(This paper is reprinted from: Proceedings, Athens International Conference , Urban Regional Environmental Planning and Informatics to Planning in An Era of Transition (T. Sellis and D. Georgoulis eds.). National Technical University of Athens, Faculty of Architecture Dept. of Urban and Regional Planning, PP453-463, 1997) 1 INTRODUCTION: FROM MAXIMUM TO MINIMAX Approaches toward sustainability in planning can be discussed in two categories, the maximization-optimization approaches and minimax-constraint approaches. Each category is further differentiated according to economic criteria and ecological criteria, resulting in a 2 X 2 matrix (Table 1) Table 1 Major approaches to the sustainability of environment and development Maximization-optimization approaches Minimax-constraint approaches Economic criteria Economic efficiency, maximum difference between total social benefits and total social costs, based on cost-benefits analysis. Avoiding very high social costs, including: Safe minimum standards (SMS), Sustainable constraints (CS), Precautionary principles (PP),Development threshold costs, etc. Ecological criteria Ecological fitness, optimum relationship, based on suitability analysis Avoiding ecological irreversibility, including:Carrying capacity, ultimate environmental threshold (UET), etc. Both economic-maximization and ecological-optimization approaches follow the model of rationality, which depends on full information to make the best choice and a belief that knowledge can lead to the best action. The common rationale behind the minimax-constraint approaches is not to seek the best solution but to avoid the worst case. 1.1 Economic Maximization Approaches In the economic maximization approaches, projected monetary benefits and costs are used to allocate environmental and man-made capital. The basic criterion is cost efficiency, and cost-benefits models are used to search for the maximum net social benefits of using or preserving environmental assets. By means of monetary terms, depleted environmental assets can be substituted by trade-offs. The reliability of these approaches in conservation and sustainable use of the environment is, however, doubtful (Foy 1990; Pearce 1994). 1.2 Ecological Optimization Approaches The ecological-optimization approaches are based on suitability and capability analysis of the land according to physical attributes such as geology, hydrology, soil, vegetation and so on. The goal of planning is to search for the fittest environment for individual land use and activities. The ecological-optimization approaches are well known through McHarg's Design With Nature (McHarg 1969), which can be summarized as "all systems aspire to survive and succeed. This state can be described as syntropic--fitness--health. Its antithesis is entropic--misfitness--morbidity. To achieve the first state requires systems to find the fittest environment , adapt it and themselves" (McHarg 1981). The objective of landscape planning is thus the fittest plan, where individual uses of landscape match the intrinsic values of the landscape. Though widely used, ecological-optimization approach has been criticized for being physically deterministic and technocratic (Litton and Kieiger 1971). The common feature between ecological-optimization approaches and economic-maximization approaches is their technocracy, or rationality and deterministic. They are based on the assumption that there is a best solution at which planning should be aimed, and which can be revealed through complete information and full knowledge through systematic analysis. It was argued, however, that all human knowledge is fallible and uncertain, knowledge simply does not show us what we must do ( Davidoff 1965; Faludi 1987). No real decision-making process can meet the demands of rationality: complete information and the simultaneous consideration of all possible alternatives. Man does not optimize, he "satisfices", that is he looks for a course of action that is "good enough" or "satisficing" (Simon 1957, 1976). Furthermore, the economic-maximum approaches and ecological-optimum approaches are, incompatible (Pearce 1973). This recognition leads to the advancement of putting constraints on the maximization process of development as discussed in the next sections. 1.3 Economic Principle of Minimax-constraint Approaches (a) Safe Minimum Standards (SMS) Among various development constraints proposed by the economists, the concept of Safe Minimum Standard (SMS), first developed by Ciriacy-Wantrup and further developed by others (Ciriacy-Wantrup 1968; Bishop 1978), is one of the most widely discussed. It aims at minimizing the potential of the worst case such as the extinction of a species, which is irreversible and whose social cost is uncertain. The SMS principle was to deal with the problem of endangered species. It is argued that species are renewable resources within limits but have a threshold or critical zone. Once that critical zone is reached, further depletion is irreversible. This reduces the reservoir of potential resources for humanity. The long-run implications of permitting the reservoir to be reduced are unpredictable because of both social and natural uncertainties. A solution to prevent this potential catastrophe or the worst case from happening is the adoption of a safe minimum standard at which enough habitat is preserved to avoid such a catastrophe. The SMS is an application of the minimax principle rooted in game theory (von Neumann and Morgenstern 1947; Luce and Raiffa 1957). In addition to the minimax principle and SMS, several other similar and closely related concepts have been proposed by economists, such as "sustainable constraints" (CS), the "precautionary principle" (PP) and "reserved rationality" which implies the commitment of resources now available to safeguard against the potential adverse future outcomes of some decisions (Foy 1990; Perrings 1991; Pearce 1994). In all cases there is a presumption to conserve the environmental assets unless the social cost of conservation is "very high." This argument is further based on the assumption that environmental degradation beyond certain limits causes "the worst case" of large social loss for its uncertainty and irreversibility. Two questions arise: The first question is about the "worst case." The "worst case" for decision making purposes in the minimax approach is identified as being based on incomplete information, it can not be the most extreme of a known range of outcomes, since the range of outcomes is not known, nor can it be the worst imaginable case. It is thus always possible, as is argued by some scholars (Perrings 1991), to construct an excuse for any policy, in which the worst case environmental costs are infinite, but such a construction would not only paralyze all activity, it would fail utterly to discriminate between different policies . The second question concern "very high" social cost. Neither SMS nor other similar approaches have defined what is meant by "very high". It is noted that SMS, PP, and SC approaches look fine when one is considering small changes in developed countries, but the reality in the developing world is that the social cost of further land conversion to meet the need of population growth will probably be very high, but so is the cost of not converting (Pearce 1994). (b) Development Threshold Development threshold analysis was originated by Malisz in the early sixties, and further advanced by Kozlowski and others (Kozlowski 1986; United Nations 1977; Kozlowski and Hill 1993). It was first proposed for urban planning especially residential planning, when it was recognized that development often encounters some physical limitations imposed by the environment. These limitations cause discontinuity in the development processes expressed by a slowing down or even a stopping of those processes unless the limitations are overcome by additional costs of development, namely threshold costs. These threshold costs, which may be high, would not only be investment costs but also social and ecological costs. The concept of discontinuity and of additional cost are considered development thresholds. It is recognized that some thresholds, named as critical thresholds, impose distinctly greater constraints on further development in an analyzed area than do others. Overstepping of these thresholds may involve unusual difficulties (excessively high cost), and may be of critical significance in the formulation of development strategies. Thresholds that cannot be overcome by accessible technical means (at a given level of technology) or which can be overcome only at the expense of serious and irreversible damage to the geographic environment can be described as ultimate (or boundary) thresholds. These thresholds indicate the "final" boundaries of possible location, safe scale, type and timing of particular developments (Kozlowski 1986). Several limitations of threshold analysis have been recognized (Kozlowski 1986). It is basically a quantitative technique in which, conventionally various alternatives are compared by means of a single common denominator, namely threshold costs. Though other social and environmental costs are supposed to be taken into account, threshold cost virtually takes only economic cost into consideration. 1.4 Ecological Constraints of Development (a) Carrying Capacity (CC): The concept of carrying capacity (CC) is the most widely used as an ecological or environmental constraint over development. CC was originally used in biology to indicate the maximum number of individuals of a particular species that can be supported by a given area (e.g. Odum 1971). Defined as "capacity of an ecosystem to support healthy organisms while maintaining its productivity, adaptability, and capability of renewal" (IUCN/UNEP/WWF 1991), it is now broadly applied to describe the limits of environment or ecosystems to accommodate development and specific activities. The concept of CC implies that : "If we apply to our lives the rules we seek to apply when managing other species, we should try to leave a substantial safety margin between our total impact and our estimate of what the planetary environment can withstand. This is more essential because while we know that the ultimate limits exist we are uncertain at exactly what point we may reach them." (IUCN/UNEP/WWF 1991, p.43). Like the concept of safe minimum standard, the CC concept is useful in landscape and environmental planning until an operationally meaningful definition can be found. Unfortunately, efforts to determine such a definition is, in most cases, far from being successful. Carrying capacity is in most cases not an inherent and fixed quality of the site. Definitions of what constitutes an unacceptable level of impact vary with management objectives. Consequently, depending on one's objectives, any area has many possible capacities, and there are as many opinions as there are "capacities" (Held, Brickler et al. 1969). The issue here virtually becomes the acceptability of human impacts on environmental quality, and the trade-offs between loss of environment and the increase of uses, in addition to the investment required. This is especially the case when human uses of landscape in many cases have surpassed the biophysical capacity. (b) Ultimate Environmental Thresholds (UETs): The approach of ultimate environmental thresholds is a new development of the threshold analysis that is originally developed in urban planning emphasizing economic aspects of development. The new scope of threshold analysis directly addresses the regenerative capability of the environment and ecosystems. This leads to the recognition of environmental thresholds or thresholds imposed directly by natural resources some of which represent significant and specific development limits. Among these thresholds there are some final boundaries, called Ultimate Environmental Thresholds (UETs) which are: "the stress limit beyond which a given ecosystem becomes incapable of returning to its original condition and balance. Where these limits are exceeded as a result of the functioning or development of particular tourist or other activities, a chain reaction is generated leading towards irreversible environmental damage of the whole ecosystem or of its essential parts." (Kozlowski 1986, p.146). It has been argued that the definition of UETs, which can be considered as having final (boundary, ultimate) character, is of key importance at any planning level. They should play a critical role in defining concrete and "absolute" ecological limits and establishing an ecologically sound "solution space" within which development proposals would have to be generated and contained. This space, is considered to be the contribution of planning towards defining "carrying capacity." It is stated that planning must both safeguard the appropriate conservation of nature and, at the same time, guide or even stimulate socio-economic development. This contradiction can be dealt with by subdividing the planning process into two independent but mutually related strands: restrictive and promotional (Kozlowski and Hill 1993). In the restrictive strand, priority is given to ecological conservation and resource protection. In the promotional strand, planning should concentrate on elaborating a whole range of development options (scenarios) to be contained and fostered within the "solution space" determined by the restrictive strand. The main objective of the UETs method, as well as being the main contribution to planning for sustainability, is the definition of ecologically sound "solution space." Beyond this space, priority should go to the protection of natural resources. UETs, however, can hardly be used to address such a situation when planning must to face the survival of humanity as well as the survival of other species, when definitions of an economically sound "solution space" are no less important than an ecologically sound "solution space". This is an issue particularly important in the developing countries, where survival is still the most important goal of the planning. The related difficulty comes about when the two "solution spaces," the ecological (survival of species) and the economic (survival of humanity), overlap and compete with each other. Should the priority go to the sustainability of natural resources, or to the temporary survival of humanity? Above literature reviews lead to the following observations: (i) It is extremely difficulty (if possible at all) to have planning either aimed at ecological optimum or economic maximum, i.e. planning is not determined nor absolute in terms of environmental and economic criteria, it is defensible. (ii) Environment may impose some "ultimate" or "absolute" constraints on development, which planning has to come to terms with, but such limits and constraints are hardly definable nor acceptable, and therefore, have limited usefulness in planning, i.e. planning is in need of some defensible strategies to rationalize the planning process. 2. THE APPROACH OF SECURITY PATTERNS: THE CONCEPT It is assumed that there is some disproportionate irregularity, or step-type quality in the effects of landscape spatial variables on certain processes (e.g. the ecological processes of species dispersal, spread of disturbances, the visual process of perception and preference, the economic process of agricultural conversion, etc.). At some thresholds (in terms of scale, shape, number, portions or positions of landscape components), changes in landscapes result in disproportionate impacts on the processes (e.g. disproportionately impede or promote species dispersal). These thresholds are thus critical in safeguarding the processes. They are not the ultimate defense lines of the process, but they are "stop signs" for decision making, which signals process defenders and decision makers to "slow down and carefully consider". Overstepping these thresholds will disproportionately undermine the security of certain processes, and dramatically increase the risk of the irreversibility of decision making. Landscape and environmental planning should, therefore, pay attention to identifying those defense thresholds that are critical and more effective in safeguarding the landscape processes of our concern, reducing the risk of irreversibility of decision making. Each of the thresholds is associated with a certain security level for the process that one wants to safeguard. It is assumed that in correspondence with individual security levels, there exists certain spatial patterns, namely security patterns (SPs) --strategic portions, positions and their relationships-- that have or potentially have critical influences on landscape processes. SPs can most or more effectively safeguard landscape-related processes while maximally providing possibilities for changes at a certain security level. Using a case study, the next paragraph is to demonstrate that there are some kind of thresholds that can be used to define security levels and security patterns. These SPs may be useful for planners and defenders of a certain process, e.g., ecologists, to safeguard the process in a more effective way. 3. IDENTIFY SECURITY LEVELS AND SECURITY PATTERNS: A CASE The case is Red Stone National Park, China, 313 square kilometers in size. The ecological processes of species dispersal and maintenance are threatened by tourism development and agriculture . Landscape planning is aimed at providing defensible solutions for the landscape change so that the ecological processes can be safeguarded effectively at the minimal sacrifice of economic cost. The SP approach is proposed. Four steps are followed in identifying security levels and security patterns: Step 1 The target species and processes Three groups of species are targeted: medium-sized mammals (Cervidae and Viverridae families), pheasants (Phasianidae family) and amphibians (Cryptobranchidae and Ranidae families). These species are native to this region and have an endangered status. Step 2 The Source for the process Remnant forest patches are identified as the source of the process of native species dispersal and maintenance Step 3. The resistance surface of the ecological process The resistance surface that represents the dynamics of the ecological process are developed using a minimum cumulative resistance(MCR) model (Knaapen, et al 1992; Yu, 1995a), this model conceives the dynamics of species dispersal as a function of sources, distance and intermediate landscapes. Native remnant forest patches of the target species are taken as sources of dispersal. Intermediate landscapes are evaluated for their resistance to the dispersal of species, and the dynamics of the dispersal process is simulated based on the minimum cumulative resistance to the dispersal of a certain species. Comparative resistance values are assigned to various landscape attributes. Various factors such as cover, slope, elevation and aspect may contribute to the resistance value of each cell of the landscape. Figure1-2 show a resistance surface for the pheasants (other resistance surfaces see Yu, 1996a). Step 4. Security levels and security patterns From figure 1, two charts can be developed that represent two mathematical relationships (Figure 3-4): Chart 1 (Figure 3) is developed when a section is made across the resistance surface. It represents a relationship between distance (from the source) and cumulative resistance. A step-wise irregularity is recognized. It suggests that at some points away from the source, namely points a and b, resistance may increase dramatically when the species move any further. Within a certain range, landscape resistance only increase slowly and linearly. These steps are thresholds for a certain species or groups of species. Any points on the sections are actually resistance isoline on a surface (Yu, 1996a). It is therefore, argued that these thresholds (resistance isolines) can be used to define buffer zones of ecological protection at various security levels . Chart 2 (Figure 4) is developed when a histogram is developed that represents the relationship between area and resistance level in the surface. Some kind of step-wise irregularity is also recognized (see Yu, 1996a for the realistic charts). Increase in the area for ecological protection is associated with the decrease of use frequency by species in the periphery areas. At some points, the increase of the protected area may dramatically increase the accumulative resistance, namely, the chance for the species to use the increased area is dramatically reduced. These thresholds may be very useful in determining the effective area of the protected landscapes. These accessibility thresholds not only imply relatively effective points for the buffering in order to reach a certain level of ecological security, but also indicate a certain shape of effective buffer zone bounded by isolines. Buffering according to resistance further indicates where and how to change the landscape so that a better shape of buffered area can be achieved by reducing landscape resistance or smooth the intermedium, but not by simply drawing a boundary labeled as "buffer zone" (Yu, 1996a). When integrating the information from both chart 1 and 1 (Figure 3-4), the mathematical thresholds can be translated into spatial patterns, namely ecological security patterns, defined through lower resistance portions and positions of the landscape and their inter relationships. More components of security patterns can also be identified according the properties of the resistance surface. The surface model developed by W. Warntz (1966, 1967) are especially useful in spatial analysis of security levels and security patterns (Yu, 1995b, 1996a). Figure 2,5,6 are three security patterns for the protection of pheasants at three security levels: low, moderate and high. These ecological SPs can be used by ecologists as defensive frontiers for the defense of the ecological processes at various security levels in the process of landscape planning and change(Yu, 1995a,c, 1996b). 3. DISCUSSION Various alternative change models can be developed based on SPs. They include: differentiating management concentrations based on SPs; strengthening landscape infrastructures based on SPs; modifying introduced change models or trade-off SPs and exercising spatial bartering based on SPs. The essence of putting the concepts of security, patterns and changes together is that landscape planning and change is considered a procedure of defending, spatial bartering and gaming among defenders of various processes. It is the difference in values that makes landscape planning a defensible procedure (Steinitz 1979), and SPs are spatial representations of these values for individual processes represented by individual defenders. Each side of the game would probably claim his SPs as frontiers of defense and expansion. Loss of these frontiers means a dramatic threat to the security level of certain processes. Spatial bartering may result in the retreat or advance of frontiers that will alter the security level, but may also result in a rearrangement of spatial patterns without losing or reducing the security level (Yu, 1996b). SPs do not imply the final defense boundaries of the security of certain processes. The minimum safe standard, or ultimate environment thresholds, if definable, may contribute to the definition of one level of security pattern, which is, however, not the only strategic frontier, neither the first nor the last. The definition of security patterns is flexible in the sense that it corresponds to the security levels one expects to achieve. SPs are not the "dead ends", but "stop signs" where decision makers as well as defenders of various processes should "slow down and carefully consider" and maximally explore alternatives within a certain security level, and only knowingly overstepping a certain security level when none of the alternatives within this security level is "satisficing." This paper is therefor concluded with the following arguments: (i) landscape and environmental planning are not for the optimum (neither ecologically nor economically); (ii) planning should not depend on the "ultimate" or "absolute" boundaries to defend the environment or ecologically processes, since such boundaries may already have been passed or socially unacceptable or does not exist; there may be no "worst" case; (iii) planning is a defensible process and defending using some strategic frontiers at security levels (spatially security patterns) may be more reasonable and effective; (iv) security levels and patterns are definable based on the disproportionate and irregularity quality of the dynamics of the ecological processes. ACKNOWLEDGE: Thanks are due to Carl Steinitz, Stephen Ervin and Richard T. T. Forman at Harvard University for their advice and support of this research, Hugh Keegan and other staffs at the ESRI (Environment Systems Research Institute) for their support.