School of Geography

University of Leeds, LS2 9JT

Tel: +44 113 2333318

Fax: +44 113 2333308

Abstract: This paper discusses the difficulties of defining wilderness and describes how the wilderness continuum concept can be used to help identify the wilder areas of Britain. Geographical Information Systems and multi-criteria evaluation techniques are used to map the wilderness continuum for Britain and evaluate existing protected areas.

Despite there being no real wilderness areas in Britain; at least not in comparison to those found in such places as Greenland, Antarctica, Alaska and Siberia, it is argued that it is possible, using Geographical Information Systems (GIS), to identify a continuum from the most altered and accessible to the most natural and remote. Wilderness is just one extreme on this continuum or environmental modification spectrum from the "paved to the primeval" (Nash, 1982, p.1; Hendee et al., 1990). With this in mind, this paper applies GIS and multi-criteria evaluation (MCE) techniques to map the wilderness continuum for Britain. From this mapping of the continuum it is possible not only to identify the wildest parts of the country but compare the geographical location of these 'wild lands' and the boundaries of existing protected areas with a view towards identifying areas requiring new or additional protection.

Wilderness, Britain and the continuum concept

Wilderness as an entity is notoriously difficult to define with various attempts having been made in the literature. Leopold (1921), Nash (1982), Hendee et al (1990) and Oelschlaeger (1991) have all attempted academic definitions, while formal definitions such as that given by US Wilderness Act (1964) have been written for legislative purposes and are in active use. Most definitions stress the natural state of the environment, the absence of human habitation and the lack of other human related influences and impacts.

Clearly few such areas as defined by Leopold and the Wilderness Act exist in Britain today, and where they do they take the form of small and isolated pockets where a natural ecosystem has remained largely unaltered by human activity. Go back of few hundred thousand years, however, and the whole of Britain was a wilderness with no human settlement. It was only with the arrival of early humans across the then land bridge between Britain and the European continent that this wilderness began to be eroded by human incursion, settlement and forest clearance. Go back two thousand years and many parts of the country were still home to various species of wild animals commonly associated with North American wilderness; wolf, beaver, bear and lynx (Watson, 1983). Perversely, go back just a few hundred years and the areas of Scottish Highlands we may be tempted to call wilderness today were the basis of a thriving rural economy. It was only the 'clearances' of the early nineteenth century that erased these traditional crofting communities and re-instated a kind of secondary wilderness (Ridley, 1992). It must be recognised that the wild areas of Britain are part of a constantly changing environmental mosaic and that we cannot hope to maintain and preserve "vignettes" of wilderness as proposed by Leopold for North America's wild lands (Leopold, 1949).

Applying the definition of wilderness used in the Wilderness Act to the British countryside would merely result in a blank map, especially if size criteria like those outlined by Marshall (1930) and McCloskey and Spalding (1989) are applied. True wilderness simply does not exist any more in Britain. Yet, for any given area of the world it should be possible, in theory at least, to identify the most wild tract of land within its boundary regardless of the presence or absence of true wilderness areas. In this context, the above definitions of wilderness are both too qualitative and too restrictive to be useful. More flexible definitions of the wilderness continuum based on quantifiable indices and personal values are required in order to effectively map those environmental characteristics pertaining to wilderness quality.

The wilderness continuum concept states that true, pristine wilderness is just one extreme on the environmental modification spectrum (Hendee et al., 1990). At the opposite end of this spectrum is the indoor and totally urbanised environment of the city centre shopping mall or office block where a person is entirely isolated from the natural world. At all stages in between it is possible to identify various environments with varying levels of human modification and naturalness. Identifying the point on the spectrum at which wilderness begins in absolute terms is the hard part when trying to define wilderness using the continuum concept. This is because the idea of wilderness is largely one of personal experience and perception and not one of measurable standards.

Developing a GIS approach to mapping the wilderness continuum

Several authors have shown GIS to be a valuable tool for wilderness management (Hendee et al, 1990; Lesslie, 1993; Lesslie and Maslen, 1995; Carroll & Hinrichsen, 1993; Ouren et al, 1994), particularly for mapping, monitoring and analysis. A good example of using GIS to identify and map wilderness areas is given by the Australian Heritage Commission's National Wilderness Inventory (NWI). Here the Australian's have used GIS to successfully identify wilderness areas on the basis of four factors: remoteness from settlement, remoteness from access, apparent naturalness and biophysical naturalness (Lesslie, 1993; Lesslie and Maslen, 1995; Miller, 1995). These are mapped and combined by GIS overlay procedures to define a wilderness quality index. Minimum indicator thresholds are then applied in the NWI to exclude areas which do not meet minimum levels of remoteness and naturalness, as well as excluding cultural areas, thus making an absolute distinction between wilderness and non-wilderness land use.

If the more open ended approach to wilderness definition advocated by Nash (1982) is adopted, then a GIS and MCE approach to mapping the wilderness continuum is much more appropriate than other more prescriptive methods commonly used in GIS-based analyses. This is because, like the continuum concept itself, MCE methods are not restricted by the necessity to specify rigid thresholds or criteria in defining where an entity like wilderness begins and ends.

To meet a particular objective (in this case the mapping of wilderness quality) it is often necessary to consider and evaluate several criteria. This is multi-criteria evaluation or MCE. The basic aim of MCE techniques is to investigate a large number of choice possibilities (geographical locations) in the light of multiple and conflicting objectives (wilderness values). In doing so it is possible to generate rankings of the alternatives according to their attractiveness or suitability (in this case their overall wilderness quality). MCE techniques, originally developed in the planning and operations research fields for evaluating discrete decision choices between a limited number of choice alternatives, have recently been adapted for use with GIS and continuous datasets for site search and suitability mapping applications (Janssen and Rietveld, 1990; Carver, 1991; Eastman et al., 1993).

The Australian approach may be adapted, using similar factors, to fit within a GIS and multi-criteria evaluation framework to identify the wilderness continuum in Britain. The raison d'être behind this is that whereas Australia still retains large tracts of wild and primitive landscapes around which lines can reasonably be drawn for management and preservation, Britain cannot possibly afford such a luxury. By far the greater majority of the British landscape have been altered in some way, such that few really wild and primitive areas remain. In addition, the vast majority of the landscape has some economic value and even in those areas that may be considered wild this goes far beyond tourism and 'wilderness' dependent activities such as mountaineering and ski touring. For example, the wilder areas of the Scottish Highlands are much valued as a land resource for deer stalking and fishing. This makes the definition of policy boundaries on purely conservation and environmental grounds difficult and even undesirable. There are areas of the country that do, however, retain certain wilderness qualities; lack of obvious human structures, near natural self-seeded vegetation patterns, remoteness and inaccessibility. These can be identified using the continuum concept and highlighted for appropriate management. It is suggested here that a GIS/MCE approach is best suited for this purpose since it is flexible and allows individual perceptions to be taken into account.

GIS-based MCE routines are adapted and used here for describing and evaluating datasets and opinions relevant to the problem. This work focuses on defining the wilderness continuum for Britain rather than trying to identify discrete boundaries for Britain's wildest areas. By adopting a GIS/MCE approach to mapping, the minimum threshold constraints used in the Australian NWI can be safely dropped allowing the full variation in the data and statements of individual preference, made via factor weights, to be used to best effect. The key issue is reflected in Nash's statement that "one man's wilderness may be another's roadside picnic ground" (1982, p.1) in that different people have different perceptions as to what wilderness is. This not only refers to that point along the continuum at which a person considers that wilderness begins, but also to the relative importance a person may place on particular factors affecting the wildness of the landscape. Only through the incorporation of MCE type techniques can GIS effectively cope with such variability in the data and how it is applied.

Data and analysis

Several existing digital datasets are used to create four factor maps describing remoteness from population, remoteness from access, apparent naturalness and biophysical naturalness. A 1 km (0.39 square miles) resolution raster is used for the storage and subsequent analysis of all datasets.

Remoteness from population is based here on the 1991 UK Census of Population. Rather than base this factor simply on the strongest impact of a distance weighted settlement size function (Lesslie & Maslen, 1995), remoteness from population is based here on an exponential distance weighted decay model for all populated cells within a 25 km (15.5 mile) radius of the target cell. This means that the impact of a cell laying inside the 25 km radius on the target cell decreases by an exponential factor 2 (e.g. the impact of a cell located 2 km away from the target cell is 4 times less). This radius is chosen since it represents the distance an individual can reasonably walk in a day over rough terrain. The factor map shown in figure 1 is therefore more representative of true population accessibility as it is not influenced by the strongest impact of the distance weighted settlement size function but more by the total population of settlements within 25 km of a given site and the distance of those settlements from that site.

Remoteness from access is also based on an exponential weighted distance model. All forms of transport except air traffic are taken into account, namely road, rail and ferry. The weighting schemes for road, rail and ferry are shown in table 1a and 1b. The remoteness from access map is shown in figure 2.

Again, for mapping apparent naturalness the upper developed distance decay function was used by taking into account all highly visible non natural features such as radio masts, railway lines, roads, settlements, urban areas, etc. For this purpose the outline of the urban areas of the Bartholomew’s dataset was used. The weight of all the urban areas was chosen as very high (10) and the same weightings as shown in Table 1a, 1b are used for the transportation network. All other features are listed in table 2 with the specified weighting scheme. The apparent naturalness map is shown in figure 3.

Biophysical naturalness is defined as the naturalness of the land use and associated vegetation cover. This is derived here from land cover data available on the Department of the Environment's Countryside Information System (CIS). The Landsat TM derived land cover data provided by the CIS gives the total hectares (2.47 acres) per square kilometre of particular land cover types found within each 1 km grid square in Britain. These figures have been used to derive a map indicating the likelihood of finding natural or near natural ecosystems based on the weighted sum of areas of different land cover types in each grid cell. To do this land cover types were grouped into one of four 'naturalness' classes and weighted as shown in table 3. The resulting factor map is shown in figure 4.

All the above datasets were created and analysed using the GRID module in the Arc/Info GIS. Factor maps describing remoteness from population, remoteness from access, apparent naturalness and biophysical naturalness for the whole of Britain were all standardised onto a 0 to 255 scale and combined using user-specified factor weights and a simple MCE model.

The MCE method used here is a very simple one called weighted linear summation. This is shown below in mathematical form, but in plain English it simply multiplies the standardised factor maps by their user-specified weights and adds these up to create the final continuum map. Standardisation of the factor maps onto a common scale is necessary in order to allow the direct comparison of factor maps measured on different scales.

where:

Wsum = position on wilderness continuum

wj = jth user-specified factor weight

eij = standardised score

n = number of factors

Other, more complex, MCE routines exist (Carver, 1991), but a weighted linear summation model is used here for the sake of simplicity and to illustrate the principles involved.

Example results from the analysis described are shown in figures 5 and 6. By applying different factor weights, different continuum maps can be produced reflecting different people's experiential values concerning their ideas of wilderness using the MCE model. Figure 5 shows a continuum map based on user weights that stress remoteness from population and access, while figure 6 shows a continuum map based on user weights that stress apparent and biophysical naturalness. The weights used are shown in table 4.

The differences between the two maps are obvious and serve to illustrate just how different people's perceptions affect the mapped outcome. Yet, the overall geographical pattern is similar showing how the wilder ends of the continuum focus on particular regions of the country, notably the north west and central Highlands and Border regions of Scotland, the Lake District and North Pennines in England and the mountains of mid-Wales. The key point to note is that both maps are essentially correct, at least in the eyes of their originators, and both could equally be used as a basis for further work on wilderness and associated landscape conservation policy in Britain.

Evaluation of existing protected areas

Despite the focus of this work being on developing techniques for mapping the wilderness continuum and despite the problem associated with defining where along this continuum wilderness begins and ends, it is still tempting to use the above maps as a basis for identifying the wildest areas of country. This is necessary if the continuum is to be used as the basis for evaluating the coverage of existing protected areas.

The 'wild' areas shown in figure 7 are derived from the continuum shown in figure 5 by re-selecting the wildest 10% of the country. For the purpose of discussion these are overlaid with the boundaries of existing protected areas within Britain. These include National Parks, Areas of Outstanding Natural Beauty (AONBs), Heritage Coasts, National Scenic Areas, Environmentally Sensitive Areas and Regional Parks. Looking at this map it can be seen that while existing protected areas may contain landscapes of high wilderness value, a significant proportion of the wildest areas of the country are not formally protected by conservation area status. However, small parts of these areas may classified as SSSI's (Sites of Special Scientific Interest) or NNR's (National Nature Reserves) not covered by the conservation area boundaries shown here. Together with the fact that the greater majority of Britain's wildest areas are within private and not public ownership, the results shown in figure 7 have certain implications for the preservation of what is left of Britain's wild landscapes. Notably, the majority of Britain's wild lands occur in the north west Scottish Highlands. Many of these landscapes may be regarded as secondary wilderness, created during the 'clearances' and maintained subsequently by land management practices focused on deer stalking, grouse shooting, sport fishing and sheep farming. As long as land management practises are responsible and sympathetic to the environment, then these wild areas will be protected without the need for formal policy. However, as suggested earlier in the paper, the landscape mosaic of Britain is constantly changing, so some form of vigilance is required concerning land use pressures affecting the wilder parts of the country.

Policies and action plans specific to the preservation and re-creation of wild landscapes within Britain are currently being formulated by relevant organisations and conservation groups. Much interest has been generated, for example, in plans by the Royal Society for the Protection of Birds (RSPB) and Scottish Natural Heritage (SNH) for re-creating parts of the old Caledonian Pine Forest across Scotland (Featherstone, 1996), while more wide ranging studies regarding the scope for wilderness in England have been made by the Countryside Commission (Countryside Commission, 1994).

It is maintained here that if the wilder areas of the country are not formerly identified and protected then we run the risk of loosing them to the pressures of development, such as from forestry and non-wilderness dependent forms of tourism. It is recommended that some form of wild land survey and associated 'wilderness quality' index is incorporated in current management practices in protected areas such as national parks and that the results from national and local level surveys are used to help define and map new protected areas.

Carroll, C., and D. Hinrichsen. 1993. Monitoring ecological responses in wilderness using geographic technologies. International wilderness allocation, management and research. Proceedings of the 5th World Wilderness Congress. 304-308.

Carver, S. 1991. Integrating multicriteria evaluation with GIS. International Journal of Geographical Information Systems 5(3): 321-339.

Countryside Commission. 1994. The Scope for Wilderness. Report prepared for the Countryside Commission by Landscape Design Associates, Peterborough.

Eastman, J.R., P.K.A.Kyem, J.Toledano, and J.Weigen. 1993. GIS and Decision Making: explorations in Geographic Information Systems technology. UNITAR, Geneva.

Featerstone, A.W. 1996. Regenerating the Caledonian Forest: restoring ecological wilderness in Scotland. International Journal of Wilderness 2(3): 36-41.

Janssen, R., and P.Rietveld. 1990. Multicriteria analysis with GIS: an application to agricultural landuse in The Netherlands. In Geographical Information Systems for Urban and Regional Planning, H.J.Scholten and J.C.H.Stillwell, eds. Kluwer, Dordrecht.

Hendee, J.C., G.H.Stankey, and R.C.Lucas. 1990. Wilderness Management. Fulcrum Publishing, Colorado.

Leopold, A. 1921. The wilderness and its place in forest recreational policy. Journal of Forestry 19(4): 718-721.

Leopold, A. 1949. A Sand County almanac, with essays on conservation from Round River. Ballantine Books, New York.

Lesslie, R. 1993. The National Wilderness Inventory: wilderness identification, assessment and monitoring in Australia. International wilderness allocation, management and research. Proceedings of the 5th World Wilderness Congress. 31-36.

Lesslie, R. 1994. The Australian National Wilderness Inventory: wildland survey and assessment in Australia. Wilderness: the spirit lives. Proceedings of the 6th National Wilderness Conference. 94-97.

Lesslie, R, and Maslen, M., 1995. National Wilderness Inventory, Handbook and Procedures, Content and Usage (Second Edition), Commonwealth Government Printer, Canberra

McCloskey, J.M., and H.Spalding. 1989 A reconnaissance-level inventory of the amount of wilderness remaining in the world. Ambio 18(4): 221-227.

Marshall, R. 1930. The problem of the wilderness. Scientific Monthly. 30: 141-148.

Miller, J. 1995. Australian approaches to wilderness. International Journal of Wilderness. 1(2): 38-40.

Nash, R. 1982. Wilderness and the American Mind. Third Edition. Yale University Press, New Haven.

Oelschlaeger, M. 1991. The idea of wilderness: from prehistory to the age of ecology. Yale University Press, New Haven.

Ouren, D.S., J.Hummel, M.Eley, M.Sestak, and A.Riebau. 1994. Advanced technologies for wilderness monitoring and management. Wilderness: the spirit lives. Proceedings of the 6th National Wilderness Conference. 61-64.

Ridley, M. 1992. What do we do with the mountains? Daily Telegraph, December 5, 1992.

Watson, D. 1983. A brief history of the origins of the Scottish Wildlands. Wilderness: the way ahead. Proceedings of the 3rd World Wilderness Congress. 105-115.