Long-distance trade with the coast between c. 28,000 and c. 17,000 b.p. comes in the form of baler shell (Melo sp.), pearl shell (Pinctada sp.) and mud clam (Polymesoda coaxans). O'Connor (1990:349) has argued that the dates for Koolan illustrate its abandonment by c. 23,000 b.p. followed by its reoccupation at 10,850 [+ or -] 160 b.p. The Widgingarri shelters replicate this trend, but abandonment occurs later and closer to the height of the Last Glacial Maximum. The inference that this gap between c. 19,000 and c. 11,000 b.p. represents a hiatus in occupation is supported by a lack of lag deposits or any obvious unconformity in the sites.

Whether abandonment follows regional aridity, a retreating sea, or both, is difficult to determine. Faunal analysis from Widgingarri provides evidence for truly arid species, such as bilby (Macrotis sp.) in spits immediately predating the height of the Glacial Maximum (Prince 1991). What is clear, however, is that the sites are reoccupied when the sea is again near by. Because of the steep offshore profile at Koolan, this is at c. 10,500 b.p. at Widgingarri; with a lower declination slope, this occurs at c. 7500 b.p.

Significantly, both the Widgingarri shelters and Koolan have much lower artefact and sediment rates during the early Holocene (when adjusted for volume) than they do during the final period prior to their abandonment (O'Connor et al. 1993:99). It is not until the late Holocene that artefact discard rates exceed average rates registered prior to abandonment.

Excavation of a rock-shelter and large open midden on the small and adjacent High Cliffy Island has revealed early island occupation from 6700 b.p. (O'Connor in press). Reduction debris from local silcretes occurs in the shelter between dates of c. 3500 and c. 2500 b.p.; it reflects a much higher rate of production than quartz reduction debris recorded from the earlier units of Koolan (O'Connor 1990). Over the 30,000-year time-span that the west Kimberley has evidence for occupation, the time of greatest change lies within the Pleistocene-Holocene transition. This appears to go beyond simply adjustments to sea-level fluctuations.

The only other dated sequences from the southwest Kimberley come from recent excavations south of Broome and on the Dampierland Peninsula (O'Connor & Veth 1993). Mainly based on reworked linear and mound middens, these have returned dates between c. 3500 and c. 1000 b.p. Such middens of late-Holocene age appear to be very common along this semi-arid coast-line. Many middens which were likely formed during the mid Holocene have probably been destroyed by geomorphic processes.

East Kimberley and Mitchell Plateau

The excavations of Miriwun and Monsmont Shelters in the Ord Valley, east Kimberley (Table 1) provide evidence for a Pleistocene-aged riverine adaptation which has been argued to display continuities through time (Dortch 1977). Late-phase lithic technologies at c. 3000 b.p., such as prismatic blade production and invasive flaking, are seen as added components to an already well-established maintenance tool kit (Dortch 1977:123). Economic fauna including murids, marsupials, reptiles, fish, birds and freshwater molluscs alter little (in kind) over 18,000 years of occupation. Although Miriwun has at least one stratigraphic unconformity, in combination with Monsmont the sites attest to use of riverine resources at the height of the Glacial Maximum and afterwards during the late Holocene. Excavations of other rock-shelters in the region (Table 1) such as Kununurra, Pincombe Range, Pilchowski Crossing and Canyon have all provided recent dates falling between c. 3500 and >1000 b.p.

On the Mitchell Plateau, Veitch (pers. comm.) has excavated six rock-shelters and open middens, some of the latter taking the form of extremely large monospecific shell mounds. All sites are Holocene with dates falling between c. 6000 and c. 1500 b.p. (Table 1).

On the available evidence, major change in cultural assemblages from Kimberley sites dating from the Pleistocene-Holocene transition is seen only in those sites from the southwest region where increased aridity associated with the Last Glacial Maximum is likely to have been an impinging factor on human ecology.

Northwest Australian regions during the

transition

Any characterization of regional patterns based on a limited number of excavations and sampling points will have limitations; however this should not deter archaeologists from identifying emerging and consistent patterns in regional prehistories (cf. Hiscock 1988;) Smith & Sharp 1993; Veth 1993). At the very least these empirical generalizations can give rise to alternative mid-range theories, which can be tested, and which might otherwise have only originated from general theory (after Trigger 1989:22).

With reference to the various `demographic' models discussed earlier (Brown 1987; Hiscock 1988; Morse 1993a; O'Connor 1990; Smith 1988; Veth 1993), the regional syntheses illustrate several general trends in occupation deposits during the Pleistocene-Holocene transition. * Sites located at some time within or adjacent to arid lowlands provide evidence for discontinuity of occupation either coinciding, or substantially overlapping, with the Last Glacial Maximum. Sites characterized as having a hiatus in occupation include the Widgingarri and Koolan 2 rock-shelters in the southwest Kimberley and the Silver Dollar Site at Shark Bay. It may also be that parts of the sequences in the Montebello sites demonstrate a hiatus in occupation coinciding with the Last Glacial Maximum, when the coast is distant. This interpretation must await further dating and excavation of the Noala 2 sequence. * Where cultural deposits are present, demonstrably lower artefact discard and (in some cases) sediment accumulation rates occur during the Pleistocene-Holocene transition period, than in periods before or after. Sites which show this trend include Pilgonaman Rockshelter and the Widgingarri and Koolan 2 rock-shelters (O'Connor et al. 1993). It might be that accumulation rates are also reduced during this time at the Hamersley rock-shelters if the vertical closeness of widely separated dates is considered. * The existence of regional economies and long-distance exchange networks in the Pleistocene has been advocated (e.g. O'Connor 1990). Long-distance transport of marine molluscs, ochres and probably some lithics occurred before the transition period in a number of the northwest sites. Importantly, we see a discontinuity between c. 20,000 and c. 10,000 b.p. in the exchange networks which before and after this period transported ochres into the Cape Range sites.

These preliminary regional syntheses support the argument that the Pleistocene-Holocene transition in northwest Australia is a time of change - change in regional settlement patterns, in site functions and, possibly in exchange/redistribution systems. While the Pleistocene-Holocene distinction is geological in origin, this does not render it culture-less.

Smith & Sharp (1993:54) have argued that the Late Pleistocene record of northern Australia can be characterized as relatively stable; they draw attention to the `anomalous' sequences of northwest Australia. Rather than expecting these sequences to conform to a northern Australian `bloc', it would seem much more useful to examine why they have been interpreted, by different workers, as having contrasting patterns of occupation. The central thesis of this paper is that increased regional aridity during and after the Glacial Maximum, in conjunction with sea-level fluctuation, provided the major impetus for observed changes. This represents an unashamed `first-order' explanation (after Gould 1990).

Table 1 provides data for 197 radiocarbon determinations from northwest Australia; 21, or approximately 11%, are greater than 17,000 b.p. When the distribution of these dates is examined by millennia one unavoidable conclusion is reached.

The only absence of dates for a period more than a millennium over a 30,000-year period occurs between 13,000 and 17,000 b.p., where no dates are registered at all, the gap extending from 17,410 [+ or -] 330 to 12,100 [+ or -] 620 b.p.

This is clearly not a randomly located gap in the data. Have archaeologists, through flawed methodology and sample bias, managed to miss over 5000 years of the prehistory of northwest Australia? Is it simply chance that the complete absence of dates in this period coincides almost exactly with the timing of the height of the Last Glacial Maximum and the period immediately following (see also Hiscock 1988)? I think not.

Smith & Sharp argue (1993:55) that in northern Australia `the terminal Pleistocene-early Holocene was a time when the rate of sediment accumulation in rock-shelters was very much lower or when existing deposits were destroyed by erosion'. The proposed widespread erosion of rock-shelter deposits in northern Australia (Smith & Sharp 1993: figure 12), peaking between 9000 and 6000 b.p., occurs well after the hiatus in the northwest dates and does not explain why dates are consistently recorded in earlier millennia from 17,000 to 30,000 b.p. Natural erosional processes are not a satisfactory explanation for the 5000-year time-gap.

The substantially lower rate of both sediment and artefact accumulation during the Pleistocene-Holocene transition in northwest Australia is argued to reflect cultural responses to enhanced full glacial aridity (see demographic modelling in O'Connor et al. 1993). The high correlation between intensity of human occupation and sedimentation rate has been clearly demonstrated from a number of Australian rock-shelter contexts (Hughes & Lampert 1982). A decrease in the frequency and duration of visits to rock-shelter sites by the prehistoric inhabitants of northwest Australia, due to demographic shifts and increased residential mobility, would result in diminished deposits. In some cases these reduced deposits are likely to have been reworked by human and natural processes to the extent that minor depositional episodes are masked, particularly in sandy matrices.

The effects on site patterning of aridity and subsequent climatic amelioration during the Pleistocene-Holocene transition has recently been described for the tropics of Australia (cf. Morwood & Hobbs, this volume). Likewise, the present paper has argued that regions actually located within the expanded arid core of the Last Glacial Maximum, such as the Pilbara, would have been subject to significant resource re-structuring and stress. For northwest Australia, within the arid zone and containing one of the longest arid littoral zones on the globe, the Pleistocene-Holocene transition can reasonably be expected to have been a significant period of demographic and cultural transformation.

References

Baynes, A. & B. Jones, 1993. The mammals of Cape Range Peninsula, northwestern Australia, Records of the Western Australian Museum, Supplement 45:207-25. Birdsell, J.B. 1977. The recalibration of a paradigm for the first peopling of Greater Australia, in J. Allen, J. Golson & R. Jones (ed.), Sunda and Sahul: prehistoric studies in southeast Asia, Melanesia and Australia. 113-67. London: Academic Press. Bowdler, S. 1990. The Silver Dollar Site, Shark Bay: an interim report, Australian Aboriginal Studies 2:60-3. 1992. Homo sapiens in Southeast Asia and the Antipodes: archaeological versus biological interpretations, in T. Akazawa, K. Aoki & T. Kimura (ed.), The evolution and dispersal of modern humans in Asia: 559-89. Tokyo: Hokusen-sha. Bowler, J.M. & R.J. Wasson. 1984. Glacial age environments of inland Australia, in J.C. Vogel (ed.), Late Cainozoic palaeoclimates of the Southern Hemisphere: 183-208. Rotterdam: A.A. Balkema. Brown, S. 1987. Toward a prehistory of the Hamersley Plateau, northwest Australia. Canberra: Department of Prehistory, RSPaS, Australian National University. Occasional Papers in Prehistory 6. Dortch, C.E. 1977. Early and late stone industrial phases in Western Australia, in R.V.S. Wright (ed.), Stone tools as cultural markers: change, evolution and complexity: 104-32. Canberra: Australian Institute of Aboriginal Studies. Gosden, C. 1993. Understanding the settlement of Pacific islands in the Pleistocene, in Smith et al. (ed.): 131-6. Gould, R.A. 1977. Puntutjarpa Rockshelter and the Australian desert culture. New York (NY): American Museum of Natural History. Anthropological Papers 54. 1980. Living archaeology. Cambridge: Cambridge University Press. 1990. Recovering the past. Albuquerque (NM): University of New Mexico Press. Harrison, S.P. 1993. Late Quaternary lake level changes and climates of Australia, Quaternary Science Reviews 12:211-31. Hiscock. P. 1988. Prehistoric settlement patterns and artefact manufacture at Lawn Hill, Northwest Queensland. Unpublished Ph.D thesis, Department of Anthropology and Sociology, University of Queensland, Brisbane. In press. Technological responses to risk in Holocene Australia, Journal of World Prehistory. Hiscock, P. & P. Veth. 1991. Change in the Australian desert culture: a reanalysis of tulas from Puntutjarpa Rockshelter, World Archaeology 22(3):332-45. Hubbard, N.N. N.d. In search of regional palaeoclimates 1: Australia (18,000 BP). Unpublished manuscript. Hughes, P.J. 1992. Marandoo Salvage Archaeology Project: 2nd progress report. Unpublished manuscript. Hughes, P.J. & R.J. Lampert. 1982. Prehistoric population change in southern coastal New South Wales, in S. Bowdler (ed.), Coastal archaeology in eastern Australia: 16-28. Canberra: Department of Prehistory, RSPaS, Australian National University. Jennings, J.N. 1975. Desert dunes and estuarine fill in the Fitzroy Estuary, northwestern Australia, Catena 2:215-62. Jones, R. & J. Bowler. 1980. Struggle for the savanna: northern Australia in ecological and prehistoric perspective, in R. Jones (ed.), Northern Australia: options and implications: 3-31. Canberra: RSPaS, Australian National University. Kendrick, G. W. & K. Morse. 1982. An Aboriginal shell midden deposit from the Warroora coast, northwestern Australia, Australian Archaeology 14:6-12. Lau, J.E., D.P Commander & G. Jacobson. 1987. Hydrogeology of Australia, Bureau of Mineral Resources Bulletin 227:1-21. Lorblanchet, M. 1977. Summary report of fieldwork, Dampier, WA, Australian Institute of Aboriginal Studies Newsletter new series 7. Mabbutt, J.A. 1971. The Australian arid zone as a prehistoric environment, in D.J. Mulvaney & J. Golson (ed.), Aboriginal man and environment in Australia: 66-79. Canberra: Australian National University Press. Morse, K. 1993a. Who can see the sea?: Aboriginal occupation of the Cape Range Peninsula, Records of the Western Australian Museum, Supplement 45:227-42. 1993b. New radiocarbon dates from Northwest Cape, Western Australia: a preliminary report, in Smith et al. (ed.): 155-63. 1993c. Shell beads from Mandu Mandu Creek rock-shelter, Cape Range peninsula, Western Australia, dated before 30,000 b.p., Antiquity 67:877-84. 1994. West Side story: towards a prehistory of the Cape Range Peninsula. Unpublished Ph.D thesis, Centre for Archaeology, University of Western Australia, Perth. Murphy, A. 1994. Comments on Malia Rockshelter (Bee Gully Shelter 4b). Unpublished document. O'Connor, S. 1990. Thirty thousand years in the Kimberley: a prehistory of the islands of the Buccaneer Archipelago and adjacent mainland, West Kimberley, Western Australia. Unpublished Ph.D thesis. Department of Archaeology, University of Western Australia, Perth. 1992. The timing and nature of prehistoric island use in northern Australia, Archaeology, in Oceania 27(2): 49-60. In press. A 6700 BP date for island use in the west Kimberley, WA: new evidence from High Cliffy Island, Australian Archaeology. O'Connor, S. & P. Veth. 1993. Where the desert meets the sea: a preliminary report of the archaeology of the southern Kimberley coast, Australian Archaeology 37:25-34. O'Connor, S., P. Veth & N. Hubbard. 1993. Changing interpretations of postglacial human subsistence and demography in Sahul, in Smith et al. (ed.): 95-105. Pederson, B.J.T. 1983. A preliminary examination of the vegetation history of Dragon Tree Soak from 6380 BP to the present. Unpublished BA (Hons.) thesis, Department of Geography, University of Western Australia, Perth. Pulsford, T. 1994. Montebello repast: early Holocene environments and economy on the Montebello Islands, Pilbara Desert Coast, Western Australia. Unpublished BA (Hons.) thesis, Department of Anthropology and Archaeology, James Cook University of North Queensland, Townsville. Ross, A., T. Donnelly & R. Wasson. 1992. The peopling of the arid zone: human-environment interactions, in J. Dodson (ed.), The naive lands: prehistory and environmental change in Australia and the southwest Pacific: 76-114. Melbourne: Longman Cheshire. Smith, M.A. 1988. The pattern and timing of prehistoric settlement in central Australia. Unpublished Ph.D thesis, Department of Archaeology and Palaeo-anthropology, University of New England, Armidale. 1993. Biogeography, human ecology and prehistory in the sandridge deserts, Australian Archaeology 37:35-50. Smith, M.A. & N.D. Sharp. 1993. Pleistocene sites in Australia, New Guinea and island Melanesia: geographic and temporal structure of the archaeological record, in Smith et al. (ed.): 37-59. Smith, M.A., M. Spriggs & B. Fankhauser (ed.). 1993. Sahul in review: archaeology of Australia, New Guinea and island Melanesia. Canberra: Department of Prehistory, RSPaS, Australian National University. Occasional Papers in Prehistory 24. Smith, M.A., E. Williams & R.J. Wasson. 1991. The archaeology of the JSN site: some implications for the dynamics of human occupation in the Strezelecki desert during the late Pleistocene, Records of the South Australian Museum 25:175-92. Stafford Smith, D.M. & S.R. Morton. 1990. A framework for the ecology of arid Australia, Journal of Arid Environments 18:255-78. Trigger, B.G. 1989. A history of archaeological thought. Cambridge: Cambridge University Press. Van der Kaars, W.A. 1991. Palynology of eastern Indonesian marine piston-cores: a late Quaternary vegetational and climatic record for Australasia, Palaeogeography, Palaeoclimatology, Palaeoecology 85:239-302. Veitch, B. & L. Warren. 1992. A report of an archaeological recording and salvage programme at Tubridgi, near Onslow, Western Australia. Unpublished Report prepared for Doral Resources NL. Veth, P.M. 1987. Martujarra prehistory: variation in arid zone adaptations. Australian Archaeology 25:102-11. 1989. Islands in the interior: a model for the colonization of Australia's arid zone, Archaeology in Oceania 24(3):81-92. 1993. Islands in the interior: the dynamics of prehistoric adaptations within the arid zone of Australia. Ann Arbor (MI): University of Michigan. International Monographs in Prehistory, Archaeological Series 3. 1993. The Aboriginal occupation of the Montebello Islands, northwest Australia, Australian Aboriginal Studies 2:39-50. In press a. The occupation of arid coastlines during the terminal Pleistocene of Australia. Proceedings of the Australian Coastal Archaeology Symposium, Brisbane, 1994. In press b. Marginal returns and fringe benefits: characterizing the prehistory of the lowland deserts of Australia (a reply to Smith), Australian Archaeology. Veth, P., T. Pulsford & A. Chappel. In preparation. Before the blast: Aboriginal occupation of the Montebello Islands, northwest Australia. Veth, P.M. & B. O'Brien. 1986. Middens on the Abydos Plain, northwest Australia, Australian Archaeology 22:45-59. Vinnicombe, P. 1987. Dampier archaeological project: resource document, survey and salvage of Aboriginal sites, Burrup Peninsula, Western Australia. Perth: Department of Aboriginal Sites, Western Australian Museum. Wasson B. 1984. Late Quaternary palaeoenvironments in the desert dunefields of Australia, in J.C. Vogel (ed.), Late Cainozoic palaeoclimates of the Southern Hemisphere: 419-32. Rotterdam: A.A. Balkema. Woodroffe, C.D. 1990. The impact of sea-level rise on mangrove shorelines, Progress in Physical Geography 14:483-520. Wyrwoll, K.-H. 1979. Late Quaternary climates of western Australia: evidence and mechanisms, Journal of the Royal Society of Western Australia 62:129-42.