Coastal sites and severe weather in Cape Range peninsula, northwest Australia.
The occurrence of Tropical Cyclone Vance, which entered the Exmouth Gulf on March 22 1999, presented an opportunity to study in detail the effects of extreme storm conditions on coastal sites in northern Cape Range peninsula. A post-cyclone survey of all sites in the study area was undertaken in July 1999, four months after the cyclone struck and aimed to quantify and characterise the effects of a single extreme cyclonic event on coastal archaeological sites. This paper presents the results of this survey and implications for studying coastal archaeological sites in Australia.
All open sites in Australia are exposed to various weather conditions, the severity and duration of which is dependent on geographical location and season. Archaeological sites located in tropical and northern Australia are subject to seasonally severe weather in the form of tropical cyclones, which can be extremely powerful and destructive. They generally occur between November and March. While many archaeologists acknowledge the potential impact of atypical, unpredictable or stochastic climatic events on sites (eg Hughes and Sullivan 1974, Meehan 1982, Beaton 1985, Rowland 1989, Sullivan and O'Connor 1993, Fullagar et al. 1999, Clune in prep.), research investigating and assessing the effects of severe weather on archaeological sites in the Australasian region has been limited. A notable exception is Bird (1992) who described the destruction of a number of coastal shell middens in Queensland by two tropical cyclones in quick succession. Bird found that from an original sample size of 93 middens, more than 50% had been obliterated by the storms. Similarly, Spennemann has discussed the impact of cyclonic storms on archaeological sites from a number of Polynesian islands (1987, 1998). In an alternative approach Craib and Mangold (1999) argue that the impact of cyclonic storms on archaeological sites has been `overestimated', and damage by storm erosion is used as a convenient explanation by archaeologists to explain stratigraphic anomalies. They conclude that the general lack of empirical data and geoarchaeological evidence contributes to the apparently persistent assumption by archaeologists that cyclonic erosion is potentially damaging to sites (Craib and Mangold 1999: 305).
The study area
The study area for this research is a 15km section of arid Western Australian coastline, at the northernmost tip of Cape Range peninsula (Figure 1). Cape Range peninsula is a narrow protrusion of rugged limestone range projecting into the Indian Ocean at the western extent of the Australian mainland. It is flanked on both sides by coastal plains and sand dunes. The western coastal stretch to a distance of 100m to 300m from the shoreline (depending on the width of the dune system) of the study area was surveyed for archaeological sites in 1998. Over 15km of coastal dune was surveyed on foot, and 60 shell midden sites were located and recorded. All of the recorded shell middens in the Cape Range region are unstratified surface sites, ranging from dense surface accumulations of marine shell and artefactual stone to small sparse scatters of one or two species of shell (see also Morse 1996). To date, no shell middens located in Cape Range peninsula extend more than two or three centimetres below the surface, usually no more than the depth of a shell deposited edgeways in the sand.
FIGURE 1 OMITTED]
Northern Australia, constituting approximately one third of the Australian continent, is susceptible to tropical cyclones. Occasionally the effects of particularly powerful or prolonged cyclones can be felt considerably further south, reaching as far as Perth in the west and Sydney in the east. The average cyclone season lasts from November to May, and can bring to a region more than the average annual amount of rainfall in one storm. Cyclonic activity is related to the El Nino-Southern Oscillation, with fewer cyclones occurring during El Nino events (Nicholls et al. 1998). Between 1909 and 1975 454 cyclones were recorded in the Australian region, 253 of which made landfall (although this figure increases to 390 when multiple land crossings of individual cyclones are included) (Lourensz 1977:6). In Australia, cyclones are rated on a scale of severity from 1 to 5, the most powerful and destructive storms earning a category of 5.
Cyclones produce a number of potentially destructive forces, including sustained high wind speeds, violent gusts, scouring of the coast by waves, damage from wind borne debris, torrential rain and flooding. One of the most important, and damaging, cyclonic effects in coastal regions is wave action, particularly storm surge. Storm surge can be defined as the "super-elevation of the still-water surface that results from the transport and circulation of water induced by wind stresses and pressure gradients in an atmospheric storm" (Simpson and Riehl 1981: 234). The wave surge is caused by the cyclone moving from deeper water to shallow water, where the deep water wave action is maintained relatively unhindered (except for friction from the sea floor) thus creating unusually extreme wave patterns. This can be particularly accelerated in shallow coastal waters located close to deep oceanic waters, as is the case in Cape Range Peninsula, where the shallow coastal and reef system is located only 10km away from the deep waters of the continental shelf. In coastal areas located further from deep oceanic waters, the cyclonic storm surge can be reduced in its overall effect, as the further the cyclonic wave and storm tide has to travel, the more effect friction from the sea floor and increasing barometric pressure will have on the swell. Due to its high energy, storm surge can erode beaches and coastal land formations such as dune systems very quickly, drawing loose material away and out to sea by strong currents (Simpson and Riehl 1981: 245-246).
On March 22, 1999, tropical Cyclone Vance entered the Exmouth Gulf and the small town of Exmouth was devastated by the ensuing storm. This was a Category 5 cyclone that was at the time the most powerful storm ever to be recorded in Australia. The cyclone had originated further north, offshore in the Kimberley region, and in two days swept 800km down the coast. Wind speeds reached up to 267km/hour, and severe flooding over the entire Cape Range peninsula was caused by torrential rain and storm surge (the cyclone coincided with a king tide, producing a surge of up to 8m with a 12m swell). In comparison, Cyclone Tracy, which devastated Darwin in 1974, was a Category 3 cyclone where wind speeds reached 217 km p/h. Queensland cyclones `Aivu' in 1989, and `Althea' in 1971, whose effects on coastal archaeological sites were documented by Bird (1992), were also Category 3 cyclones, with wind speeds in both cases of less than 200km p/h (Bird 1992).
Predicted cyclonic outcomes on archaeological sites
The kinds of damage associated with cyclones on archaeological sites can be placed into four categories: archaeological material is removed, material is exposed, material is buried and material is redistributed within the sites. The aim of the post-cyclone survey was to assess and quantify the effects of the cyclone on the midden sites with these four hypothetical processes in mind, and to look at the implications of these processes for understanding the complex archaeological site formation process of coastal sites in Australia.
Archaeological material is removed
This effect ranges from the removal of midden material from the site to the complete destruction of sites.
Archaeological material is exposed
The effect of this process is that `new' sites appear where there were none previously visible. This is caused by the removal of surface sediment exposing archaeological material that has been buried by previous erosional processes. Bird (1995) reported that several sites in the Upstart Bay region of the northeast Queensland coast had been exposed by continuing dune erosion caused by cyclonic events a few years previously. Her findings highlight the potential effect on the archaeological record by long-term geomorphological processes resulting from cyclonic erosion.
Archaeological material is buried
This effect is the only process that would not necessarily be visible on the surface of the site. The burying of midden material by redeposited sediment would perhaps be most obvious and quantifiable by excavation. However this is not always possible or desirable. Indications of newly buried midden material can be signified by visible clues on the surface. Absence of material together with other features such as new mounding of sediment in the vicinity of the original site or decreased midden margins (where material has been buried with sediment at the edges of the existing site) would also suggest burial of material.
Archaeological material is redistributed
This process effects both the archaeological material and the site matrix, and causes mixing of sediment and artefacts. This effect can also be observed in stable, mounded middens where stratigraphic inversions or confused date sequences are obtained.
The study area was first surveyed for archaeological sites as part of my doctoral research during September to November in 1998. Cyclone Vance struck the coast four months later in March 1999. In July 1999, four months after the cyclone struck, I was able to return to Cape Range peninsula to assess the impact of the cyclone on the sites in the study area. As with the original survey, the post-cyclone coastline of the research area was re-surveyed on foot. Particular attention was paid to relocating the eight middens that had been subject to surface collection during the initial survey, and identifying any new sites that may have appeared.
All eight middens that had been sampled were relocated, and seven of the eight were re-sampled using the same procedure as before with the aim of providing quantitative information about the extent of the storm damage to the sites. The eighth midden, Vlaming Head Midden 1, was not included in the post-cyclone analysis. The site is considerably smaller that the others, and only 1[m.sup.2] had been collected from the site before the cyclone struck. It was felt that a sample of this small size would not provide meaningful data about cyclonic disturbance. Each midden was first photographed from the same position and angle as it had been in 1998. Then, the midden was re-mapped, with any alterations in its form, vegetation cover or density of remains noted. The approximate locations of the 1998 collection squares were re-located using maps and photographs. In the original survey the midden material was collected and removed from the site for later analysis in the lab, leaving an area of clean sand behind.
In the post-cyclone survey several 1x1m area counts of archaeological material were made in each site from a number of different locations within the midden. Counts were made inside the area where the original collection square had been in order to provide a measure of the redistribution of midden material within each site (ie midden material that had been transported into the area that had previously been left devoid of material). Counts immediately adjacent to the original collection squares were made to see how much midden material had been lost, or gained, from the effects of the cyclone. In the post-cyclone survey all midden material was left in situ, and so a great deal of care was taken not to overlook artefactual material that was small in size and fragmentary.
Of the four generalised cyclone induced site alterations outlined above, coastal middens in Cape Range Peninsula were affected by two: archaeological material was removed from sites, and material was redistributed within and around sites. There were no middens that had disappeared completely; all of the re-surveyed sites retained at least their basic shape, if somewhat broader or deeper than before. No `new' archaeological sites that had been exposed by the storm were encountered. Nor was there any reason to believe that archaeological material had been buried by sediment. I did observe that some rocky shores had been covered with sand and some sandy beaches were somewhat elevated as a result of wave action and floodwater. Redeposition of sediment in this context however was limited to the immediate shoreline and did not appear to affect the middens, all of which were located behind the high tide line.
The first and most apparent effect of the cyclone was that middens close to the sea were by far the most disturbed. Storm surge and wave action appeared to have been the most damaging cyclonic effect, causing material to have been actually removed from sites and either swept out to sea or redeposited further down the beach at the high tide zone. Sites located within the reach of storm surge tended to be deepened in the centre and enlarged at the peripheries, where the combined water and wind action had caused dunes to collapse and loose surface sediment to be lost. Middens located further from the beach and higher in the dunes were more affected by strong winds and rain, and so preservation of midden material and site structure tended to be slightly, and in some cases considerably, greater further from the tidal zone. Distinctive patterns in the surface sand, such as channels and ripples, were left behind where water and wind had damaged the surface of sites.
Shell midden case studies
Quantitative data documenting the effects of the cyclone on seven of the sampled shell middens are presented here. For each site, the average amount of archaeological material per square metre both pre- and post-cyclone was calculated and compared. Post-cyclone counts made from within the location of the original collection squares and the results averaged. This provides an indication of the degree to which midden material was redistributed within the site, as the collection squares were completely cleared of archaeological material prior to the cyclone.
Babjarrimannos Midden 1
This large midden was located in a deep blowout behind the first few dunes, approximately 25m from the shore. The surrounding dunes were lightly vegetated, and the surface of the blowout itself was scattered with occasional herbaceous plants. The midden contained large quantities of turtle (Chelonia mydas) and dugong (Dugong dugon) bone fragments, a feature no other shell midden in the study area shared. The midden also contained a diverse range of marine mollusc species from both rocky platform and sandy bottom environments, in addition to the ubiquitous baler (Melo amphora), chiton (Acanthopleura gemmata) and giant clam (Tridacna maxima). Smaller artefacts such as fish otoliths and teeth also occurred on the midden. A large range of artefactual stone was found on the site, most significantly considerable quantities of artefacts with secondary working, formal tool types such as tula and burren adzes and backed pieces, and a high proportion of fine grained raw materials.
The area where the collection was made the previous year, at the base and lower incline of a sand dune in the northern outer third of the midden, was completely removed by the cyclone. It had been located on the slope of a lightly vegetated dune, but this had collapsed, and the place where the collection had been made was now level. Behind the now level former dune was a precarious 2m high erosional scarp, held together by the remaining vegetation. The midden surface was rippled with erosion channels apparently caused by water and wind action. Where there had been soft loose surface sand and midden material was now a hard consolidated sand surface. Initial inspection of the site suggested that only the larger artefacts remained, such as chunks of giant clam and baler shell, and larger pieces of bone and artefactual stone.
Most of the archaeological materials that were removed from the site were the smaller, more fragmented pieces such as Turban (Turbo sp.) shell and opercula, fragments of crustacean and small bivalves (Table 1). Larger, heavier objects such as baler shell, giant clam, cowries and chiton were less affected by the storm, and tended to remain within the site. Several pieces of worked marine shell were found on this site during the initial survey. No worked shell was recorded at the site during the post-cyclone survey.
The most striking change is the dramatic reduction in the amount of bone fragments left on the site. Prior to the cyclone the average number of bone fragments per [m.sup.2] on this midden was 638. More than 98% of the bone was removed from the site, leaving a post-cyclone average number of 10 per [m.sub.2] Movement of archaeological material within the site, as indicated by the mean number of fragments redistributed per [m.sub.2] (Table 1), was mainly limited to the larger and/or heavier objects such as baler shell and cowries, or to more abundant remains such as bone fragments.
Babjarrimannos Midden 2
This midden was located well back from the current beach line, approximately 175m east of the high tide zone. The midden scatter was concentrated in a bowl shaped dune blowout, surrounded by densely vegetated consolidated sand dunes. The midden material was evenly scattered over the site with a concentration of shell (mainly chiton) at the uppermost north-eastern edge. The archaeological assemblage of Babjarrimannos Midden 2 was dominated by chiton shell and turban opercula. Small quantities of other marine molluscs occurred in the site, including baler, bivalves and Terebralia palustra. Also found in relatively small quantities were artefactual stone and worked marine shell.
The midden retained its same basic shape following the cyclone, but had been swept clear of loose sand and midden material by strong winds and rain. The centre of the site was the most affected, and had been deepened by approximately 25cm due to the removal of surface sand. Spinifex clumps to the north of the midden were packed full of sand, where the exposed plant roots and spines had acted as sediment traps (Figure 2). The packing of sand into spinifex bushes continued further north for at least 100m. Very small fragments of former midden material, such as turban opercula and small chiton fragments were thinly scattered over the surface of the dunes surrounding the blowout where there had been none the previous year. The site was located too far from the beach to be affected by storm surge. The collection of sediment in surrounding vegetation suggests that the major cyclonic action that affected the site was strong winds.
[FIGURE 2 OMITTED]
Counts were made from several locations in the site. The very centre of the midden was completely devoid of artefactual material, and only sand remained. The outer two thirds of the site had fared a little better, particularly in the southern half of the site. A high proportion of the smaller, more fragmentary archaeological materials, such as small bivalves and bone fragments, were removed from the site by the cyclone (Table 2). The larger and heavier objects, such as artefactual stone, baler, giant clam and chiton fared considerably better. Only 40% of the giant clam, 60% of baler, 67% of chiton and 70% of Terebralia palustra was removed from the site by the cyclone, compared with 100% of many other shell species. Artefactual stone was reduced by 80%, and no fragments of worked giant clam were found at all. The species that were redistributed over the site were the most commonly occurring and the smallest in size, particularly chiton and turban opercula.
Vlaming Head Midden 2
Vlaming Head Midden 2 was the largest shell midden located in the study area, with a midden scatter measuring 125m by 40m. The northernmost edge of the midden scatter was located 30m south of the shore, which was predominantly rocky with large boulders shelving steeply to the reef. Baler, chiton and turban opercula were the most commonly occurring marine molluscs in the midden deposit. Several other species, including large and small bivalves, limpets and oysters were found in smaller quantities. Bone and emu eggshell fragments were also found in the site. Artefactual stone at Vlaming Head Midden 2 included retouched pieces, formal tools (such as tulas, burrens, backed blades and scrapers), debitage and grinding stones. Many of the stone artefacts were made using a distinctive silcrete quarried from an outcrop located immediately south of the site.
Following the cyclone, the entire surface of the site was stripped of surface sand and archaeological material, which had the effect of deepening and broadening the blowout in which the midden was located. Along the northern margin of the site the beach had been considerably elevated by deposited sediment, and the section of coast immediately adjacent to the site was completely covered with sand, obscuring what had previously been a rocky shore strewn with large boulders (Figure 3). It is estimated that the beach was elevated by up to 1.6m by redeposited sediment. Sediment had also been redeposited creating an underwater sandbar that stretched from west of Vlaming Head eastward to Surfer's Beach, some 1500m. The sandbar, which was clearly visible from the shore, was a direct result of the cyclone and was not present during the 1998 survey.
FIGURE 3 OMITTED]
Storm surge, which is estimated to have been up to 8m with a 12m swell, was the major process contributing to the damage of this site. The entire 125m northern margin of the site ran parallel to the beach, and was afforded little protection from wave action by low foredunes. No sign of the foredunes was visible in the post-cyclone survey. The surface of the midden showed very little patterning (such as rippling), and was almost entirely smooth, sloping gently down to the newly elevated beach.
Most of the material was completely removed from the site by the cyclone (Table 3). The only archaeological remains that survived on the midden were the larger, heavier and most numerous objects including baler, chiton, giant clam and smaller, more fragmentary but profuse bone and turban opercula. Between 4% and 11% of the original amount of baler, chiton and artefactual stone per [m.sup.2] was counted in Vlaming Head Midden 2 during the post-cyclone survey. Prior to the storm numerous fragments of bone, turban opercula and chiton were contained in the site, of which between 85% and 95% was removed.
Surfer's Beach Midden 1
This shell midden was located in stable dunes some 60m from the high tide zone in a dune escarpment high above the rocky shore. The swale in which the midden was located was heavily vegetated, making surface visibility low in some sections of the site.
The midden did not appear to have been substantially altered in shape and dimension by the cyclone, and despite the vegetated nature of the site was easily relocated. As with most other middens, the surface of the site had been cleared of loose surface sand resulting in the removal of some artefactual material. Sediment had collected in thick mounds at the base of shrubs, and was trapped in spinifex hummocks throughout the dunes in the surrounding area. This suggests that the dominant action in this area was strong winds, particularly as the site was located well above and behind the potential storm surge zone.
The contents of this site fared considerably better than most other middens in the study area, although several marine shell species, a majority of which were small in size or fragmentary, were not found on the site in the post-cyclone survey (Table 4). Of the twenty species and artefact types identified on the midden before the cyclone, nine were still present in the second survey. The material that survived the storm was either large and heavy, such as baler shell, giant clam and artefactual stone, or numerous within the site, such as bone fragments and turban opercula.
Surfer's Beach Midden 2
This midden was located in a deep circular blowout in moderately vegetated dunes. The surface of the midden was composed of loose sandy sediment and was sparsely vegetated, while the surrounding dunes were more consolidated and stable. The midden was some 45m from the high tide zone, and was situated in the elevated dunes of a high dune scarp, rising steeply from the beach below.
Surfer's Beach Midden 2 was largely unaltered in shape and dimension by the cyclone. The surface of the midden within the blowout was however, like all the other sites in the survey, stripped of loose surface sediment and vegetation. The removal of surface sediment however did not appear to have been as thorough as with most other sites, and some loose sand was still present on the surface. The surface sand of the blowout was rippled and channelled, indicating that the site had not escaped water and wind action (Figure 4). The flanking dunes retained their structure and shape, and did not appear to have been adversely affected.
[FIGURE 4 OMITTED]
Although the structure of the site appeared to be intact, the post-cyclone counts indicated that a considerable amount of artefactual material was removed from the site. Baler, giant clam and turban were the only species of marine shell that were preserved in the site (Table 5). No bone fragments were counted in any of the post-cyclone squares. Artefactual stone was reduced by 75%, from an average of >40 artefacts per [m.sup.2] to less than 10. Surfer's Beach Midden 2 was the only site in the study area that was observed to contain worked shell in the post-cyclone survey. The amount of worked shell was reduced in this site from 3 artefacts per [m.sub.2] to 1, a reduction of 67%.
North West Cape Midden I
North West Cape Midden 1 was a large midden nestled in a blowout between high, lightly vegetated dunes. The midden scatter extended toward the high-tide line, the site terminating within 20m of the water. Sand dunes south of the site were more densely vegetated than those to the north and east, but lower in height and not as steeply inclined. The midden was large, covering a rectangular area of approximately 45m by 30m, the north western end opening straight to the beach with no foredunes in between.
This midden was dense in parts, and several areas of discrete artefact scatters occurred in sections with more consolidated sands. Although dense, the midden scatter did not extend more than 2cm below the surface. A wide range of faunal and artefactual material occurred in the site, including marine shell, bone fragments, artefactual stone, worked clam and small fragments of emu eggshell. Like Babjarrimannos Midden 1 and Vlaming Head Midden 2, this site contained a large number of retouched stone tools and formal tools, such as tula and burren adzes and scrapers, and a high proportion of fine grained lithic raw material.
The most striking change in this site was that the surface sand had been swept away, leaving behind a more consolidated sandy surface, and deep channels had been created by the action of wind and water. The removal of surface sand and midden material meant that the overall surface area of the blowout was enlarged and deepened. The high walls of the surrounding dunes had collapsed, and the collapsed sand had been removed with the rest of the surface sand. The collapsed dunes ringing the site were now precarious scarps of sand more than 3m high. In addition to being marked by channels, the blowout surface was scattered with `pedestals' of sand supporting artefactual and non-artefactual objects where the loose sand had been removed (Figure 5). The objects perched on pedestals tended to be larger in size and/or relatively heavy, such as giant clam and fist sized pebbles. The pedestals were generally between 20cm and 25cm high, although some were up to 50cm. This indicates that at least 20cm of surface sand and midden material has been removed from the site. I also observed that a sandbar had formed approximately 40m off shore, which was visible from the crest of the dunes. The sandbar had not been there the previous year.
[FIGURE 5 OMITTED]
North West Cape Midden 1 lost a considerable amount of midden material (Table 6). The only material found on the midden after the cyclone was baler shell, giant clam, chiton and artefactual stone. Chiton and artefactual stone were the least affected by the storm, where only 77% and 80% respectively was removed from the site. All the small size (or fragmentary, such as pearl oyster) and light objects were entirely removed from the midden scatter, and redeposited with dune sediments out to sea creating the new sandbar. Movement of material within the site once again favoured larger and/or heavy faunal remains, such as chiton, baler and stone.
North West Cape Midden 2
This was a small midden located behind moderately vegetated low coastal foredunes, which effectively separated the swale in which the midden scatter is situated from the beach and water. This midden consisted almost entirely of Terebralia paulstra, although small quantities of other marine species were present. The midden also contained a small number of stone tools (primarily debitage). The midden surface was littered with numerous hand sized upper grindstones and larger pecked limestone slabs. Some of the slabs were encrusted with chalky residues and shell grit. The shoreline here was rocky, with a reef extending approximately 50m out to sea, which was exposed during low tides.
The low dunes between the midden and the high tide line were breached by storm surge, and cultural material bearing sand was both swept out to sea and redeposited on the beach, covering the rocky shoreline and reef for at least 15 metres out to sea. A large drainage channel carved into the sand extended from the site to the beach though the dunes via the breach, which was approximately 60cm deep at the deepest part. The midden and adjacent areas were riddled with water erosion channels leading generally toward the breach between the midden scatter and the beach. Pedestals of sand also formed at North West Cape Midden 2, as described for North West Cape Midden 1. The pedestals of sand (almost all with a medium to large size object perched on top) were generally between 10cm and 15cm in height. This area was also covered in storm litter; parts of buildings and caravans, road signs, polystyrene, timber and lumps of bitumen.
All midden material except Terebralia palustra, turban opercula and artefactual stone had been removed from the site (Table 7). T. palustra was the most abundant marine mollusc in the site, and was reduced by more than 90%, from an average of 637 per [m.sub.2] to 61 per [m.sub.2] All artefactual stone observed on the site in the post-cyclone survey was large grinding and processing limestone slabs. No stone tools other than the slabs were recorded either in the counts or in general survey of the site. Overall, the average number of stone artefacts (including grinding and processing slabs) was reduced from 6 per [m.sub.2] to 0.5 per [m.sub.2], a reduction of 92%. The only archaeological remains found within the former collection square area was T. palustra and turban opercula, which suggests that the only movement of material within the site was of the most abundant and smallest in size.
Two of the four predicted outcomes of the cyclone on seven archaeological sites in the study area have been identified. These are removal and redistribution of archaeological material. Archaeological material was also redistributed on the surface of formerly archaeologically sterile dune crests above some sites, particularly Babjarrimannos Midden 2 where midden sediment and artefactual contents was scattered thinly over a wide area immediately north of the site. No sites were completely destroyed by the cyclone; although many sites lost more than 90% of their contents, the sites were still visible and were easily re-located. There was no indication that archaeological sites or material was buried in situ in the study area. Similarly, no archaeological sites were located in the post-cyclone survey that were not recorded in the pre-cyclone survey.
Cyclone Vance had a devastating effect on the coastal middens in Cape Range peninsula. More than 70% of the total amount of archaeological material was removed (Table 8) from all shell middens analysed in this study, with a mean loss of 86%. Three of the seven middens surveyed had less than 10% of their archaeological material remaining. Sediment and archaeological material that was removed from the middens ended up in one of three locations, depending on the proximity of the midden to the beach. Sites within reach of storm surge, particularly those afforded only minimal protection by low foredunes, tended to have loose surface sediment, artefacts and faunal material either removed out to sea or redeposited lower down on the beach. Along the northern stretch of coast in the study area offshore sandbars were created by the removal and redeposition of dune sediment. Where sediment had been redeposited lower on the beach, it was clear that beaches and rocky shores were elevated, in some cases by up to 1.5m.
In all middens the material that tended to survive the cyclone was typically the largest and heaviest objects, or the most numerous. Baler shell, giant clam, artefactual stone and chiton were the most prevalent archaeological materials contained in the post-cyclone middens. In some cases, small and fragmentary faunal remains such as turban opercula and bone fragments were still present, when most other small and fragmentary artefacts had been removed entirely. This pattern was most pronounced in sites where these artefacts occurred in significant quantities before the cyclone.
Artefact counts made during the post cyclone survey within the original 1998 collection squares showed that although these areas had been cleared of archaeological material before the cyclone, material had been redeposited into all of these squares. This suggests that sediments and material were lost from the sites gradually throughout the duration of the cyclone, rather than by a single catastrophic event during the storm, such as a particularly powerful wave or wind gust. This also suggests that although the loss of midden material is perhaps the most immediately obvious effect of extreme coastal weather, the movement of midden material within an archaeological site also often occurs. This has important implications for understanding intra and inter-site spatial patterning of archaeological materials at coastal archaeological sites in Australia.
The middens in the study area generally occurred in three geographical locations. Type 1 middens were found in sparsely vegetated dune blowouts extending from behind a usually low foredune to close to the high tide zone. Type 2 middens were located in swales or blowouts behind lightly to moderately vegetated, more substantial foredunes with the extent of the midden deposit located further from the high tide zone. Type 3 middens were generally furthest from the shore, situated in swales or blowouts well behind the current foredunes in more consolidated sands and surrounded by moderately to densely vegetated, stable dunes. In general, middens located closer to the current shore are mostly younger, dating to the late Holocene, while middens situated further inland, Types 2 and 3 middens, tend to be older, dating to the mid Holocene (for further discussion of the dating of the midden sites see Przywolnik 2002).
These results suggest that the extent to which sites were altered by the cyclone is associated with the distance of the midden from the beach and the dune system. Those middens located close to the high tide zone, Type 1 middens, were the most prone to inundation and erosion by storm surge, and as a result suffered the most in terms of loss of surface archaeological material and damage to the surrounding dune structure. The further sites were located from the beach and away from the storm surge zone, the more surface archaeological material was retained, with Type 3 middens retaining the most (Table 8).
The stability and size of foredunes were also influential in protecting some sites from cyclonic erosion. Both the Type 1 middens were located behind low and sparsely vegetated foredunes that afforded no protection from storm surge. Similarly, Vlaming Head Midden 2, a Type 2 midden, was located behind low foredunes that were no effective barrier to waves. As a result more than 90% (97% for North West Cape 1) of the contents of each of the three middens was removed. The structure of the surrounding dunes in these three sites was also adversely affected by the cyclone, where secondary dunes were truncated by wave action, forming a steep and highly unstable scarp of sand.
Type 2 middens with higher and more stable, vegetated foredunes were afforded slightly more protection from wave action. Located behind low but moderately vegetated foredunes, North West Cape Midden 2 was damaged slightly less, 88% of the archaeological material being removed. Another Type 2 midden, Surfer's Beach 2, was located in very high, vegetated dunes above the beach and lost only 83% of its artefactual contents. Both Type 3 middens in this analysis were located more than 50m from the high tide zone in well vegetated, stable secondary dunes. These two sites were the most well preserved in the survey, losing between 73% and 76% of the midden contents, a difference of 20% in comparison with the Type 1 middens.
All sites, including Type 3 middens, showed evidence of wind and water erosion in the form of surface patterning (ripples, channels, pedestals of sediment) and mounding of trapped sediment around vegetation. The most damage to the middens, in terms of artefact removal and damage to site structure, occurred where sites were closest to the high tide zone. This suggests that storm surge was the most destructive cyclonic process. Bird (1992) and Spennemann (1998) also conclude that storm surge caused by cyclonic conditions was responsible for more damage to coastal sites than any other erosional process in their respective study areas.
This analysis has shown that severe storms have the potential to be extremely destructive of archaeological sites and associated coastal landforms. It must be noted however that `Vance' was an abnormally powerful cyclone, the worst recorded storm to have struck the Australian mainland. The degree of devastation amongst the middens, some retaining less than 10% of their original contents prior to the cyclone, suggests that storms as violent as `Vance' did not occur with any frequency in the study area, otherwise midden sites would not have survived at all. Considerably less powerful cyclones do however cross the coast in the area several times in every decade, suggesting that similar, if smaller scale, processes would occur consistently (seasonally) over time.
The annual cycle of cyclonic storms has had a strong impact on coastal sites in Cape Range peninsula, and until the increase in human traffic along the coastal strip in the 1960s degraded and eroded the northern dune system, cyclones were the most important factor influencing the way in which the midden sites were preserved. The only middens that occur in Cape Range peninsula are surface scatters found in dune blowouts and swales, the contents of which do not extend any further below the surface than a few centimetres. Seasonal cyclonic activity has meant that sediment and artefacts are given little opportunity to accumulate in sites before the next season of storms. Those sites protected by stable, vegetated flanking dunes and distance from storm surge have increased potential for greater preservation of artefactual material, although the pattern of human activity represented at the midden sites suggests that sites closest to the contemporary shoreline (i.e. those that are most vulnerable to cyclonic weather) were those most often used.
In other parts of northern Australia coastal open sites located in sheltered areas, such as in estuaries, behind mangroves, on sheltered headlands or behind well established dunes, have a better chance of survival than those located on exposed featureless beaches (Bailey 1977, Woodroffe et al. 1988, O'Connor 1989, Bird 1992, O'Connor and Sullivan 1994, Veitch 1996, Ulm and Lilley 1999). Similarly, mounded middens occur only in parts of northern Australia protected from cyclonic processes, in estuaries, behind low-energy coasts (such as mangrove mudflats) and on the leeward side of offshore islands (O'Connor 1999, Veitch 1999, Clune in prep). Open sites located on high-energy, exposed coast lines in northern Australia tend to be surface scatters only (Veth and O'Brien 1986, Morse 1996, Bowdler 1999), a result of cyclonic erosional processes. This study has been able to quantify the effects of a single extreme weather event on archaeological sites located on a high-energy exposed coastline. The results of the study confirm and shed some light on the details of why mounded middens rarely occur in such environments. Further it has been able to quantify the less visible effects of extreme weather on coastal archaeological sites in the movement of midden material within archaeological sites during cyclonic activity.
There is ample evidence that the recent trend towards global warming will have repercussions for coastal environments. It is predicted that the cumulative effect of global warming will increase the frequency and intensity of cyclones and floods, and change sea level (Wendland 1977, Evans 1991, Rowland 1992). While it is impossible to ensure the long term protection of all sites along the cyclone coastline in northern Australia, this research shows that the stability and integrity of coastal dune systems has a moderating effect on cyclonic erosional processes. The implication for cultural heritage management is that dune rehabilitation and stabilisation is vital for the preservation of vulnerable midden sites such as those discussed in this analysis. This would be best achieved by limiting human (and livestock in some regions) access to degraded dunes to prevent further deterioration, and re-vegetating dunes to assist with stability. This in turn would offer some buffer for exposed sites and reduce the impact of wind and water erosion during extreme weather occurrences. Alternatively, but perhaps less desirably, structures along the coast designed to protect sites from storm erosion could be built, such as baulks and sea walls (see for example Garrett 1983).
Tropical cyclones have the potential to be extremely destructive forces, affecting more than one third of the Australian continent. This analysis demonstrates the capability of cyclonic erosional processes to effect changes in open archaeological sites. In this study all the midden sites discussed were reduced of their artefactual content by between 73% and 97%, with a mean of 86%. The unusually powerful movement of water and wind caused by Cyclone Vance resulted in two of the four predicted outcomes altering the coastal middens in northern Cape Range peninsula: archaeological material was removed from the sites, and was redistributed in and around the sites.
This is the first time the effects of a severe cyclone on coastal shell middens has been quantified in this manner. The effect of cyclonic storm surge on coastal midden sites was found to be the single most important factor. Sites furthest from the modern shoreline behind the modem foredune system were the least affected.
Cyclones are arguably the most influential Holocene post-depositional process for coastal sites in northern Australia, and have been instrumental in shaping how the shell middens appear today. The impact of extreme seasonal weather on archaeological sites located on high-energy, exposed coastlines in northern Australia is the most likely explanation for the lack of mounded midden sites in such areas. The implication of the study for cultural heritage management is that dune rehabilitation and stabilisation is vital for the preservation of vulnerable midden sites such as those discussed in this analysis.
Table 1. Babjarrimannos Midden 1: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 72 13 81 7 stone Flaked 5 0 100 0 giant clam Baler 81 28 66 26 Giant 5 0.2 96 0 clam Chiton 13 4 70 2 Terebralia 5 0 100 0 Mussel 29 6 79 0.5 Limpet 5 1 80 0 Pearlshell 5 0 100 0 Periwinkle 10 5 55 0.2 Cowrie 5 0 100 1 Surf clam 5 0 100 0.2 Cockles 5 0 100 0 Turban 40 0 100 0 opercula Turban 5 0 100 0 shell Cone 5 0 100 0 Crustacean 14 0 100 0 Barnacle 7 1 86 1 Fish 5 0 100 0 otoliths Bone 638 10 98 5 Fish teeth 9 0 100 0 Table 2. Babjarrimannos Midden 2: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 10 2 80 1 stone Flaked 5 0 100 0 giant clam Baler 5 2 60 0.5 Giant 5 3 40 0.3 clam Chiton 169 56 67 28 Terebralia 5 1 70 0 Mussel 5 0 100 0 Limpet 5 0 100 0 Rock oyster 5 0 100 0 Periwinkle 5 0 100 0 Cowrie 5 0 100 0 Surf clam 5 0 100 0 Cockles 5 0 100 0 Turban opercula 800 144 82 7 Turban shell thais 76 8 89 3 Crustacean 5 0 100 0 Bone 5 0 100 0 Table 3. Vlaming Head Midden 2: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 23 1 96 2 stone Flaked 2 0 100 0 giant clam Baler 9 1 89 3 Giant 2 0 100 0 clam Chiton 195 13 93 23 Terebralia 2 0 100 0.3 Mussel 2 0 100 0 Limpet 2 0 100 0 Pearlshell 2 0 100 0 Rock oyster 2 0 100 0 Cowrie 2 0 100 0 Turban 192 22 88 5 opercula Turban 3 0 100 0 shell Thais 2 0 100 0 Cone 2 0 100 0 Crustacean 2 0 100 0 Barnacle 2 0 100 0 Fish 2 0 100 0 otoliths Bone 64 9 86 3 Fish teeth 2 0 100 0 Emu egg 11 0 100 0 shell Table 4. Surfer's Beach Midden 1: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 19 3 82 3 stone Flaked 2 0.2 87 0 giant clam Baler 6 2 65 1 Giant 2 0.5 69 0 clam Chiton 1 0 100 0 Terebralia 5 2 62 1 Mussel 12 0 100 2 Limpet 2 0 100 0 Pearlshell 2 0 100 0 Rock oyster 3 0 100 0.2 Cowrie 2 0.5 71 0.2 Surf clam 2 0 100 0 Turban 57 5 91 3 opercula Turban 1 0 100 0 shell Thais 1 0 100 0 Cone 1 0 100 0 Crustacean 1 0 100 0 Barnacle 1 0.2 80 0.2 Fish 1 0 100 0 otoliths Bone 33 6 82 0 Table 5. Surfer's Beach Midden 2: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 40 10 75 4 stone Flaked 3 1 67 1 giant clam Baler 17 3 82 2 Giant 4 2 50 2 clam Chiton 2 0 100 0 Terebralia 5 0 100 0 Mussel 3 0 100 0 Limpet 2 0 100 0 Pearlshell 2 0 100 0 Rock oyster 2 0 100 0 Cowrie 2 0 100 0 Cockles 2 0 100 0 Turban 205 37 82 9 opercula Turban 2 0 100 0 shell Thais 2 0 100 0 Cone 2 0 100 0 Crustacean 2 0 100 0 Barnacle 2 0 100 0 Bone 20 0 100 0 Fish teeth 2 0 100 0 Table 6. North West Cape Midden 1: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 43 8 80 1 stone Flaked 1 0 100 0.5 giant clam Baler 78 5 94 2 Giant 9 1 92 0 clam Chiton 136 31 77 19 Terebralia 1 0 100 0 Mussel 4 0 100 0 Limpet 1 0 100 0 Pearlshell 3 0 100 0 Rock oyster 1 0 100 0 Periwinkle 1 0 100 0 Cowrie 2 0 100 0 Surf clam 1 0 100 0 Cockles 1 0 100 0 Turban 805 0 100 0 opercula Turban 1 0 100 0 shell Thais 1 0 100 0 Cone 1 0 100 0 Crustacean 4 0 100 0 Barnacle 9 0 100 0 Fish 5 0 100 0 oltoliths Bone 130 0 100 0 Fish teeth 10 0 100 0 Emu egg shell 1 0 100 0 Table 7. North West Cape Midden 2: number of fragments per [m.sup.2] pre- and post-cyclone Archaeological Pre-cyclone: Post-cyclone: % lost Mean number material mean number mean number of fragments of fragments of fragments redistributed per [m.sup.2] per [m.sup.2] per [m.sup.2] Artefactual 6 0.5 92 0 stone Giant 5 0 100 0 clam Chiton 5 0 100 0 Terebralia 637 61 90 7 Mussel 6 0 100 0 Limpet 5 0 100 0 Pearlshell 5 0 100 0 Rock oyster 5 0 100 0 Cowrie 5 0 100 0 Surf clam 5 0 100 0 Cockles 5 0 100 0 Turban 197 41 79 6 opercula Turban 5 0 100 0 shell Thais 5 0 100 0 Crustacean 5 0 100 0 Bone 5 0 100 0 Table 8. Pre- and post-cyclone mean total artefact number per [m.sup.2] and percentages of archaeological material remaining each site Midden Site pre-cyclone post-cyclone % % type remaining lost 1 Babjarrimannos 46 3 7 92 Midden 1 1 North West 52 2 3 97 Cape Midden 1 2 Vlaming Head 25 2 8 92 Midden 2 2 Surfer's Beach 15 2 17 83 Midden 2 2 North West 54 6 12 88 Cape Midden 2 3 Babjarrimannos 59 16 27 73 Midden 2 3 Surfer's Beach 8 2 24 76 Midden 1 Total 37 5 14 86 mean
This analysis was undertaken as part of doctoral research at the Centre for Archaeology, UWA. I would like to thank the Aboriginal community groups in Carnarvon, Exmouth and Cardabia, particularly Sid Dale, Mary Cooyoo and Fred Clinch. Funding for this project was generously provided by AIATSIS and the Centre for Archaeology, UWA. Amy Gardos and Rodney Harrison braved the sun, wind, sand and bugs and assisted with the fieldwork, for which I am grateful. Rodney Harrison drew Figure 1, and I thank him for his comments and suggestions for this paper.
Bailey, G. N. 1977. Shellmounds, shellmiddens and raised beaches in Cape York Peninsula. Mankind 11: 132-143.
Beaton, J. M. 1985. Evidence for a coastal occupation time-lag at Princess Charlotte Bay (North Queensland) and implications for coastal colonization and population growth theories for Aboriginal Australia. Archaeology in Oceania 20: 1-20.
Bird, M. K. 1992. The impact of tropical cyclones on the archaeological record: an Australian example. Archaeology in Oceania 27: 75-86.
Bird, M. 1995. Coastal morphodynamics and the archaeological record: Further evidence from Upstart Bay, North Queensland. Australian Archaeology 40: 57-58.
Bowdler, S. 1999. Research at Shark Bay, WA, and the nature of coastal adaptations in Australia. In Hall, J and McNiven, I. J. (eds). Australian Coastal Archaeology, pp. 79-86. ANH Publications, Department of Archaeology and Natural History, RSPAS, The Australian National University, Canberra.
Clune, G. In prep. Abydos: an archaeological investigation of adaptations on the Pilbara coast, northwest Western Australia. Unpublished PhD thesis, Centre for Archaeology, University of Western Australia.
Craib, J. L. and Mangold, G. R. 1999. Storm in a test pit: Effects of cyclonic storms on coastal archaeological sites in western Micronesia. In Hall, J and McNiven, I. J. (eds). Australian Coastal Archaeology, pp. 299-306. ANH Publications, Department of Archaeology and Natural History, RSPAS, The Australian National University, Canberra.
Evans, J. L. 1991. Tropical cyclones and climate change. Search 22: 26-28.
Fullagar, R., Head, L., Bryant, T., and Beaman, S. 1999, Tsunami disturbance on coastal shell middens: McCauleys Beach, New South Wales. In Hall, J and McNiven, I. J. (eds). Australian Coastal Archaeology, pp. 235-238. ANH Publications, Department of Archaeology and Natural History, RSPAS, The Australian National University, Canberra.
Garrett, S. 1983. Coastal erosion and archaeological resources on National Wildlife Refuges in the southeast. Archaeological Services Branch, National Park Service Southeastern Region, Atlanta, Georgia.
Hughes, P. J. and Sullivan, M. E. 1974. The re-deposition of midden material by storm waves. Journal and Proceedings of the Royal Society of New South Wales 107: 6-10.
Lourensz, R. S. 1977. Tropical Cyclones in the Australian Region July 1909 to June 1975. Department if Science, Bureau of Meteorology: Meteorological Summary. Australian Government Publishing Service, Canberra.
Meehan, B. 1982. Shell Bed to Shell Midden. Australian Institute of Aboriginal Studies, Canberra.
Morse, K. 1996. Coastal shell middens, Cape Range peninsula, Western Australia: an appraisal of the Holocene evidence. In Veth, P. and Hiscock, P. (eds). Archaeological of Northern Australia: Regional Perspectives. Tempus Volume 4, pp. 9-25. Anthropology Museum, The University of Queensland, St Lucia.
Nicholls, N., Landsea, C. and Gill, J. 1998. Recent trends in Australian region tropical cyclone activity. Meteorological and Atmospheric Physics 65: 197-205.
O'Connor, S. 1989. Contemporary island use in the west Kimberley, Western Australia, and its implications for archaeological site survival. Australian Aboriginal Studies 1989 2: 25-31.
O'Connor, S. 1999. A diversity of coastal economies: Shell mounds in the Kimberley region in the Holocene. In Hall, J and McNiven, I. J. (eds). Australian Coastal Archaeology, pp. 37-50. ANH Publications, Department of Archaeology and Natural History, RSPAS, The Australian National University, Canberra.
O'Connor, S. and Sullivan, M. 1994. Distinguishing middens and cheniers: a case study from the southern Kimberley, W. A. Archaeology in Oceania 29: 16-28.
Przywolnik, K. 2002 Patterns of occupation in Cape Range Peninsula (WA) over the last 36,000 years. Unpublished PhD thesis, Centre for Archaeology, University of Western Australia.
Rowland, M. J. 1989. Population increase, intensification or a result of preservation? Explaining site distribution patterns on the coast of Queensland. Australian Aboriginal Studies 1989/2: 32-42.
Rowland, M. J. 1992. Climate change, sea-level rise and the archaeological record. Australian Archaeology 34: 29-33. Simpson, R. H. and Riehl, H. 1981. The Hurricane and Its Impact. Louisiana State University Press, Baton Rouge and London.
Spennemann, D. H. R. 1987. The impact of cyclonic surge on archaeological sites in Tonga. Indo-Pacific Prehistory Association, Bulletin 7, 75-87.
Spennemann, D. H. R 1998. Conservation management and mitigation of the impact of tropical cyclones on archaeological sites. In Spennemann, D. H. R and Look, D. W. (eds). Disaster Management Programs for Historic Sites, pp. 113-132. Association for Preservation Technology (Western Chapter), San Francisco.
Sullivan, M. E. and O'Connor, S. 1993. Middens and cheniers: implications of Australian research. Antiquity 67: 776-788.
Ulm, S. and Lilley, I. 1999. The archaeology of the southern Curtis Coast: an overview. Queensland Archaeological Research 11: 59-84.
Veitch, B. 1996. Evidence for mid-Holocene change in the Mitchell Plateau, Northwest Kimberley, Western Australia. In Veth, P. and Hiscock, P. (eds). Archaeological of Northern Australia: Regional Perspectives. Tempus Volume 4, pp. 66-89. Anthropology Museum, The University of Queensland, St Lucia.
Veitch, B. 1999. Shell middens on the Mitchell Plateau: A reflection of a wider phenomenon? In Hall, J and McNiven,
I. J. (eds). Australian Coastal Archaeology, pp. 51-64. ANH Publications, Department of Archaeology and Natural History, RSPAS, The Australian National University, Canberra.
Veth, P. and O'Brien, B. 1986. Middens on the Abydos Plain, Northwest Australia. Australian Archaeology 22: 45-59.
Wendland, W. M. 1977. Tropical storm frequencies related to sea surface temperatures. Journal of Applied Meteorology 16:477-481.
Woodroffe, C. D., Chappell, J. M. A. and Thom, B. G. 1988. Shell middens in the context of estuarine development, South Alligator River, Northern Territory. Archaeology in Oceania 23: 95-103.
Cultural Heritage Division, NSW National Parks and Wildlife Service, PO Box 1967, Hurstville NSW 2220. email@example.com
|Printer friendly Cite/link Email Feedback|
|Publication:||Archaeology in Oceania|
|Date:||Oct 1, 2002|
|Previous Article:||Beyond the limits of location: human environmental disturbance prior to official European contact in early colonial Australia.|
|Next Article:||Old boundaries and new horizons: the Weipa shell mounds reconsidered.|