Where are stone artefacts found? Testing hall's predictive model of artefact density in the south-east Australian highlands.
We use a non-site survey methodology and statistical significance testing to evaluate Hall's (1992a) predictive model of stone artefact distributions in the south-east Australian highlands. The study area is a 19.5 [km.sup.2] rectangular corridor in the Limestone Creek valley, north-east Victoria. The study confirms a number of Hall's predictions about the relationship between stone artefact density and environmental factors. The strongest relationship is between artefact density and slope. This relationship is probably a reflection of both cultural and non-cultural processes, which have promoted higher artefact density on the more subdued terrain. We also found that less common artefact types such as retouched tools and cores had the least predictable spatial distributions, although this may be a reflection of the small sample size.
Keywords: non-site archaeology, stone artefacts, predictive modelling, south-east Australian highlands.
The south-east Australian highlands have been defined by Frith (1984: 2) as a distinct biogeographical zone bordered by the upland open plain near Canberra to the north, the foothills of the Victorian Alps to the south, the coastal plain to the east and the inland plains to the west (Figure 1). The region is dominated by mountain ranges that have been formed through uplift and erosion, resulting in an elevated terrain consisting of steep ridges and valleys, as well as relatively flat tablelands. The highest peak, Mount Koskiusko, is 2250 metres a.s.l., and most of the region is over 600 metres a.s.1. The climate varies according to the season and the locality, but is generally harsh when compared with the surrounding lowland plains.
Since much of the region is not easily accessible, management of its cultural heritage is a challenging task. Despite this, numerous large-scale archaeological surveys have been conducted, often for the purpose of developing heritage management plans (Argue 1995). These plans have been designed for application over broad areas such as state forests and national parks, and so archaeologists have become increasingly concerned with the construction and refinement of predictive models. In some cases, rigorous sampling strategies have been employed, and site distribution analysis has been accompanied by concurrent studies of artefact distributions. The most detailed example of this approach is Hall's (1992a) model, developed from his survey of the Snowy River National Park.
Hall, like others before him (e.g. Byrne 1984; Flood 1980), considered that variations in the density of artefacts and sites across the highlands landscape reflected the influence of natural resources on the intensity of Aboriginal occupation. However, he believed that previous studies bad not been systematic enough in defining and sampling relevant environmental units. Hall believed that the factors that would have a strong bearing on Aboriginal use of an area would be overall resource productivity; abundance and distribution of resources, including water, plants and animals; ruggedness of terrain affecting mobility; and geology (Hall 1992a: 30). He felt that forest type was the environmental variable most likely to encompass the influence of all these factors. However, he also collected independent data on other environmental variables, including landform (Hall used the term "topography"), proximity to streams, elevation and slope.
Hall's survey covered a total of 82 kilometres of transects and recorded the distribution of over 1000 stone artefacts. Yet he acknowledged that the sample was still too small to evaluate the significance of many types of environment. For this reason, he stressed that his model was merely a first attempt at predicting the most archaeologically sensitive areas and that further survey work was desirable. Hence Hall's model consists of a series of hypothetical statements, to be subjected to ongoing testing and refinement. These statements related to the density of both sites and stone artefacts and are summarised hereunder (see Hall 1992a: 62):
* A high density of substantial archaeological sites is likely to occur within the valleys of major streams that have riparian forest.
* Artefacts and sites will occur at medium to high density on the flats and immediate river banks (fourth-order streams and above) if they are above the flood level.
* Artefacts and sites will occur at a high density on the low slopes (base of ridgesides), low spurs and creek banks that run down to rivers.
* On ridgesides adjacent to a riparian zone, artefacts and sites fade rapidly from a high density to a low density with increasing distance from the stream.
* Away from major streams, artefacts on ridgesides are negligible except where they are only gently sloping, where they occur at very low density.
* Artefacts and sites occur at a low density on ridgelines and at a high density on spurs in drier forests, although the situation in any particular locality varies a lot according to proximity to perennial water.
* Artefacts occur at a low density on the banks of first-and second-order creeks.
* In wet sclerophyll forest, artefacts are non-existent on ridgesides away from riparian zones and occur at a very low density on ridgelines.
Our aim in this study is to independently test these hypotheses through archaeological survey of a different part of the south-east Australian highlands, the area around Limestone Creek in north-east Victoria. We note that stone artefact densities are relatively lower in highlands than most other regions; thus Hall's (1992a: 61) measure of high artefact density is > 5 artefacts per 100 square metres and his measure of low artefact density is 1 to 1.9 artefacts per 100 square metres. While such distinctions may have their uses, in this study we are more concerned with estimating the general likelihood of stone artefacts occurring in particular environmental contexts. Hence we employ methods that are designed to enable the significance of relationships between environmental variables and artefact density to be tested.
THE LIMESTONE CREEK STUDY
The study area
The Limestone Creek valley is located in the east Victorian section of the Australian Alps (Figure 1). The creek itself originates on the Native Cat Tablelands (1150 metres a.s.l.) and flows for about 20 kilometres in a north-east direction until it meets the upper reaches of the Murray River (850 metres a.s.1.). A number of tributaries flow into Limestone Creek from either side, dissecting the surrounding slopes and forming a network of ridges and spurs. Some of these slopes rise to over 1400 metres, and peaks of over 1800 metres lie within 8 kilometres of the valley. Even within this relatively small area, a number of different vegetation communities flourish (Figure 2). Riparian forest and, at higher altitudes, alpine wet heathland occur close to the banks of Limestone Creek. Patches of montane forest, montane sclerophyll woodland and snowgum woodland are represented on the surrounding slopes.
Environmental stability or instability?
Hall's model predicts that the distribution of stone artefacts will be primarily influenced by the intensity of past Aboriginal occupation across the landscape. While this is a reasonable proposition and a good starting point, we need to consider that Aboriginal occupation of the south-east Australian highlands occurred over thousands of years, and that complex changes in climate and environment occurred during this period (Hope et al. 2004: 119-120; Kamminga 1992: 114-117). Hall (1992b: 129) has argued that the configuration of resource locales in areas of high relief are relatively constant through time, but the use of present-day observations of environmental conditions as key factors in the model remains a potential source of error (Holdaway & Fanning 2010). The model is only likely to be reliable if most stone artefacts observable through ground surface observation were discarded in the relatively recent past or, alternatively, if environmental conditions have been fairly stable at a regional level throughout the period of Aboriginal occupation.
Changes to the environment not only affect patterns of stone artefact discard; they can also underlie post-depositional processes that rearrange artefact distributions. Erosion and colluvial movement of sediments are characteristic geomorphological processes of highland areas, and they mitigate against the long-term survival of surface archaeological sites (Kamminga 1992:119; Witter 1984). These processes are a natural product of the steep terrain, but they are exacerbated by other natural processes, especially wildfires. Various activities by humans can also lead to increased rates of erosion.
It is therefore important to note that, since European contact, the Limestone Creek area has been subjected to significant changes in land use. By the 1860s, pastoralists were combing the area for suitable
grazing land. Gold prospecting was also being conducted by the late nineteenth century and mineral exploration has continued intermittently to the present day. Almost all of the valley currently lies within the boundaries of the Alpine National Park, and is used for limited cattle grazing, logging and recreational activities (Land Conservation Council (Vic.) 1982: 11-18, 299-301; VandenBerg et al. 1984: 23-24).
The potential effect of each of these activities on the landscape should not be underestimated. By global standards, rates of erosion in Australia's eastern highlands are low (Young 1983; Young & Wray 2000). Nevertheless, intensification of land use in areas with steep gradients often results in significant increases in erosion (Rees 1996: 34-36). Erosion leads not only to the burial of artefacts, but also may be responsible for their movement and re-exposure to the surface (Fanning et al. 2007, 2008). If these processes cause significant changes to the surface distribution of stone artefacts, then they may also affect the validity of Hall's model.
However, it is important to understand the nature and purpose of Hall's model. It draws upon some basic assumptions and observations to predict the most likely locations where stone artefacts will occur. It is intended to be a management tool, but because of the limitations of its foundations, it needs to be constantly tested and refined. As we have noted above, it is easy to provide examples of factors that may compromise the integrity of the model, but it is much more difficult to quantify the effects of these factors at a scale that can be useful for regional modelling. It is only through repeated testing of the model that we can get an indication of how significant they are.
Prior to 1996, no formal archaeological survey of the area had taken place. The data used in this paper were obtained during surveys conducted in 1996 and 1997 by third-year honours students from La Trobe University, and during subsequent survey work by the authors. For these field exercises, a rectangular corridor in the bottom half of the valley, measuring approximately 13 kilometres long and 1.5 kilometres wide, was designated as the study area (Figure 1).
The Aboriginal archaeology of forested highland areas usually consists of a wide range of site types, including rockshelter occupations, art sites, quarries, axe-grinding grooves, scarred trees, stone arrangements, bora rings, burials and camp sites. In the south-east highlands, these relatively rare site types are complemented by small surface scatters of stone artefacts, which occur "more or less continuously across the landscape" (Hall & Lomax 1996: 35). Foley (1981a,b) has noted that this is a typical characteristic of the archaeological record of small-scale, mobile societies and argued that the regional structure of this record is described most appropriately in terms of variable artefact density across the landscape. He and others (e.g. Bettinger 1977; Ebert 1992; Fanning & Holdaway 2001, 2004; Holdaway et al. 1998; Robins 1997; Thomas 1973, 1975) have developed field methods that concentrate on characterising artefact distributions across the landscape without recourse to the concept of archaeological "sites." Since the Aboriginal cultural material detected during the Limestone Creek surveys was limited to relatively small stone artefact scatters, we decided that a non-site method would best serve our research objective. This was to test whether the environmental factors incorporated in Hall's model--landform, water sources, forest type, slope and elevation had statistically significant relationships with artefact density, and to ascertain the strength of any such relationships.
Two important problems were considered during the design stage of the survey. First, it would be necessary to ensure that the area that was surveyed adequately sampled the various states of each environmental factor. Second, a means for controlling differences in ground surface visibility had to be established. A graded vehicle track that ran the length of the study area provided a reasonable solution to both of these problems (see Figure 1). Its course followed ridgelines, creeklines and intervening hill slopes and did not bypass any significant environmental zones (although no expansive plateaus were present in the study area). The track was 2-3 metres wide and generally free from ground litter, and so provided a fairly uniform, high standard of ground surface visibility. The use of vehicle tracks as survey transects in forested areas is not uncommon (Schiffer et al. 1978: 7); in fact, this was a strategy used by Hall (1992a: 45) in his study.
However, it should not be implied that the use of graded tracks is a simple solution to the problem of poor ground surface visibility. The process of grading disturbs archaeological sites and its effects may vary, depending upon the location. Since it strips the surface, it may cause surface artefacts to be buried or moved downslope. Concurrently, it may lead to greater exposure of subsurface artefacts. Hence it is possible that this strategy provides a more accurate indication of subsurface, rather than surface, artefact densities. We acknowledge that this is a potential source of error, but at Limestone Creek and many similar localities, the thickness of the vegetation precludes alternative survey strategies.
To enable an assessment of the relationship between artefact density and the environmental variables, a suitable standard sampling unit was required. After considering several practical constraints, it was decided that the survey transect should be divided into segments measuring 50 metres in length. For each segment, field workers recorded the number and type of stone artefacts, along with assessments of the landform and slope. Slope was originally measured with a clinometer, and subsequently converted into an ordinal scale variable with the values gentle (0-4 degrees), moderate (5-10 degrees), steep (11-15 degrees) and very steep (> 15 degrees) (following Hall 1992a: 61). Details of the proximity to streams and elevation were calculated from 1:250000 and 1:100000 topographic map sheets, respectively, consistent with the approach taken by Hall. A similar stream-order classification system was also applied, with the smallest unbranched streams on the 1:250000 map sheet classed as first-order streams, and stream order increasing with the number of upstream branches (Hall 1992a: 38-39; Land Conservation Council 1989: 207). Details of vegetation communities were obtained from 1:100000 scale vegetation maps, also consistent with Hall's approach (Hall 1992a: 40; National Herbarium of Victoria 1985). All of this information was compiled into a database, with the 344 segments forming the records.
Analysis of artefact distribution
A total of 479 stone artefacts were located during the survey. Of the artefacts recovered, 74% were unretouched flakes, 14% were cores and 12% were formal retouched tools. The retouched tools included notched, thumbnail and end scrapers, Bondi points and geometric microliths. The most common raw materials were quartz (which became the dominant raw material in the south-east Australian highlands over the past 2000 years: Flood 1980) and rhyolite. Other rock types represented in the assemblage included silcrete, quartzite, chert and chalcedony. Apart from the silcrete, these raw materials could have been obtained from sources within the study area. The closest known silcrete outcrop is more than 50 kilometres away (Hall 1992a: 71; A. VandenBerg pers. comm.).
Artefact density varied considerably along the transect (Figure 3). Most of the segments (77%) contained no artefacts, but in the remaining segments artefact counts ranged from 1 to 39. Figure 3 clearly shows that transects with the highest artefact densities were found near Limestone Creek, the main channel of the survey area. However, some transects with moderate artefact densities were also located some distance from streams.
The chi-squared test was used to investigate the influence of environmental variables over artefact density. Artefact density was converted to a nominal scale variable by indicating, for each segment, whether artefacts were present or absent. A more sensitive measure of artefact density would not have been able to produce adequate sample sizes.
Each environmental factor was examined separately to determine the nature of its relationship with artefact density. Chi-squared and phi-squared values were calculated, comparing the frequency of artefact-bearing and non-artefact-bearing segments among the different states of the environmental variables (Table 1). While chi-squared values indicate whether statistically significant relationships exist, the phi-squared statistic provides a means for comparing the relative strengths of these relationships (Shennan 1988: 77-79).
When all artefacts were used in the analysis, all of the environmental factors demonstrated significant relationships with artefact density. Slope provided the strongest relationship ([[phi].sup.2] = 0.33), reflecting the scarcity of artefacts on steeper surfaces. Figure 4 illustrates this relationship. It shows that artefacts were found on 45% of segments located on gentle slopes, but at the other extreme artefacts were found on only 5% of segments located on very steep slopes. A moderate relationship ([[phi].sup.2] = 0.23) was found with vegetation, with riparian forests containing a much higher proportion of artefact-bearing segments than the other vegetation communities (Figure 5).
The initial calculations of proximity to water ignored first-, second- and third-order streams, and focused only on the major channels (Murray River, Limestone Creek and Dead Horse Creek). A significant relationship was found, but artefact density did not simply decrease with distance from water as had been expected. It was felt that these irregularities may reflect the influence of smaller streams, and so a second proximity to water variable was tested, this time including third-order streams (Painter Creek, Stony Creek, Mac Creek and Greenwood Creek). In this case, there was a clear gradient in the distribution of artefacts, with decreases occurring as distance from the streams increased, particularly beyond 200 metres (Figure 6).
Weaker relationships were evident with elevation ([[phi].sup.2] = 0.16) and landform ([[phi].sup.2] = 0.16). Relatively few segments with artefacts were found at high altitudes (Figure 7), while alluvial terraces appeared to have higher artefact densities than the other landform categories (Figure 8). Relatively few segments on ridgelines, saddles and spurs bore artefacts, but we caution that saddles (ten segments) and spurs (18 segments) were not highly represented in the survey area. Moderate numbers of artefacts were found on ridgesides, but mainly on subdued slopes close to terraces or floodplains.
Further statistical tests were then conducted to analyse the distribution of three specific categories of artefacts--unretouched flakes, retouched tools and cores. The results of these tests suggest that these categories do not all conform to the same distribution pattern (Table 2). Flakes obtained very similar results to those of all artefacts, showing significant relationships with all environmental variables. In contrast, both retouched tools and cores only recorded significant relationships with slope and proximity to streams (third-order and higher), and showed no relationship with vegetation, elevation, landform or proximity to streams (fourth-order and higher). We acknowledge that the tests involving retouched tools and cores are based on small numbers of artefacts, and so their different distributions may merely be a reflection of the small size of the sample. However, it is possible that they reflect different discard contexts for these artefact types. Variation in the spatial distributions of different kinds of artefacts has been noted in numerous other non-site studies (e.g. Cochrane 1998: 71-115; Doelman et al. 2001; Douglass et al. 2008; Holdaway et al. 1998, 2000, 2004; Shiner 2006). We would expect that retouched tools and cores would be more heavily curated in the south-east Australian highlands and the data suggests that they were discarded in a greater variety of locations.
The results of the Limestone Creek survey lend qualified support to Hall's model. All of the environmental factors that feature in Hall's model have statistically significant relationships with artefact density. This means that, with respect to all of these factors, we can assert that stone artefacts are more likely to be found in some conditions and less likely to be found in others.
With regard with Hall's specific hypothetical statements, we find general agreement with the following:
* Artefacts and sites will occur at medium to high density on the flats and immediate fiver banks (fourth-order streams and above) if they are above the flood level.
* On ridgesides adjacent to a riparian zone, artefacts and sites fade rapidly from a high density to a low density with increasing distance from the stream.
* Away from major streams, artefacts on ridgesides are negligible except where they are only gently sloping, where they occur at very low density.
* Artefacts occur at a low density on the banks of first-and second-order creeks.
Like Hall, we also found that artefacts occurred in low densities on ridgelines, although we did not find high densities of artefacts on spurs or low slopes. We were unable to investigate artefact densities in wet sclerophyll forests because this vegetation type did not occur within our survey area. The prevalence of Aboriginal archaeological sites in wet sclerophyll forest remains a contentious issue (Hall 1992a: 30).
The use of significance testing in our study helps us to weigh the relative influence of each environmental variable. Slope is clearly the variable that has the strongest influence on artefact density. This is worth noting because in many landscape studies, including Hall's, the starting expectation is that the most important variables would be those that closely reflect differences in resource availability or productivity, such as forest type. In Hall's study, there was clearly a very strong relationship between slope and artefact/site density. Of all of the sites found during his survey, 66% were located on flat or gentle slopes (< 4 degrees). Almost all of the remaining sites were found on moderate slopes (5-15 degrees). Yet Hall (1992a: 61) still concluded that distance from resources was a more important constraint on site location than slope.
The relative influence of slope and other environmental variables is a matter that can be clarified through further testing of the model. Clearly, slope is an important factor, and probably for a number of different reasons. Hall (1992a: 61) emphasised the influence of slope on Aboriginal mobility; since flatter terrain was easier to traverse it would be more heavily occupied and therefore have a denser coverage of discarded stone artefacts. However, as we noted above, erosion has the potential to cause significant post-depositional changes to the surface distribution of stone artefacts, and the differential nature of these effects is related, at least in part, to slope. The use of graded tracks as survey transects ensures that the area inspected has undergone significant disturbance, perhaps promoting the downslope movement of artefacts on steeply sloping surfaces.
Despite this uncertainty about the relative roles of cultural and non-cultural factors in producing our observed patterns of artefact distributions, it is important to remember that the primary role of the model is to predict present-day site and artefact locations for the purpose of heritage management and preservation. It is true that Hall himself has interpreted correlations between predicted and actual artefact densities as confirmation that present-day surface artefact distributions accurately reflect the intensity of past Aboriginal occupation (Hall 1992b). However, we maintain that in testing the model, we are not testing a hypothesis about the land-use strategies of Aboriginal people; we are testing a hypothesis about where artefacts will be located. We have found that the model is relatively reliable, but this does not necessarily imply that present-day surface site and artefact locations are a simple reflection of past Aboriginal occupation. More likely, it shows that post-depositional processes in highland regions have similar effects to past cultural processes. Specifically, they promote higher artefact densities in flat or gently sloping terrain. More targeted investigations could help to clarify the respective roles of cultural and post-depositional processes. In their recent studies of the landscape archaeology of the Cumberland Plain near Sydney, White and McDonald (2010) have demonstrated that extensive surface and subsurface artefact distributions can be compared by applying a systematic program of excavation. Where excavations have been conducted in the south-east Australian highlands, they have revealed interesting contrasts between surface and subsurface artefact densities (e.g. Kamminga 1992: 109; Kamminga et al. 1989). A more extensive program of excavations would probably help us gain a more comprehensive understanding of the nature of the stone artefact record and its formation processes in this region.
A note of caution for heritage managers who wish to use predictive models lies in the different distribution patterns that we have detected for unretouched flakes, retouched tools and cores. In essence, this suggests that the less common types of artefact have less predictable spatial distributions. This implies that the most archaeologically significant artefacts will also be the most difficult to find. We acknowledge that the patterns are based on small sample sizes, but they are consistent with similar patterns noted elsewhere. This is certainly a subject that should be explored through further testing of the model.
Notwithstanding these caveats, the results of the Limestone Creek survey do confirm that factors reflecting resource availability do have predictive value. This is particularly the case with regard to proximity to water. One point that we have been able to clarify is the significance of third-order streams. Hall was unable to assess the influence of third-order streams in his model because they were rare in his study area. Our study has shown that proximity to third-order steams and higher has a very clear influence on artefact density. Vegetation type is also significant, although in our study this was largely driven by large numbers of artefacts found in riparian forest, which is obviously related to proximity to water.
This study confirms that Hall's predictive model is, for the most part, a useful guide for those who wish to assess the likelihood of finding stone artefacts in the variable terrain of the south-east Australian highlands. However, it also demonstrates that the spatial distribution of stone artefacts is not necessarily a simple reflection of the intensity of past Aboriginal occupation of resource zones, but more likely a complex product of a variety of factors, including ease of access through the rugged terrain, the discard contexts of particular artefact types and post-depositional processes. Further testing of the model in different localities and in subsurface contexts is required to explore these matters in greater detail.
We thank Phillip Edwards, Richard Cosgrove, Rudy Frank and the many La Trobe University students who contributed towards the collection of the data that was used in this study. We also thank Barry Kenny and Aboriginal Affairs Victoria for their assistance.
Argue, D. 1995. Aboriginal occupation of the Southern Highlands: Was it really seasonal? Australian Archaeology 41: 30-36.
Bettinger, R.L. 1977. Aboriginal human ecology in Owens Valley: Prehistoric change in the Great Basin. American Antiquity 42: 3-17.
Byrne, D. 1984. A survey strategy for a coastal forest. In S. Sullivan and S. Bowdler (eds), Site Surveys and Significance Assessment in Australian Archaeology, pp. 61-70. Australian National University, Canberra.
Cochrane, G.W.G. 1998. Toward an Unfamiliar Past: New Approaches to Spatial Analysis at Stud Creek, Northwest NSW. Unpublished BA (Hons) thesis, La Trobe University, Bundoora.
Doelman, T., Webb, J. and Domanski, M. 2001. Source to discard: Patterns of lithic raw material procurement and use in Sturt National Park, northwestern New South Wales. Archaeology in Oceania 36: 15-33.
Douglass, M.J., Holdaway, S.J., Fanning, P.C. and Shiner, J.I. 2008. An assessment and archaeological application of cortex measurement in lithic assemblages. American Antiquity 73: 513-526.
Ebert, J.I. 1992. Distributional Archaeology. University of New Mexico Press, Albuquerque, NM.
Fanning, P.C. and Holdaway, S.J. 2001. Stone artifact scatters in western NSW, Australia: Geomorphic controls on artifact size and distribution. Geoarchaeology 16: 667-686.
Fanning, P.C. and Holdaway, S.J. 2004. Artifact visibility at open sites in western New South Wales, Australia. Journal of Field Archaeology 29: 255-271.
Fanning, P.C., Holdaway, S.J. and Rhodes, E.J. 2007. A geomorphic framework for understanding the surface archaeological record in arid environments. Geodinamica Acta 20: 275-286.
Fanning, P.C., Holdaway, S.J. and Rhodes, E.J. 2008. A new geoarchaeology of Aboriginal artefact deposits in western New South Wales, Australia: Establishing spatial and temporal geomorphic controls on the surface archaeological record. Geomorphology 101: 524-532.
Flood, J. 1980. The Moth Hunters. Aboriginal Prehistory of the Australian Alps. Australian Institute of Aboriginal Studies, Canberra.
Foley, R. 1981 a. A model of regional archaeological structure. Proceedings of the Prehistoric Society 47:1-17.
Foley, R. 1981 b. Off-site Archaeology and Human Adaptation in Eastern Africa. An Analysis of Regional Artefact Density in the Amboseli, Southern Kenya. BAR International Series, 3. Cambridge Monographs in African Archaeology, Oxford.
Frith, H.J. 1984. Birds in the Australian High Country. Angus and Robertson, London.
Hall, R. 1992a. An Archaeological Survey of the Snowy River National Park. Occasional Report No. 42. The Victoria Archaeological Survey, Melbourne.
Hall, R. 1992b. Artefact density patterns in areas of high relief. A case study from tar East Gippsland. In B. Scougall (ed.), Cultural Heritage of the Australian Alps. Proceedings of the Symposium Held at Jindabyne, New South Wales, 16-18 October 1991, pp. 125-140. Australian Alps Liaison Committee, Canberra.
Hall, R. and Lomax, K. 1996. A regional landscape approach to the management of stone artefact sites in forested uplands in Eastern Australia. Australian Archaeology 42: 35-39.
Holdaway, S., Witter, D., Fanning, P., Musgave, R., Cochrane, G., Doelman, T., Greenwood, S., Pigdon, D. and Reeves, J. 1998. New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia. Archaeology in Oceania 33: 1-19.
Holdaway, S.J. and Fanning, P.C. 2010. Geoarchaeology in Australia: Understanding human-environment interactions. In P. Bishop and B. Pillans (eds), Australian Landscapes, pp. 71-85. Special Publication 346. The Geological Society, London.
Holdaway, S.J., Fanning, P.C. and Witter, D. 2000. Prehistoric Aboriginal occupation of the rangelands: Interpreting the surface archaeological record of far western New South Wales, Australia. The Rangelands Journal 22: 44-57.
Holdaway, S.J., Shiner, J.I. and Fanning, P.C. 2004. Hunter--gatherers and the archaeology of discard behavior: An analysis of surface stone artefact scatters from Sturt National Park, New South Wales, Australia. Asian Perspectives 43: 34-72.
Hope, G., Kershaw, A.P., van der Kaars, S., Xiangjun, S., Liew, P.-M., Heusser, L.E., Takahara, H., McGlone, M., Miyoshi, N. and Moss, P.T. 2004. History of vegetation and habitat change in the Austral-Asian region. Quaternary International 118-119: 103-126.
Kamminga, J. 1992. Aboriginal settlement and prehistory of the Snowy Mountains. In B. Scougall (ed.), Cultural Heritage of the Australian Alps. Proceedings of the Symposium Held at Jindabyne, New South Wales, 16-18 October 1991, pp. 101-124. Australian Alps Liaison Committee, Canberra.
Kamminga, J., Paton, R. and Macfarlane, I. 1989. Archaeological Investigations in the Thredbo Valley, Snowy Mountains: Report to Faraba Pty Ltd. Unpublished report. ANUTECH Pty Ltd, Canberra.
Land Conservation Council (Vic.). 1982. Report on the Alpine Study Area. Land Conservation Council of Victoria, Melbourne.
Land Conservation Council (Vic.). 1989. Rivers and Streams. Land Conservation Council of Victoria, Melbourne.
National Herbarium of Victoria. 1985. A Floristic Vegetation Map of East Gippsland. Department of Conservation, Forests and Lands, East Melbourne.
Rees, D.B. 1996. Land Inventotory of East Gippsland: A Reconnaissance Survey. Centre for Land Protection Research Technical Report No. 23. Land Protection Branch, Department of Conservation and Natural Resources, Melbourne.
Robins, R.P. 1997. Patterns in the landscape: A case study in nonsite archaeology from southwest Queensland. Memoirs of the Queensland Museum, Cultural Heritage Series 1 (1): 23-56.
Schiffer, M.B., Sullivan, A. and Klinger, T.C. 1978. The design of archaeological surveys. World Arehaeology 10: 1-28.
Shennan, S. 1988. Quantifying Archaeology. Edinburgh University Press, Edinburgh.
Shiner, J.I. 2006. Artefact discard and accumulated patterns in assemblage composition in surface archaeological deposits from Pine Point and Langwell Stations, western New South Wales. The Rangelands Journal 28: 183-196.
Thomas, D.H. 1973. An empirical test for Steward's model of Great Basin settlement patterns. American Antiquity 38: 155-176.
Thomas, D.H. 1975. Nonsite sampling in archaeology: Up the creek without a site? J.W. Mueller (ed.), Sampling in Archaeology, pp. 61-81. University of Arizona Press, Tucson, AZ.
VandenBerg, A.H.M., Bolger, P.F. and O'Shea, P.J. 1984. Geology and Mineral Exploration of the Limestone Creek Area, Northeast Victoria. Geological Survey Report, 72. Geological Survey of Victoria, Department of Minerals and Energy, Melbourne.
White, B. and McDonald, J. 2010. Lithic artefact distribution in the Rouse Hill Development Area, Cumberland Plain, New South Wales. Australian Archaeology 70: 29-38.
Witter, D.C. 1984. Providing a regional context for management archaeology. In S. Sullivan and S. Bowdler (eds), Site Surveys and Significance Assessment in Australian Archaeology, pp. 48-54. Australian National University, Canberra.
Young, R.W. 1983. The tempo of geomorphological change: Evidence from southeastern Australia. The Journal of Geology 91: 221-230.
Young, R.W. and Wray, R.A.L. 2000. Contribution to the theory of scarpland development from observations in central Queensland, Australia. The Journal of Geology 108: 705-719.
GRANT W.G. COCHRANE
Cutting Edge Archaeology and Department of Archaeology, University of Sydney
Departnwnt of Archaeology, University of Sydney
Matrix Archaeological Services Pty Ltd
Niche Environment and Heritage
Correspondence: Grant W.G. Cochrane, Cutting Edge Archaeology, 15/52-58 Meta St, Mooloolaba, 4557, Qld, Australia. Email: grant @ cearchaeology.com.au
Table 1. Chi-squared and phi-squared values comparing the frequency of artefact-bearing and non-artefact-bearing segments across environmental variables (all artefacts). Environmental [chi square] d.f. P [[phi].sup.2] variable Slope 36.59 3 <0.001 0.33 Vegetation 18.54 3 <0.001 0.23 Proximity to water 19.91 5 0.001 0.24 (fourth-order plus) Proximity to water 22.05 3 <0.001 0.25 (third-order plus) Elevation 8.93 3 0.03 0.16 Landform 8.35 3 0.04 0.16 Table 2. Chi-squared and phi-squared values comparing the frequency of artefact-bearing and non-artefact-bearing segments across environmental variables (unretouched flakes, cores and tools). Environmental variable [chi square] d.f. P [[phi].sup.2] Unretouched flakes Slope 23.22 2 <0.001 0.26 Vegetation 16.61 2 <0.001 0.22 Proximity to water 12.03 3 0.007 0.19 (fourth-order plus) Proximity to water 16.62 2 <0.001 0.22 (third-order plus) Elevation 7.12 2 0.03 0.14 Landform 7.50 2 0.02 0.15 Retouched tools Slope 16.54 2 <0.001 0.22 Vegetation 0.07 2 0.97 Proximity to water 5.62 3 0.13 (fourth-order plus) Proximity to water 8.68 2 0.01 0.16 (third-order plus) Elevation 1.36 2 0.51 Landform 3.80 2 0.15 Cores Slope 23.25 2 <0.001 0.26 Vegetation 4.41 2 0.11 Proximity to water 3.47 3 0.33 (fourth-order plus) Proximity to water 8.32 2 0.02 0.16 (third-order plus) Elevation 1.60 2 0.45 Landform 5.96 2 0.051
|Printer friendly Cite/link Email Feedback|
|Author:||Cochrane, Grant W.G.; Doelman, Trudy; Greenwood, Simon; Reeves, Jamie|
|Publication:||Archaeology in Oceania|
|Date:||Jul 1, 2013|
|Previous Article:||Niche production strategies and shell matrix site variability at Albatross Bay, Cape York Peninsula.|
|Next Article:||Polynesian chickens in the New World: a detailed application of a commensal approach.|