Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Social costing for preservation decisions : the nature, significance and elicitation of wilderness values Roessler, Craig 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_roessler_craig.pdf [ 13.19MB ]
JSON: 831-1.0086291.json
JSON-LD: 831-1.0086291-ld.json
RDF/XML (Pretty): 831-1.0086291-rdf.xml
RDF/JSON: 831-1.0086291-rdf.json
Turtle: 831-1.0086291-turtle.txt
N-Triples: 831-1.0086291-rdf-ntriples.txt
Original Record: 831-1.0086291-source.json
Full Text

Full Text

SOCIAL COSTING FOR PRESERVATION DECISIONS: THE NATURE, SIGNIFICANCE AND ELICITATION OF WILDERNESS VALUES by CRAIG ROESSLER B.S.F., The University of British Columbia, 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (School of Community and Regional Planning)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA October 1993 © Craig Roessler, 1993  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of Cavtifeemii^/(16 6 /a4/4( /Jz,..9441//A,* -  The University of British Columbia Vancouver, Canada  Date U  DE-6 (2/88)  Cr  / z/ /73  -  .  ii ABSTRACT  This thesis is designed to serve the following purposes: (1) to increase awareness of values and wilderness values; (2) to highlight the significance of meaningfully including stakeholder values in allocation decision processes; and (3) to propose the use of one nonmarket valuation method which shows promise for gaining improved insight into complex land use decisions. Value concepts based on preferences offer a framework within which values can be understood and assessed. Wilderness values are thus expressed here in preference-related terms, and include human demand and spiritual values and ecological values. A taxonomy built upon these fundamental wilderness values is advanced in order to heighten understanding of them and underscore their importance in decisional and evaluative settings. Allocation decisions involving conflicting objectives, limited resources and shifting values have focussed rising attention on the need to formulate widely supported tradeoffs incorporating all values. The problem facing the planner is to provide a reasonable assessment of wilderness values, given that many of them are non-market based, complex and uncertain. Two broad classes of non-market elicitation techniques have been promoted in the literature: (1) those based on the indirect revealed preferences of consumers in markets with related commodities; and (2) those based on the direct expressed preferences of individuals. Indirect approaches outlined and assessed here are the travel cost and hedonic price methods. Their foremost strength is that they measure behavior directly. However, their utility for eliciting wilderness values is limited by their incapacity to evaluate non-use values. Direct approaches outlined and assessed here are the contingent valuation method (CVM) and multiattribute utility technology (MAUI). Expressed preference techniques share the advantage of retaining flexibility to elicit values under a range of hypothetical scenarios. However, a number of significant errors and biases have been linked to the CVM. Its  iii overriding weakness is that it typically requires holistic dollar measures for complex, incommensurable, uncertain and multidimensional values. MAUT represents a 'decomposition' approach which seems to be more in harmony with the constructive nature of human values and the coping strategies of respondents dealing with complexity. Although empirical evidence is sparse, there appears to be considerable scope for MAUT's use in allocation decision contexts. Two preservation decisions workshops were conducted to explore the feasibility of using MAUT and CVM to inform allocation decisions. Feedback from the workshops revealed that respondents, on average, found MAUT much more acceptable than CVM for eliciting their preferences. Nevertheless, they had reservations about attaching dollar figures to wilderness values with either approach, found MAUT at times confusing and generally felt that more context-building realism was needed. MAUT offers an improved means of clarifying complex decisions, but before MAUT is formally used in allocation processes there will need to be revisions in the prototype presented and larger scale trials involving stakeholders to test for its wider applicability.  iv TABLE OF CONTENTS Page  ^ii  Abstract Table of Contents ^  iv  List of Tables ^  ix  Acknowledgement ^ Chapter I. NATURE AND PURPOSE OF THE STUDY ^ Introduction Problem ^ Purpose and Objectives ^ Organization of the Thesis  1 ^1 2 4 ^5  H. AN OVERVIEW OF THE GENERAL CONCEPTS OF VALUE ^ 7 Introduction Preference-Related Value Concepts ^ Overview of Values as Preferences ^ Values as Beliefs Values as Choices and Judgments ^ Choices Judgments Held Values Values as Relations Assigned Values Intrinsic Value Summary ^ Summary of Values as Preferences III. WILDERNESS VALUES ^ Introduction Definitions Benefit Wilderness ^ Wilderness Value ^ Purpose of Constructing Taxonomy ^ Taxonomy of Wilderness Values ^ Human Demand-Related Values ^ Market Based Human Demand Values ^ Genetic Information for Products ^ Direct Use  ^7 8 8 ^9 9 ^10 ^12 ^13 ^14 ^14 ^15 17 ^18 19 ^19 ^19 ^19 20 23 24 26 26 27 28 ^29  V  Page  Chapter Ill. (cont.) Use as Natural Blueprints Use as Genetic Material Scientific Information for Resource Products Environmental Information for Media ^ Tourism Opportunities ^ Non-Market Based Human Demand Values ^ Outdoor Recreation ^ Aesthetic Existence / Option Existence Option ^ Cultural / Historical Human Spiritual Values ^ Rebuild Spiritual Health Reshape Preferences ^ Ecological Values Recoionize Damaged Areas ^ Intensify Diversification Contribute Support to Other Species Circulation / Cleansing of Air and Water Air Quality Water Quality/Quantity Climatic/Atmospheric Regulation Physical Influence Chemical Influence Summary ^ IV. INDIRECT APPROACHES TO ELICITING VALUES AND THE NATURE OF VALUE TRADEOFFS ^ Introduction Value Tradeoffs in Preservation / Harvesting Conflicts Indirect Non-Market Value Elicitation Techniques Travel Cost Method ^ Procedure ^ Applications ^ Strengths Weaknesses ^ Time Cost ^ Multiple Destination/Purpose Trips ^ Substitutes Site Congestion and Quality ^ Non-Use Values ^ Other Weaknesses ^ Hedonic Price Method ^  ^34 ^34 ^38 43 44 46 46 ^49 ^50 ^51 53 ^54 57 ^58 60 ^62 64 ^66 ^69 ^72 ^73 ^73 ^74 ^75 ^75 79  81 ^81 ^82 ^90 90 90 92 ^93 93 93 94 ^94 95 95 96 98  vi Page  Chapter IV. (cont.) Procedure Applications Strengths Weaknesses ^ Summary ^ V.  DIRECT APPROACHES TO ELICITING VALUES ^ Introduction Direct Non-Market Value Elicitation Techniques Contingent Valuation Method Procedures Open-Ended Method ^ Iterative Bidding Method Payment Card Method ^ Dichotomous Choice Method ^ Applications Strengths Weaknesses Testing Validity and Reliability^ Willingness to Pay / Willingness to Accept Discrepancy Design Biases Starting Point Bias Range Bias Payment Vehicle Bias Relational Biases Benchmarks Sequencing Embedding ^ Information Bias Motivational Biases Interaction Bias Strategic Bias Hypothetical Bias Cognitive Biases Anchoring Bias Overconfidence Bias Contextual Bias Multiattribute Utility Technology Procedure Structuring the Problem Step 1: Identify the Stakeholders to be Involved Step 2: Identify the Decision to be Made Step 3: Identify the Alternatives to be Evaluated Step 4: Identify the Relevant Dimensions of Value  ^98 ^99 ^100 100 102 104 ^104 ^104 ^104 ^104 106 ^107 107 108 ^109 ^110 ^111 111 ^114 ^121 ^121 ^123 ^124 ^126 ^126 ^127 127 ^130 ^132 ^132 ^133 ^135 ^137 ^139 ^139 ^140 ^142 ^142 ^143 ^144 ^144 ^145 ^146  ^  vii Page  Chapter V. (cont.) Determining the Importance of Dimensions ^148 ^148 Step 5: Rank the Dimensions in Order of Importance Steps 6 and 7: Translate the Rankings to Ratings and Normalize^148 ^150 Measuring Alternatives on the Dimensions Step 8: Measure the Relative Value of Alternatives on each 150 Dimension ^ Step 9: Calculate Overall Utilities for Alternatives ^153 Choice ^153 Step 10: Choose the Alternative ^153 ^155 Applications Strengths ^156 ^159 Weaknesses 161 Summary ^ VI.  WILDERNESS ALLOCATION DECISIONS WORKSHOP PROCEDURES ^ 163 Introduction ^163 Workshop Procedures ^164 Introduction / Context to Wilderness Issue ^164 Elicitation of Wilderness Values ^165 Question Sheet 1: Contingent Valuation ^165 ^165 Household Willingness to Pay for Increased Preservation ^166 Provincial Willingness to Pay for Increased Preservation ^167 Question Sheets 2 to 13: Multiattribute Utility Technology Propose / Structure Wilderness Objectives ^167 ^168 Costs Associated with Increased Preservation (Case 1) ^170 Benefits Associated with Increased Preservation ^171 Different Perspectives on Preservation Benefits ^173 Tradeoffs for Present Generation Within the Province ^174 Tradeoffs for Future Generations Within the Province Tradeoffs for B.C. Perspectives at Various Levels of Preservation ^175 Costs Associated with Increased Preservation (Case 2) ^177 Tradeoffs ^177 Feedback ^177  VII. WILDERNESS ALLOCATION DECISIONS WORKSHOP FEEDBACK QUESTIONNAIRE RESULTS ^ Introduction Feedback Questionnaire Results Summary ^ VIII.  SUMMARY AND CONCLUSIONS ^ Summary ^ Summary of the Literature Summary of the Workshop Procedures and Findings  178 ^178 ^179 192 194 194 ^194 ^198  viii Page Conclusions ^  Bibliography ^  200 206  Appendices I. EXAMPLES OF WILDERNESS DEFINITIONS ^ H. WORKSHOP QUESTION SHEETS ^ III. FEEDBACK QUESTIONNAIRE ^ W. FEEDBACK QUESTIONNAIRE RESULTS ^  247 250 264 268  ix  LIST OF TABLES Page Table Hierarchy of Wilderness Values 1.  ^27  2.  Mean Acceptability Scores  ^180  3.  Mean Ease of Use Scores  ^182  4.  Mean Comfort Scores for Assigning Dollar Amounts  ^185  5.^Mean Closeness Scores Reflecting Respondents' True Values  ^185  6^Mean Relevancy Scores for Directing Public Policy  ^186  ACKNOWLEDGEMENT  To Mona Sugawara, my lifepartner and soulmate  1 Chapter I NATURE AND PURPOSE OF THE STUDY  Introduction  Values associated with wilderness have become a central concern in British Columbia. The Stein, Carmanah and Tsitika Valleys, Clayoquot Sound, South Moresby and Tatshenshini are notable examples of wilderness areas that harbor qualities highly valued by many individuals. A diversity of values, ranging from instrumental ones accruing exclusively to humans to intrinsic ones accruing to wild species and ecological systems, have been cited by wilderness advocates as justification for preserving these forested wilderness ecosystems. The people, industries and communities which depend directly or indirectly on the extractive value of timber resources often advance positions that conflict those advocating wilderness preservation. The emotional strength of these firmly held values has led to divisive, highly publicized land-use disputes, in which wilderness advocates and forest industry interests promote objectives that appear intractable. Lobbying, public education campaigns, blockades, protests and, in rare circumstances, "ecotage" tactics have all been employed by some preservationists in an effort to stop logging and protect the wilderness values they cherish. Individuals and groups dependent on the commodity forest react in part by pointing out that timber extraction supports a high standard of living in the province by providing employment opportunities, community stability and government revenue, and that withdrawals from the commercial land base will compromise these socio-economic values. Land use decisions have often been delayed for years and even decades in such forested areas, while the provincial government has attempted to devise processes that could effectively reconcile these conflicting objectives. Their historical ability to forge broadly endorsed  2 solutions to the more protracted conflicts has to date been rare. In part these failures have been due to the inability to represent and effectively incorporate into decision processes the diverse values underlying these allocation disputes. When decisions are politically rendered, without clear consideration of the held values of interested parties, positions can become even more entrenched and the decision environment more contentious, as is occurring over the recent Clayoquot Sound decision. The disparate values that people hold are the fundamental source of such conflicts, and clearly are important to consider in such value-laden decision contexts. Understanding the nature of these values and how to most decisively elicit them would, when linked with effective public participation processes and logically structured frameworks incorporating scientific facts and sources of uncertainty, help decision makers devise more enduring and widely supported land-use allocation decisions. This thesis focuses on wilderness values, and is intended to advance an understanding of them, the value conflicts which arise from them and the evaluative methods which can be used to measure them. The remainder of this chapter provides an organizational foundation to these general themes by presenting the problem, purpose, objectives and organization of the thesis.  Problem  Although the provincial government has had difficulties grappling with land-use planning issues involving conflicting values, recent institutional developments demonstrate that it understands the seriousness of such issues and the need to address them in a responsible manner. The CORE (Commission on Resources and Environment) process is an example of this recent trend. In addition to its responsibility for developing a province-wide "land use strategy", CORE's mandate includes the coordination, facilitation and monitoring of regional planning processes, community-based participatory processes and dispute resolution systems.  3 Recognizing that traditional forms of resource allocation decision-making approaches (i.e., relying mainly on statutory decision makers) are not working in the current environment of shifting values and limited natural resources, CORE is developing "shared decision-making" approaches to land use planning and management in an effort to involve the representatives of all relevant interests with a stake in the decision (CORE 1992). Although these processes and mechanisms are only in their embryonic stage of development and use, they represent a desire on the part of the various stakeholders involved in them to resolve land-use allocation conflicts in an efficient and equitable manner. The more flexible regional framework holds particular promise for formulating land use allocation options which strive to balance general economic, social and environmental concerns. One of the more difficult challenges facing community, regional and complimentary processes is to resolve disputes centred on contentious areas, particularly old-growth forests, where both timber and wilderness values are deemed by opposing interests to be high. With the provincial government, through its "Protected Areas Strategy" (see BCMELP 1992, 1993), committing itself to an expansion in the protected areas from the existing approximate 6% of the provincial land base to 12% by the year 2000, players involved in CORE-designed and other planning processes will face increasing challenges to equitably balance the various natural resource values which lie at the core of these conflicts. There is a need, therefore, for allocation planning processes that incorporate all appropriate interests to establish effective means of explicitly exposing the values which define those interests. Given that values should be a central consideration in allocation decision contexts, the problem facing the planner is how to most validly elicit and give weight to the many values attached to natural resources in order to offer clearer, value-informed guidance to decision makers. With many being complex, poorly understood and difficult to assign monetary worth to, wilderness values are particularly difficult to elicit. Because they are often quantitatively elusive and, therefore, more likely to complicate and be a limiting factor in the  4 development of workable solutions, wilderness values and their elicitation require more thoughtful consideration in rational decision processes than they have received to date. They will thus be the predominate focus of this thesis.  Purpose and Objectives  The purposes of this thesis are to advance debate on the subject of wilderness values, underscore the importance of seriously considering them and provide some guidance on how to cogently measure them in order to help decision makers make informed tradeoffs. The specific objectives to serve this purpose are: 1. to provide a contextual basis to the specific discussion on wilderness values by sketching the general concept of "value", from both human-centered and nonhuman-centered perspectives; 2. to develop a relatively comprehensive taxonomy of wilderness values and describe the categories and sub-categories comprising it; 3. to outline the general nature of preservation / harvesting conflicts and tradeoffs; 4. to detail the methodologies of the major direct and indirect value elicitation approaches, explore where each has been applied and assess the strengths and weaknesses of each; 5. to describe the procedure of two preservation decisions workshops which elicited participants values regarding wilderness using willingness to pay and multiattribute utility techniques; 6. to provide the results of a feedback questionnaire which asked various questions of the participants regarding their thoughts and feelings on the workshop and its two elicitation methods; and 7. based upon the findings from the literature, the experiences gained from the values workshops and the lessons learned from the feedback questionnaire, to  5 deduce which approach shows the greatest promise for informing processes such as those supported by CORE, suggest possible improvements in the selected workshop method and comment on where refined approaches could be conducted to test for their wider applicability.  Organization of the Thesis  This thesis is divided into eight chapters. The first chapter presented the nature and purpose of the study. This included a brief introduction to the value-laden nature of resource use conflicts and allocation decisions, the importance of considering values and the poor success rate of the government in resolving the more polarized land-use disputes due in part to their traditional neglect of values in decision processes. The problem section then sketched the recent institutional changes and reiterated the need to incorporate elicited values into these newly emerging processes. The purpose and objectives of the thesis were then provided. Chapters II, III, N and V are the literature review segments of the thesis. Chapter II provides a generalized conceptual framework for the discussions on specific values and elicitation methods in the chapters to follow by giving an overview of the prevailing concepts of value. Although values are framed primarily within preference-based concepts, intrinsic values are also briefly sketched in the chapter. Chapter III then describes in fair detail within a hierarchically organized taxonomy the specific values of wilderness which predominate in the literature. In Chapters IV and V, the two major categories of non-market value elicitation approaches are procedurally described and their applications, strengths and weaknesses characterized, with Chapter N covering indirect, revealed preference techniques (travel cost method, hedonic price method) and Chapter V covering direct, expressed preference techniques (contingent valuation method, multiattribute utility technology). Chapter N also brings into  6 sharper focus the nature of value tradeoffs in preservation / harvesting conflicts, a topic only cursorily discussed in Chapter I. Chapters VI and VII are the study segments of the thesis. The methodology used in the workshop is discussed in Chapter VI. Chapter VII then presents the results of a feedback questionnaire which gauged participants general views of the workshop as a whole and the two direct valuation approaches used within it. The general themes from the literature review and workshop segments of the thesis are summarized in Chapter VIII. Conclusions are then drawn from theory and the empirical findings about the general value requirements of allocation decision processes and the specific approach which is proposed to help fulfill those requirements.  7 Chapter II AN OVERVIEW OF THE GENERAL CONCEPTS OF VALUE  Introduction  Philosophers addressed concepts of value when developing ethical theories well before Aristotle's time. Economists have been concerned with values since before Adam Smith. Yet, a modern concept and measurement of values did not begin until 1931, when Gordon W. Allport, Philip E. Vernon and Gardiner Lindsay attempted to measure personality traits (Rozelle 1982). A resurgence of interest in the theory of values occurred in the 1960s, with Louis Raths and his colleagues (1966) and Milton Rokeach (1968, 1973) taking the lead. Economics, philosophy, psychology, sociology, cybernetics, as well as numerous physical and biological science branches have all contributed data, concepts and research methods to the study of values (Rokeach 1979; Rozelle 1982). Such a wide array of disciplines contributing to the understanding of values means that many concepts of value have been developed and advanced. This is also true within specific disciplines. In reference to economics, Boulding (1956, 1) observes that "The word 'value' occurs in economic writing with high frequency, the frequency of meanings being about as great as the frequency of occurrence." Dictionaries themselves supply a dizzying array of meanings, running from the tangible (e.g., monetary or material worth) to the intangible (e.g., intrinsic value). The importance of the concept of value is reinforced by the central role it plays across the disparate fields concerned with human nature. Moreover, Rokeach (1979) argues that the concept of value is significant at all levels of social analyses. It will be helpful to precede the discussion on the specific values derived from wilderness with a common understanding of the general meaning of value. The development of a preference-based value theory will help allay ambiguity surrounding the concept and use of  8 the term "value". The last part of this chapter will reveal the various philosophical positions of ethicists who attach intrinsic value to non-human species and ecological systems. By broadening the understanding of value concepts beyond human-centered positions, we can see the rich array of the meanings of value.  Preference-Related Value Concepts  Overview of Values as Preferences Although many perspectives on value have arisen, most of the literature devoted to value concepts revolve around the principle of human preferences. Farnworth et al. (1981, 275-6) make the following observation: In a strict sense, all definitions of value are relativistic in either comparing one item against another item or against money. Each of these definitions establishes value as an ordinal (ranked) or cardinal (related to a standard) measure. Both ordinal and cardinal measures are relativistic means of valuation, i.e. values exist only in comparison with other things, and value exists only on the basis of human judgment and preferences. Others concur with the observation that values are generally cast in terms of human preferences (see Batson 1989; Levine and Moreland 1989; Staub 1989). Preference-based values thus factor predominantly in the definitions supplied by many authors. Williams (1979, 16) supports this assertion by stating that although the term "value" has been used to refer to interests, pleasures, duties, desires, wants, moral obligations, goals, needs, attractions, among many more, "To avoid such excessive looseness, we have insisted that the core phenomenon is the presence of criteria or standards of preference." He goes on to add the following supporting statement: All values have cognitive, affective, and directional aspects. Values serve as criteria for selection in action. When most explicit and fully conceptualized, values become criteria for judgment, preference and choice. When implicit and unreflective, values nevertheless perform "as if' they constituted grounds for decisions in behavior (p. 16). These preferential standards are organized into an individual's value system, which  9 provides guiding principles for the purposes of appropriately selecting objects and actions, resolving conflicts, invoking social sanctions and defending proposed or asserted choices (Williams 1979).  Values as Beliefs Rokeach (1973, 5) provides a consistent definition with that of Williams while also  infusing the concept of value with the concept of belief. He defines value as "an enduring belief that a specific mode of conduct or end state of existence is personally or socially preferable relative to an opposite or converse mode of conduct or end-state of existence." Many other authors foster the idea that values and beliefs are related. Equating the two in a broader definition, Epstein (1989, 4) states that "Values are beliefs about desirability that organize experience and direct behavior with respect to certain broad classes of events." What this relatedness implies is that what a person does (his behavior) is guided by what he wants (values) and what he considers to be true or likely (believes) about himself and his world (Ebert and Mitchell 1975, 49). Because values are beliefs about desirability, they correspondingly imply affect. When personal values are upheld, positive manifestations of the value are revealed through thoughts (e.g., contentment), emotions (e.g., joy) and behavior (e.g., preferences in free choice situations, or amount of effort exerted or payment made to achieve valued things or outcomes) (Epstein 1989).  Values as Choices and Judgments In addition to beliefs, preference-related concepts of human values include the interrelated ideas of choice and judgment. Most decisions that people make are not universal, but involve choices that are limited in scope. At the individual level, desirable decisions and modes of behavior occur when values and beliefs form a firm basis for judgment, which in turn allows value-informed discrete choices to be made.  10 Choices The cognitive process of choosing provides an operational conceptualization of value. Hall (1973, 50) supplies a functional definition of value when he describes it as "something that is chosen from alternatives, and is acted upon, and enhances creative integration and development of human personality." Although this defmition alludes to the fact that value, relates to the satisfaction of some notion of human needs (this will be taken up later), the emphasis here is on the process of valuing (or evaluation), as opposed to the philosophical essence of value. However, because the evaluative process is linked to that area which is subjective to human personality (i.e., fulfillment of an abstract ideal of human need), any choice that is intentionally destructive would not be considered a value (Andres 1980). Economists have extended choice theory into the domain of the concept of utility, which is defined as the satisfaction of a human want or desire. The value of a thing to any person is the utility (benefit) he could derive from it minus the disutility (cost) to him in acquiring it. Because people have limited resources they must choose among a restricted range of alternatives, the final choice depending on which alternative is most advantageous to the individual, group or society. The one chosen is said to have the greatest utility factor, given the imposed constraints, and thus the most value. The comparative values of alternative objects or deeds provide guidance for choices and decisions (Sinden and Worrell 1979; Rozelle 1982). However, a utilitarian understanding of human values restricts the rich, multifaceted nature of value to the single dimension of desire satisfaction and, correspondingly, neglects the more profound existential needs of humans. Walhout (1978, 50) alludes to the restrictiveness of this interpretation of value by stating that "Desires are clues to the existence of needs, but are not the essence of value." Utilizing the process of choosing for value clarification, Raths, Harmin and Simon (1966, 28-9) help to illuminate whether or not a person is actually holding a value or not by identifying seven criteria, which together constitute an operational definition of what they  11 consider a full value: A value must be (1) chosen freely; (2) chosen from alternatives; (3) chosen after considering the consequences of each alternative; (4) carried out, acted upon or performed; (5) acted upon repeatedly so as to develop into a pattern of life; and a value is (6) something a person feels positive about and cherishes; and is (7) publicly affirmed. An eighth criterion has since been added by Smith (1977a): (8) A value enhances a persons total growth. The authors argue that a value is not fully formed unless it fulfills all of these criteria. Moreover, as Smith (1977a, 7) asserts, "Full values are formed by a process that involves one's feelings, thoughts, desires, actions, and spiritual needs. It is a dynamic formulation, not a static one." Because an individual's values are labile, their dimensions are always changing, fluctuating, forming, developing or diminishing. As a result, most values probably do not fulfill the requisite criteria, and thus are only partial values (Smith 1977a). The magnitude of a value's property will depend on such variables as who is evaluating it and the context in which it is assessed. An assessed thing may have different value at different times, in different circumstances, under different conditions, to different people, and for different purposes (Sinden and Worrell 1979). A decision maker's opinions regarding her values will often, therefore, be far from certain, unambiguous and consistent. Supporting the declaration that preferences are often poorly defined, Tyszka (1989, 190) submits that "More and more, observations of human decision making behavior show that our preferences are far from clear in many contexts." Because such preferences form our judgments and facilitate our choices, the cognitive task of judging among and evaluating choice alternatives becomes a very tricky one. Although beyond the scope of this chapter, the difficulties inherent in the evaluative judgment process are brought up here to surround the following discussion on judgments with cautious reflection.' 'Specific difficulties in eliciting values, including those arising from the lability of values, are more fully covered in Chapters IV and V.  12 Judgments As the initial definitions indicate and the preceding discourse on choice and evaluation reveals, value judgments also factor predominantly in preference-based value concepts. In a general sense, it can be said "that one makes a value judgment when one performs the human activity of believing (accepting, subscribing to, or the like) a statement which ascribes value  or disvalue of some kind or other to a thing" (Cohen 1985, 2). Value judgments are often conceptually divided into their descriptive and normative aspects. Firstly, they describe relational situations in which some objective condition, reality or thing fulfills some human need (whether it be physiological, psychological or spiritual) or serves some inherent creative capacity that transcends need (Walhout 1978). 2 However, because people are often wholly or partially ignorant of their values, needs and creative impulses, value judgments describe situations which may or may not be real and to which human desires or likings may or may not correspond. This disharmony implies that the value situation (which includes needs and creativity), and hence the truth and falsity of the judgments describing them, is not contingent on what a person likes or thinks necessarily; rather, they are "potential value situations" which are grounded in the objective ideal of actualized values (Walhout 1978, 57). This description more or less matches Raths et al.'s (1966) and Smith's (1977a) notion of the partial value/full value dichotomy.  Secondly, as the preceding account exposes, there is a normative dimension to value judgments. Along with describing relational situations within which value lies, "...value judgments also exhibit and prescribe the norms by which we ought to weigh different value claims in order to evaluate them correctly" (Walhout 1978, 57). As Rozelle (1982, 36) notes, citing Newman and Oliver (1970), "Values are internal standards by which events or behavior are judged to be good/bad, right/wrong, fair/unfair, just/unjust. They are normative standards by which people judge the way things 'ought' to be." The normative standard by which judgments can be compared relates to the idea of 2The term 'well being' will also be used to denote both needs and creative impulses.  13 "value as fulfillment" (Walhout 1978, 49-51). Because values form the basis for human judgment, the success of the evaluative exercise will depend on whether or not the judgment results in a choice which satisfies either an existential human need (whether it be biological or spiritual, individual or social, private or common, permanent or temporary, among other forms) or a creative impulse. The essence or nature of human value can, therefore, be defined in terms of the fulfillment of either of these "springs of valuation", in all their diversity of forms (Walhout 1978, 49). The normative dimension of human values described here closely parallels Williams' (1979, 16) notion of values as "standards of preference" and Raths et al.'s (1966) and Smith's  (1977a) depiction of "full values". It also provides an overriding abstract ideal to Hall's (1973, 50) depiction of values as those qualities which enhance the growth of human personality and creativity. Because these values are values of the first order (i.e., are grounded in an abstract ideal of human well being), and are not instrumental in leading to a higher value, they are often referred to as "intrinsic values" (valuable in itself) (Godfrey-Smith 1979; Wood 1990). As will be touched on later, many individuals feel that intrinsic value can also be grounded in entities existing outside of human experience.  Held Values Brown (1984, 232) uses the term "held values" to describe these value potentialities (whether they be instrumental or intrinsic), and defines them as "modes of conduct, end-states, or qualities which could possibly be desirable." Placing held values within the "conceptual realm", Brown (1984) re-integrates human values within the organizing principle of human preferences by providing the following definition: In this sense, a value is an enduring conception of the preferable which influences choice and action. One can think of this conception of the preferable as a preference of the first order, which influences all subsequent, second order preferences (and therefore choice and action), or as that which underlies all preference (p. 232).  14  Values as Relations The discussion so far has revealed that held values (or human needs and creative capacities or human well being) are grounded in the human self. However, Walhout (1978) argues that it is in the relations between the valuing self and the entities, behaviors or events which fulfill the fundamental needs of the valuing self where human values actually exist. Value here is described not as an intrinsic quality of anything, but rather it emerges from the interaction between a subject and an object, and thus occurs within the "relational realm" (Brown 1984, 233).  Assigned Values Although the fundamental nature of human values can be either conceptually linked to the human self through held values or to the relationship between the evaluator and the evaluated, a third class of value provides a more useful means of guidance for practical decision problems. Falling within the "object realm", Brown (1984, 233) uses "assigned value" to refer to value which "is the expressed relative importance or worth of an object to an individual or group in a given context." This conception of value does not rely on an elusive absolute value standard, which can not be obtained, but instead conceives of value in terms of the worth or importance of an object, arrived at through implicit or explicit comparisons (Najder 1975; Brown 1984). Such comparative or relativistic notions of value provide a means of evaluation, where the value of an object of concern can be determined either through market mechanisms (monetary measures) or by ordinal (ranked) or cardinal (related to a standard) measures (Rescher 1969; Sinden and Worrell 1979; Farnworth et al. 1981). It is a quantitative value that can be translated into or expressed by some unit of measurement or by comparison, and can frequently be defined numerically (Najder 1975).  15 Intrinsic Value  Although the above discussion, the taxonomy of wilderness values outlined in the next chapter and the evaluative procedures delineated in Chapters IV and V concentrate on preference-based values, other more abstract values also exist for many people. The assumption so far has been that humans are the only holders of value, and thus only their needs or well being should be ultimately serviced. This anthropocentric, or human-centred, viewpoint discerns intrinsic value only in human beings, with all other values instrumental in meeting the held values of people. From this perspective, human well being is an intrinsic value because it is valued for its own sake, and is not instrumental toward any further value (Walhout 1978; Wood 1990). On the other hand, people who hold biocentric and ecocentric value systems submit that intrinsic values also respectively dwell in non-human species and natural systems, independent of whether or not humans obtain derivable value from them. In other words, non-human species and/or ecological communities have a right to exist for their own sakes, and are said to be an "end-in-themselves", not just a "means" to another's ends (Callicott 1986, 140). Resting on the broad and exceedingly complex tenets of moral philosophy and environmental ethics, a number of schools of thought have provided alternative rationales for extending moral standing to these non-human entities. One school ascribes moral standing to all sentient beings (species capable of experiencing pain or pleasure). Famous sentientists (commonly known as animal liberationists) like Pete Singer and Tom Regan share the common conviction espoused in Jeremy Bentham's famous quote: "The question is not, Can they reason? nor Can they talk? but, Can they  suffer?" (cited in Singer 1986, 25). They argue that although rationality and selfconsciousness give people moral obligations (moral agency) along with moral standing, the lack of such states, although precluding a species from assuming the role of a moral agent,  16 does not preclude their receiving moral standing. Humans thus have a moral duty to preserve sentient species for their own sakes on the basis of either their having comparable interests to those of humans (Singer 1975, 1990; VanDeVeer 1986) or their having outright inherent moral worth or "species rights" (Regan 1983, 1985, 1986b). Another group of environmental ethicists, called vitalists, expand further the moral franchise to include all living things. Because all living things are teleological centers of life (goal-seeking), such vitalists as Paul Taylor reject the notion of human superiority; vitalism asserts, therefore, that all life forms possess biologically-based interests that should be considered equally intrinsically valuable and thus be respected (Taylor 1981, 1986). Christopher Stone has even developed an ethical framework that includes the promotion of legal rights for living beings (Stone 1974, 1986). A third non-anthropocentric perspective adopts an ecocentric outlook. Labelled "holism" or "deep ecology", such a world view attributes value to the natural world, including its ecosystems and species, which is independent of instrumental value to us. Popularized by the "land ethic" advanced by Aldo Leopold (1949), the use of the biosphere and its ecosystems as ethical foundations and organizing principles has since been taken up by such authors as Callicott (1980, 1986), Lovelock (1988), Naess (1986) and Sessions and Devall (1985). Others have widened the discussion further by proposing that there are absolute or transcendental values in the universe (i.e., defmitively fixed abstractions of right and wrong), and that Humankind and other earth inhabitants are only components among many embedded within it (Reimold, Hardisky and Phillipps 1980). However, because these values lie largely outside the grasp of human apprehension, and humans thus lack the insight to elaborate them adequately, they cannot be contained within the limited scope of this discussion.  17  Summary  The above hierarchy of ethical theories illustrates that there is conviction to the belief that intrinsic value, and thus moral standing, exists not only within humans (anthropocentrism), but may also reside in non-human species (biocentrism) and natural systems (ecocentrism) as well. Indeed, they have provided insight into the formidable weaknesses in anthropocentric and related paradigms, whose lines of moral standing are often arbitrarily drawn around such criteria as personhood, rational capacities and linguistic abilities (see VanDeVeer and Pierce 1986). Without denying the merit in the various moral positions advanced, it is important to note, however, that intellectual and ethical conflicts exist even within ethical categories, resulting in competing interpretations of the qualifications required for the possession of specific rights or morally-defensible interests. Moreover, because many of these ethical frameworks have been developed without firm rooting in moral logic, critics have identified many philosophical problems with their internal logical structures. 3 The complexity and philosophical elusiveness of the morally-framed non-human intrinsic value concept make it too extensive for this paper. Therefore, the human preferencerelated theory of values will be used to provide guidance for the task of developing a taxonomy of wilderness values. Though this implies that values derived from natural areas are instrumental to human needs, it is argued that sufficient overlap exists between the fundamental interests of humans, non-humans and ecosystems to provide sufficiently strong incentives to preserve the intrinsic values dwelling in each. In addition, as we shall see in the next chapter, people may conceive that a part of such intangible values as existence value includes within its essence of meaning the intrinsic value of non-human species or ecosystems. Some individuals may also see the elements and functions of biodiversity in intrinsic terms. As such, were these 3 See  Russow (1986), Sober (1986), VanDeVeer and Pierce (1986) and Norton (1987) for overviews of the major ethical positions, along with the strengths and weaknesses of each.  18 values to be measured, their full meaning could be carried into the elicitation exercise. 4  Summary of Values as Preferences To conclude, the relationship among the preference-based realms of value will be summarized to advance an understanding of their interrelated importance. Held values (which are human needs and creative capacities grounded in the human self) act as underlying normative standards of preference which provide directional guidance for such behaviors as judgment and choice in decisions that offer choice alternatives. Besides directing behavior, upheld values are revealed through affective and cognitive responses. They also form a basis for enduring beliefs. Held values thus provide, in concert with the evaluated qualities of objects that fulfill fundamental human needs, a basis for the preference relationships which allow the relative worth or importance of objects to be expressed. Following Brown's (1984) terminology, this relativistic or comparative value is referred to as "assigned value". It is assigned value which provides an operational framework for the evaluation techniques which will be covered in Chapters IV and V. Wilderness values, on the other hand, can be conceptualized within the wider and richer notion of preference-based values, as revealed in Chapter III.  4There  is great philosophical debate over whether intrinsic value exists beyond human experience. Because the human entity remains the valuer and projects his or her values into (as well as derives satisfaction from the preservation of) the valuated entity, some would argue (see Norton 1986; Regan 1986a) that non-human entities are not being preserved for their own sakes, but rather are being protected for the psychological satisfaction of the protector, even when hidden from conscious thought; in other words, people cannot be excluded from enjoying the satisfaction attained from preserving species or natural systems. Others like Naess (1986, 505-6) take an opposite view by pointing out that "...the argument against the possibility of doing things for the sake of others is untenable."  19 Chapter III WILDERNESS VALUES  Introduction  From the tangible to the ethereal, from the marketable to the incommensurable, and  from the instrumental to the intrinsic, the range of values identified as originating from wilderness areas is extraordinary. This chapter will extract from the literature the most commonly advanced wilderness values and will describe each within an organized framework. The focus here is to provide a detailed descriptive account of the categories and sub-categories within a taxonomy of wilderness values. Before proceeding to the main topic, a few pivotal terms will be defined due to their extensive use in this chapter. This will be followed by rationales for constructing a wilderness taxonomy in the first place. The taxonomy itself is arranged into three main categories (human demand-related values, human spiritual values and ecological values), with a number of sub-categories underlying each. The general structure of the values taxonomy closely matches the one constructed by Tim McDaniels (professor, School of Community and Regional Planning, U.B.C.), in consultation with Bryan Norton (professor of public policy, Georgia Institute of Technology), as represented in McDaniels (1992, 7). The summarized set of values were also presented to and used by participants in two public values workshops conducted by Tim McDaniels (and assisted by the author) in the winter of 1992/93. The procedures of and results from the forum make up Chapters VI and VII of this volume respectively. Definitions Benefit Although economists often view benefits within the context of economic efficiency (i.e., in Pareto improvement terms), the definition here is broader in scope. A benefit is simply a desirable or improved change of state or condition, in which the gain accrues to an individual,  20 group, society, non-human species or natural system. Implicit in this definition is the notion of an advantageous consequence, which, when focussing on the issue of wilderness preservation, may take such forms as medical advances stemming from protected biodiversity, heightened spiritual connectedness or improved recreational opportunities. A benefit will also be used here to specify the prevention of a worse state from happening. Avoiding the loss of the natural functioning of an ecosystem is an example of this conception of a wilderness benefit (Driver, Nash and Haas 1985; Driver, Easley and Passineau 1990). The determination of what constitutes an improvement depends on what is judged to be preferable, which in turn is based on human beliefs and values. Values thus inform and guide the perceptual determination of whether or not a beneficial outcome has occurred. The judgment about what is beneficial will thus vary according to the person and the context. What is beneficial to one person at one time may not be to the same person at another time, and what is perceived as a benefit to one person or group may be perceived as a cost to another (Driver, Nash and Haas 1985). Wilderness  Because wilderness means different things to different people, a common definition has not been forthcoming. Roderick Nash notes this subjectivity in "Wilderness and the American Mind": "Wilderness" has a deceptive concreteness at first glance. The difficulty is that while the word is a noun it acts like an adjective. There is no specific material object that is wilderness. The term designates a quality (as the "-ness" suggests) that produces a certain mood or feeling in a given individual and, as a consequence, may be assigned by that person to a specific place. Because of this subjectivity a universally acceptable definition of wilderness is elusive. One man's wilderness may be another's roadside picnic ground (Nash 1982, 1). The literature reflects the elusiveness of the term by providing a diversity of definitions. The idea of wilderness has also been subject to the dynamic influences of society. Once considered forbidding and hostile environments, early European settlers in North America  21 often viewed wilderness as a barrier to progress and a potential defiler of the human spirit (Stankey and Schreyer 1985). Societal attitudes toward wilderness have since evolved and broadened into more accepting and positive ones, where the values of wilderness are more widely recognized. 5 Despite the elusiveness of the concept, most definitions of wilderness include the physical elements of area (size) and condition (quality). However, the magnitude of each differs between sources. The appropriate area cited as a minimum for the retention of the values sought, for example, varies from the 1000 hectares (B.C. Ministry of Forests 1989) and approximately 1900 hectares (US Wilderness Act, as cited by Bainbridge 1983) specified at the low end of the spectrum to the 5000 hectares (B.C. Wilderness Advisory Committee 1986) and even 38000 hectares (Wildlands Research Centre, as cited by McCabe 1971) at the high end. The minimum size may even be larger, depending on the area and the values that the area shelters. The B.C. Wilderness Advisory Committee (1986, 7), which was formed to look into the state of wilderness in B.C., recognized this when they suggested that "...because of the scale of natural settings and kinds of natural systems in British Columbia, units of 100,000 hectares, or larger, are often preferable for a quality experience." Besides the issue of extensiveness, the other physical attribute that all definitions of wilderness include is the condition of the natural environment. In its simplest form, wilderness denotes a condition of remoteness and naturalness. Although most of the literature include these features as basic to the meaning, controversy arises over the question of how remote and natural an area must be before it can be classified as wilderness. People involved in the purity debate ponder over such questions as "How primeval in character and primitive in appearance must an area be?" or "To what degree should any evidence of human use or occupation, such as human habitation, transportation routes or development, be allowed?" or "How remote from 5 See  Nash (1982), Stankey and Schreyer (1985), Sewell and Dumbrell (1987), Manning (1989a) and Benson (1990) for expanded accounts of the social pressures behind the development of the wilderness concept in North America.  22 settlements and lines of access must an area be?". How these questions are answered depend on the wilderness philosophies of the participants (ranging from strongly anthropocentric to strongly bio/ecocentric) and relatedly to the values that these users (or non-users) are seeking from natural areas. This brings us back to the subjective interpretation given by Nash at the beginning of this section. Agreeing that wilderness is indeed a perceived reality or state of mind, Driver, Nash and Haas (1985, 295) add, "It is not just a particular collection of natural objects; it is a collection of feelings about those objects. Thus, wilderness has more to do with the contour lines in our heads than with those on maps; it exists, in other words, in the eye of the beholder." Wilderness thus has both a concrete physical dimension and an abstract experiential dimension. Wilderness conditions or qualities may therefore be identified in a wide range of environments, including in those which may not meet the strictest definition as a physical resource (Scace and Nelson 1985). Because of the ambiguity surrounding the wilderness concept, a deliberately vague definition will be adopted for the purposes here: Wilderness is a relatively remote and intact region of land and community of life which is large enough to embody natural, self-sustainable systems, whose character and appearance are impacted and shaped primarily by the forces of nature, in which significant opportunities exist for primitive activities and solitude or scientific research, and where human's past and present influences remain largely unnoticeable, except where significant historical or cultural value exists in harmony with the surrounding natural landscape. 6  The above definition is inclusive enough to embrace a variety of natural systems and stages of successional development, with the proviso that ecosystems and disturbance patterns be formed by natural processes. Wilderness could include pioneer, early seral, late seral or oldgrowth forests. It could also be a grassland, a wetland, or any other untrammeled ecosystem. The focus in this definition is on the relationship between the ecological integrity of the system and the degree of man's influence. Human's imprint should be benign enough so as not to 6 The  elements of this definition are selectively extracted from a number of sources, with the cited source and full definition of each being offered in Appendix I.  23 disturb the natural system, disrupt wild species or their habitat, or detract from others their ability to enjoy nature. This precludes activities which require motorized vehicles within wilderness areas. This implies that accessibility should be extremely problematic for such modes of recreation. It does not rule out the existence of hiking trails. This does not imply, however, that multiple wilderness use can be accomplished on every tract of wild land. Wilderness management and single use designations may be required to separate incompatible uses. In addition, not every wilderness area harbors every wilderness value. The wilderness values that are supplied by wilderness are integrally connected to the people's motivational reasons for having a preference for a particular environment in the first place. This leads the discussion into the concept of wilderness value. Wilderness Value As has already been shown, values can be viewed within a framework which includes a person's conception of the preferable (held values), the relationship between the valuator and the evaluated, and the expressed relative importance of the evaluated entity (assigned value). Wilderness values, therefore, are those values inherent in the objects, features or processes in wilderness for which people have a comparative preference and to which they have a relational or experiential attachment. The strength of the preference and experience, and thus the significance of the wilderness value, relates to the proficiency of the wilderness environment in improving or fulfilling the well being of the attentive person. This fulfillment or enhancement of human need can be achieved through direct involvement in a wilderness-based activity (whether it be recreation or aesthetic enjoyment values) or can be derived without the present use of wilderness (such as existence or option values) (Kaplan 1984a; Pearsall 1984). The intensity or depth of the value also relates to the capacity of the wilderness to strengthen society as whole through time by perpetually supplying natural goods (such as gene resources and scientific knowledge) and sustaining natural services (such as solar energy fixation, pollutant absorption  24 and carbon and nutrient cycling) (Westman 1977; Ewert 1990b).  Purpose of Constructing Taxonomy The primary difficulty in developing a wilderness value taxonomy is that because the values are largely indivisible and their influences are often simultaneously experienced or felt, divisions within any hierarchy will not always be accurately representative of the perceptual realities of people or the ecological realities of natural systems. Yet, it is asserted that the benefits in structuring wilderness values into a hierarchy far outweigh the difficulties and limitations in doing so. The primary rationales for constructing such a taxonomy is that it promotes careful thinking about values and organizes the information on values into a logically framed and presentable format of related groupings which can be easily referred to by any number of individuals and interest groups for any number of purposes. Gaining insight into wilderness values can help interested parties in a number of ways.? Firstly, it allows individuals to promote, shape and reflect on their own personal philosophies and ethical positions. By increasing awareness into the benefits of preserving wilderness, a depth and richness is added to a person's value system. This more highly developed system of values helps instil a greater sense of confidence in a person, allowing him or her to more credibly and persuasively articulate his or her position when discussing the merits of wilderness. Secondly, clear communication of underlying values is essential for advancing debate, garnering support and developing solutions (Gibson 1966). A more knowledgeable and fluent citizenry will be more apt to participate in and be impassioned about decision-making processes. Correspondingly, they will be more widely supportive of any decision in which they played an informed and effective role. Thirdly, sounder wilderness policies and management approaches will be promoted. Having systematized information about wilderness benefits readily available will allow 7Rationales are taken from Driver, Nash and Haas (1985) and Haas (1990).  25 professionals involved in wilderness issues to develop a broad and deep body of knowledge from which informed professional judgments can be made. An advanced knowledge and skill level within the professional community will lead to more rational approaches to resource allocation decisions. Wilderness managers will also be able to define management objectives and prescriptions more clearly and set standards and guidelines for meeting those objectives more explicitly. Another advantage of promoting insight into wilderness values and benefits is that it enables professionals and laymen alike to discern which benefits are wilderness-dependent or obligatory (obtainable only from wilderness), which are mostly dependent on wilderness (obtainable from wilderness and non-wilderness, with wilderness being preferred source) and which are wilderness-independent or incidentally derived (equally obtainable from wilderness and non-wilderness alike, with indifferent preferences between them) (Dearden 1989). Such information is required to identify substitutes for wilderness uses. Fifthly, with wilderness designation and declassification decisions being undertaken within a more enlightened planning environment, there is a heightened possibility that the system of wilderness areas that are delineated will more closely reflect the public's demands and more effectively protect a greater diversity of values. Lastly, such knowledge builds and nurtures further knowledge by facilitating additional research. In addition, information on benefits and values could be used to cross-check the validity of such wilderness valuation procedures as contingent valuation, travel cost or hedonic pricing (Walsh, Loomis and Gillman 1984). Moreover, the logically-structured values taxonomy is an essential information source for participants engaged in public value forums which include multiattribute utility techniques as part of the value elicitation procedure. As such, they are essential for making value tradeoffs and evaluating options.  26 Taxonomy of Wilderness Values  It was mentioned at the beginning of this chapter that the wilderness values making up the following taxonomy were developed and structured by Tim McDaniels and Bryan Norton. Relying on Norton's expertise on the policy and philosophical issues connected to the valuing of biodiversity benefits, McDaniels conducted a telephone interview with Norton soliciting his views on the values surrounding biodiversity. The initial inventory of values were drawn from Norton's book entitled "Why Preserve Natural Variety?" (Norton 1987). After discussing the book in detail, Norton was asked to clarify his views on biodiversity values and make any modifications to the original values set rendered in his book The resulting list, drawn largely from McDaniels (1992), is shown in Table 1. This hierarchy is similar to Norton's clarified list, with some additions made during the preparation phase of the values workshop. The list contains three main categories: human demand-related values, human spiritual values and ecological values. We will now turn our attention to the elaboration of the wilderness value categories and sub-categories making up this taxonomy.  8  Human Demand-Related Values  Human demand values are those instrumental values derived from wilderness areas and their complement of floral and faunal species that either directly or indirectly satisfy some felt human preference. The magnitude of the intensity of a felt preference corresponds to the degree to which a naturally obtained or experienced good or service meets a human need. Human demand-related values can either be procured through markets or through experiences and activities lying outside market forces (Norton 1987).  8 For  good examples of alternative lists see Inland (1979), Myers (1979), Rolston (1985, 1988), Manning (1989b), Dixon and Sherman (1990), Driver, Easley and Passineau (1990), McCloskey (1989, 1990), CEAC (1991) and Spellerberg and Hardes (1992).  27 Table 1 Hierarchy of Wilderness Values Humand Demand-Related Values Market Based Genetic Information for Products Scientific Information for Resource Management Environmental Information for Media Tourism Opportunites Non-Market Based Outdoor Recreation Aesthetic Existence / Option Cultural / Historical Human Spiritual Values Rebuild Spiritual Health Reshape Preferences Ecological Values Recolonize Damaged Areas Intensify Diversification Contribute Support to Other Species Circulation / Cleansing of Air and Water  Market Based Human Demand Values A number of marketed or marketable products with tangible economic value originate from wilderness. The market prices of these various naturally-derived products generally reflect the values people place on them when the assumptions of the market model (i.e., unrestrained competition) are met. Sometimes referred to as "productive use values" (McNeely 1988; McNeely et al. 1990a), these market values are instrumental to achieving other human goals, such as when good nutrition and safe medicines, often obtained from information contained within wild genes, lead to improved health and survival (Leitzell 1986). Genetic information for products comprises the first commercial value category. The second category,  28 which also has direct informational value, is the scientific knowledge gained from researching the unencumbered functioning of natural systems. Such knowledge can be directed towards improving the manner in which commodity forests or other natural resources are managed. Thirdly, with concern and interest in nature and environmental issues rising in recent years, wilderness areas also serve as environmental source material for the electronic and print media. Lastly, natural areas in B.C. lure tourists, who come here and spend money with the aim of experiencing "Super Natural British Columbia".  Genetic Information for Products Wild species, whether they be plants, animals, fungi or microorganisms, provide and can potentially provide a vast array of usable resources for Humankind. Species reliant on wilderness are reservoirs of exploitable genetic material and sources of new domesticates. Prescott-Allen and Prescott-Allen (1986, 275) use the term "wild genetic resources" to describe both the usable genetic and species organizational levels when they define it as "any heritable characteristic of a wild plant or animal that is of actual or potential use to people. The characteristic may be disease resistance, the presence or absence of a chemical, or any other yield or quality factor. As long as it is or is likely to be of economic or social value, is found in wild species, and is transmitted genetically, it qualifies as a wild genetic resource." Wild genetic resources can be conserved in two ways: either through in situ (sustained within the natural ecosystems and habitats in which they naturally reside) or ex situ  (maintained in storage environments such as gene banks, clone banks, seed storage banks, tissue cultures and DNA libraries or in zoos, aquaria, botanical gardens, test plantations and seed orchards) strategies (Oldfield 1984; Ledig 1988; Riggs 1990). It is asserted that the genetic integrity of genetic resource products is most effectively preserved in situ, due in part  to the practical difficulties, limitations, costs and inferior genetic traits associated with the ex  situ conservation approach (Oldfield 1984; Flamm 1989). Also, it must be recognized that all of the evolutionary forces shaping the adaptive potentialities of genes occurred in the  29 wilderness, not in civilization (Nash 1982). Although wilderness preservation is a superior approach for sustaining a more complete representation of the species' original gene pool, ex  situ strategies cannot be ignored. By protecting endangered populations for reintroduction, conserving crop genetic resources (cultivars) which traditionally grow in disturbed agricultural landscapes, providing educational and public awareness services and acting as research facilities, ex situ should be considered a complimentary strategy to in situ techniques (Williams 1988; Reid and Miller 1989; Burton et al. 1992). The benefits of using wild genetic resources may be obtained in three general ways. First, a species can be utilized directly, either through direct consumption or as a direct source of wild products, domesticates or natural chemicals and compounds. Second, the genetic constituents of wild species can provide informational blueprints to researchers developing synthetic chemicals and compounds. Third, they can act as sources of genetic material from which desired genetic traits are transferred and employed using breeding or biotechnological methods (Sedjo 1992).  Direct Use Direct consumptive uses of wild species include sports or subsistence fishing, hunting, trapping and collecting. Because these activities involve the harvesting and use of plant and animal species without their passing directly through markets, estimating their dollar value to the economy is difficult. Most economists estimate the value of such activities within the purview of a whole recreation experience (McNeely 1988). For example, 84 percent of the Canadian population participates in wildlife-related recreational activities in a given year, which they declare is worth $800 million annually (Filion, Jacquemot and Reid 1985). However, such analyses are not refined enough to discern the level of dependency of the activities which wildlife supports have on wilderness. They also ignore subsistence use of wildlife, which may be supremely important from a survival and cultural point of view, particularly for some indigenous groups and individuals. Because these consumptive use values (McNeely 1988;  30 McNeely et al. 1990a) are not market based, no more will be said about them here. Their fuller importance is revealed later in the paper within the section "Non-Market Based Human Demand Values". Other examples of direct uses of wild species in which market values are attached is commercial logging, fishing, trapping and collecting. Because logging is antithetical to wilderness and commercial fishing's degree of dependency on it is unknown, only trapping and collecting will be represented here. Besides acting as an indigenous activity steeped in culture and custom, the economic benefits of the fur and skin trade to native and non-native trappers alike are significant, particularly for those living in northern communities. The average annual value (at harvest) of furs and skins from wild Canadian mammals during the period of 1976 to 1980 was $54 million U.S.. The most important wild furbearers, such as beaver (over $12 million/annum), muskrat (over $9 million), lynx (over $5 million) and red fox (approximately $5.1 million) rely heavily on undisturbed expanses of wilderness for their various habitat requirements (Prescott-Allen and Prescott-Allen 1986). There has also been a traditional association between indigenous cultures and the collecting of wild species for food. Examples of wild food species which have since been commercialized but remain undomesticated in Canada include wild blueberries, maple syrup/sugar ($483 million and $22 million respectively 9 ), wild rice and fiddlehead ferns ($1.9 million and $24 thousand respectively 10) (Prescott-Allen and Prescott-Allen 1986). Wild flora and fauna are also sources of new domesticates. Although most wild plant species were domesticated before A.D. 1900, some recent examples include cashews, kiwi fruit, lavender, wheatgrass, highbrush blueberries and wild rice, the latter four of which can be found in Canada. New and incipient animal domesticates in Canada are represented by such species as mink and fox for furs, earthworms for bait and garden soil enhancement and domesticated and semi-domesticated species of fish in the aquaculture industry (Prescott-Allen 9 Per annum approximate $U.S. earnings from 1976 to 1980. 10Per annum approximate $U.S. earnings from exports to U.S.  from 1976 to 1980.  31  and Prescott-Allen 1986). Besides providing the original base for domesticated plants and animals, wild species contain highly valued natural chemicals and compounds. The pharmaceutical industry and medical community are particularly indebted to chemical substances extracted from the wild. Up to half of the world's medicinal compounds are currently derived or obtained from plants (Oldfield 1984). These have a 1982 value of about $40 billion (Spellerberg and Hardes 1992). When the social benefits of improved health are factored in (wages saved, health care costs averted, etc.), the pharmaceutical economic value of plant-based drugs is estimated world-wide to be as high as $1.8 trillion (McNeely 1988). Although direct medicinal products from plants are generally more important in developing nations, even developed nations (which tend to focus on synthetic products) rely on biotic drug products as major contributors to public health (Oldfield 1984; Farnsworth 1988). For example, approximately 25% of all prescription drugs dispensed in the United States contain active ingredients extracted directly from plants (Farnsworth and Soejarto 1985; Farnsworth 1988; Ledig 1988; Reid and Miller 1989). In all, about one-half of all prescriptions dispensed in the U.S. contain a drug of natural origin (Myers 1979; Driver, Nash and Haas 1985). The estimated commercial value of such drugs and medicines in the U.S. in the mid 1980's ranged from $8 billion (Farnsworth and Soejarto 1985; Ledec and Goodland 1988) to $14 billion per year (Myers 1983; WCED 1987). 11 Worldwide, approximately 119 pure chemical substances extracted from vascular plants have medicinal properties (Farnsworth 1988). Many others are derived from non-plant sources. Extensive lists of these can be found in Oldfield (1984), Prescott-Allen and PrescottAllen (1986) and Farnsworth (1988). The biochemical diversity of wild and cultivated plants, animals, fungi and microorganisms have contributed analgesics, anti-cancer agents, antibiotics, cardio-active drugs, enzymes, hormones, muscle relaxants, anti-coagulants, contraceptives and abortifacients, among others (Myers 1981; Train 1981; Caufield 1985). 11 These figures include both cultivated medicinal plants and their wild ancestors.  32 Because tropical rainforests are the most biologically diverse systems on earth, the vast majority of drug-yielding plants originate from them. However, there are also examples in British Columbia of wild plants with medicinal properties. These include the Common juniper berry (77 12 , used as a diuretic), Oregon grape (40, antipyretic), Skullcap (37), Nettle (31), flower of Elder (26, diuretic, cathartic, diaphoretic), Arnica spp. (20, topical counterirritant), Pipsissewa (15, urinary antiseptic), Scouring rush (11), Senecio spp. (10, emmenagogue), among many more (Lyons 1952; Prescott-Allen and Prescott-Allen 1986). However, by far the most lucrative medicinal bioresources in the province are drugs contained within the barks of Cascara and Pacific yew. The most prolific plant in North America, in terms of numbers of drug products produced, is Cascara (Ramnus purshiana). Native to B.C., along with the western U.S., the Cascara tree contains the drug cascara sagrada within its bark. Yielding 199 products in all, cascara sagrada is best known as a laxative. With B.C., Washington and Oregon providing most of the present-thy market supply, the three account for the majority share of a U.S. laxative market which topped $75 million in 1975 (Prescott-Allen and Prescott-Allen 1986). Another example of a highly prized natural chemical is taxol, a promising anti-cancer compound which develops naturally in the bark of Pacific yew (Taxus brevifolia), a small, slow-growing understory tree dwelling mostly in old-growth forests along the slopes and valleys of the Cascade and Rocky Mountain ranges of B.C. and the Pacific Northwest of the U.S.. Approved recently by the U.S. Food and Drug Administration for general use in advanced ovarian cancer, taxol has shown promise in other types of cancers as well, including cancers of the breast, uterus, cervix, blood, colon, lungs, pancreas and prostate (Norse 1990; Hoffman 1991; Joyce 1993). The drug's remarkable anti-tumor properties have been demonstrated in 20% to over 50% of terminal cancer patients, with the high value reflecting the success rate in arresting cancer cells in breast tissue (Sedjo 1992; Joyce 1993). As a result of 12Numbers refer to numbers of products produced. Therapeutic categories are also included when available (source: Prescott-Allen and Prescott-Allen 1986, 111-28).  33 the exceptional results from experimental treatments, many researchers and physicians expect it to be approved for cancers other than ovarian soon (Joyce 1993). With recent studies revealing taxol's unique mechanism of action and abilities in treating tumors which have not responded to conventional treatments such as chemotherapy, many physicians are calling taxol the most exciting new cancer drug in more than a decade (Sedjo 1992; Joyce 1993). Because old-growth-dependent Pacific yews (Pearson 1990) are currently the only dependable source of taxol, protection of the compound will require the preservation of old-growth, at least until other sources are found or manufactured. Synthetic duplication of the compound remains a difficult task due to the complexity of taxol's chemical structure (Joyce 1993). Other methods, such as conversion of structurally similar compounds into taxol or generating it from plant tissue cultures, will not likely succeed for some years yet (Sedjo 1992). The concern of botanists and others wanting to protect the bioresource and blueprint values of taxol is that with the Pacific yew existing in fairly small numbers and with extraction of taxol requiring the stripping of the yew's bark which destroys the tree (yet yields very little drug), demand for the drug far outweighs supply (Zuckerman 1991; Burton et al. 1992). A coordinated approach of preservation, extraction and development of new sources will need to be undertaken in order to meet current and future demands and protect the natural chemical in  situ for potential alternative uses (e.g., for highly immune future strains of cancer). In any event, Joyce (1993) discloses the importance of the wild strain, which has developed over millions of years through natural selection, when he states the following: The compound has now spawned a multi-billion dollar industry and an international effort to synthesize it. Moreover, it has revived interest in medicinal botany, for millennia the centerpiece of medicine but recently relegated to natural food stores and herbalists....Taxol has made nature's case - that plants still offer surprising new chemicals. As biologists in the natural products field like to say, natural selection is a lot more inventive than human chemists (pp. 133, 136). Though some drugs formerly cultured from wild organisms are now manufactured  34 synthetically, many of the most valuable compounds, like taxol, have complex chemical structures which are impossible, or at the very least difficult and costly, to synthesize (Frome 1974; Oldfied 1984; Ledec and Goodland 1988). In addition, interest in drug compounds derived from natural sources has increased in recent years due in part to a renewed understanding of the need to replace synthetic drugs which have lost their effectiveness in combatting virulent strains of specific diseases (Oldfield 1984). In any case, natural sources remain the key building blocks from which the duplicated compounds are modelled (Frome 1974; Oldfield 1984; Ledec and Goodland 1988).  Use as Natural Blueprints Natural drug products often act as chemical blueprints for the development of related synthetic drugs. As mentioned earlier, natural, pharmacologically active compounds tend to be very complex in chemical structure. As a result, their a priori chemical synthesis is unlikely without the use of such natural model compounds. Although artificially derived drugs are rarely as effective as their parent compounds, artificial drugs carry certain advantages, such as purity and dosage control. Natural extraction and artificial synthesis should, therefore, be considered complimentary research and economic activities (Oldfield 1984).  Use as Genetic Material Wilderness areas, besides protecting species diversity, also retain the reservoir of genetic material contained in those species. Genetic diversity, the smallest level of biodiversity, is the total amount of genetic information encoded on the genes of all living organisms inhabiting an intact natural area. It maintains variability between populations of a given species and between individuals in a given population of a species. Evolving over thousands to millions of years through natural selection, such genotypic diversity enables populations to more readily adapt to environmental changes and enables individuals to maintain health, fertility and longevity (Foose 1986; Department of Natural Resources 1991). Besides producing medicinal and nutritive compounds for direct consumption or  35 derivative use, genes and gene complexes produce adaptive genetic traits which can be selectively exploited through breeding and recombinant DNA technology. By discerningly employing and manipulating the ancestral diversity contained within genes, geneticists can advance and develop new and improved varieties. Preserving gene pools in situ, therefore, safeguards the exercising of resource options by sheltering potential new products (Cutler 1980). Breeding, although not explicitly specified, was alluded to earlier when the use of wild species as domesticates was sketched out. Plant crop species, for example, have been selectively bred from wild strains on a very large scale, resulting in thousands of domesticates representing millions of genetic varieties. At one time or another, human beings have used from between 3000 to 7000 plant species in all as food (Ehrlich and Wilson 1991; Spellerberg and Hardes 1992). However, humans have narrowed the food base by cultivating on a large commercial scale only 150 species, with less than twenty producing 90 percent of the world's food supply (Myers 1979). Over half of the Earth's total cropland is devoted to growing human's three principal food items: rice, wheat and corn (Ehrlich and Ehrlich 1981). The plants that are consumed on a regular basis therefore represent a very small proportion of the 75,000 to 300,000 which are estimated to be edible (Myers 1979; Ehrlich and Ehrlich 1981; Wilson 1984; Rolston 1988). This untapped food resource means that the undiscovered or undeveloped value of wild edible plants is very high (Ehrenfeld 1976). Many unexploited and under-utilized wild plants hold as much potential value as future cultivars as the ancestral plants of today's domesticated crops did years ago (see examples given by Myers 1979; Ehrlich and Ehrlich 1981). Along with expanding the supply of directly consumable food products, these untapped biotic resources are essential for protecting the food supplies we presently have. Modern breeders, with the aim of developing high-yield varieties, have concentrated on hybridizing fewer and fewer varieties which are further and further removed from their wild ancestors.  36 This narrowing of the genetic base, along with the modern practice of large-scale monocultivation, has made crops vulnerable to pests and diseases and has reduced their tolerances to variations in environmental conditions (Myers 1979; Bunnell and Williams 1980b; Ledig 1988; Perfect 1991; Spellerberg and Hardes 1992). Given that vast areas are planted to a few major, genetically-restricted cereal species, the genetic diversity within each species is required to keep agriculture resilient and stable (Wolf 1985; Perfect 1991). Modern breeding programs have begun to recognize this by incorporating the desired genes from wild stock into crop plants through a careful process of cross-breeding and screening. Employing the natural defenses of wild plants by cross-breeding them with their domestic relatives enhances the latter's resistance to pests and diseases. Other traits, such as high productivity and tolerances to weather extremes, high salinity and low light conditions, have also been successively integrated into crop species (Myers 1979; Wolf 1985; Norse et al. 1986; Spellerberg and Hardes 1992). Though much further behind in development than agricultural breeding, tree breeding also shows immense promise in improving the survivability and tolerance of seedlings (Burton et al. 1992). The significance of genetically improving cultivars through the transferal of robustness-conferring genes will grow as biotechnology becomes increasingly more common. Recombinant DNA technology, currently experimental, will allow genetic engineers to transfer useful, adaptive traits between species that could never be crossed by conventional breeding methods (Wolf 1985; Ledig 1988). Genes that have evolved to cope with inhospitable environments can, through gene splicing and transferring techniques, be employed in new and unforeseen ways to improve human welfare and generate profit (Riggs 1990; Sedjo 1992). As biotechnology develops and genetic engineering expertise grows, the scope for utilizing the diversity of genetic information embodied in a variety of wild organisms will relatedly increase. The benefits of sustaining a rich representation of natural systems are immense since natural genetic information and the new biological technologies will likely complement each other in  37 large numbers of economic activities (Hoffmann 1991; Sedjo 1992). As the WCED (1987, 156) points out, genetic engineering "must be based on existing genetic material and makes such material even more valuable and useful." A biologically diverse community of plant and animal life is also important for providing habitat for crop pollinating insects and predators and parasites of crop and forest pests (Allen 1974; Beattie 1991). Additionally, its preservation helps maintain the productivity of plantations and managed forests by ensuring a genetically diverse seed supply for reforestation efforts (Perry and Maghembe 1989). These and other direct or indirect economic goods and services arising from wild genetic resources provide humans with trillions of dollars in products. Genetically improved cultivars also reduce the losses associated with pests, diseases and environmental extremes and reduce the agricultural industry's dependence on increasingly toxic and costly chemicals. Wild genetic resources also represent untapped evolutionary potential (Riggs 1990). With only 1.4 million species taxonomically described out of an estimated 30 million or more, and with the genetic diversity contained within these species remaining almost entirely unstudied, the conceivable discoveries of new uses appear almost limitless (Scace and Nelson 1985; Wilson 1985, 1988; Ehrlich and Wilson 1991). Although much of this diversity exists in tropical rainforests, B.C. still has an enormous variety of wild species and subspecies, many of which are restricted largely to B.C. Because it contains more microhabitats than anywhere else in Canada, B.C. is the most biologically diverse province in the country (Bunnell and Williams 1980a; Hume 1993). In addition, the temperate forests of B.C. and the U.S. Pacific Northwest harbor some of the most genetically diverse conifer species on earth (Norse 1990; Pearson 1991). As this section has revealed, this diversity embodies hidden economic value in the form of a variety of exploitable genetic resources. Such resources could be unscreened medicinal compounds, overlooked domesticates or uninvestigated genes with utilizable genetic traits. The genetic material from all wild sources will be sure to appreciate in value as new uses are found  38 for their encoded genetic information (Wolf 1985). Maintaining a diversity of wild gene pool resources is therefore a means of expanding possible future uses and applications (Ewert 1990b), and as such, should be considered an investment in the future (Sedjo 1992).  Scientific Information for Resource Management Wilderness can act as natural laboratories for researchers in such scientific disciplines as ecology, geology, botany and zoology. Research scientists can and do use the data extracted from undisturbed ecosystems and the species they shelter to develop scientific knowledge and theory (Manning 1989b). There is increasing recognition of the importance of retaining natural ecosystems of all ages for both pure and applied research (Hansen et al. 1991). Ecological studies that employ applied research methods are of particular interest here due to their economic spin-off values to society. However, it must be recognized that because the future is uncertain in terms of societal tastes and needs, even pure research may yield products of eventual significant commercial value, as formerly "useless" species and processes are applied in new and unforeseen ways. Because of this, both the direct economic benefits of applied research and the potential applications of basic research will be interwoven in this discussion. Most of the scientific values of wilderness in which resource production maintenance or enhancement is the primary management goal fall within the general category of baseline studies (Cutler 1980; Mack et al. 1983; Stankey 1985). Aldo Leopold first recognized the importance of retaining wilderness as a reference point from which to assess modified environments when he wrote "A science of land health needs, first of all, a base datum of normality, a picture of how healthy land maintains itself as an organism" (Leopold 1949, 274). Jenkins and Bedford (1973, 169) define baseline as "An accurate description of the status and workings of an ecosystem in the absence of artificial human disruptions." This deliberately general definition provides enough conceptual latitude to allow it to be used as an organizing principle in a range of disciplines. The relevancy of environmental baselines (or benchmarks) relates to the notion that measurements made and  39 studies conducted in wilderness help to establish a body of information which can be compared to conditions existing in managed environments. Such comparative data provide a basis for formulating better management or mitigation strategies (Jenkins and Bedford 1973; Parsons and Graber 1990). Applied forestry is particularly reliant on benchmarks provided by wilderness. Wilderness areas can provide representative levels of ecological baseline data if they are carefully chosen to represent different stages of forest development, landscape positions and soil types (Burton et al. 1992). The traditional level of research concern for forest managers in B.C. is at the species level. Plant species which are adapted to a restricted range of site conditions are often used as sensitive indicators of variations in site factors (Smith 1986). The presence, vigour, abundance and distributional pattern of such indicator species can reveal explicit ecological conditions of the site, including its soil moisture and nutrient status and micro-climatic condition (shade, aspect, etc.). As a result, the use of indicator plant species is central to forest land management and planning in B.C. (Green et al. 1984; Burton et al. 1992). They are used in ecosystem classification, which in turn forms the foundation for pre-harvest silvicultural prescriptions (Burton et al. 1992). Because most indicator species by nature are specialists and more reliant on the ecological conditions existing in untrammeled stands, preserved wilderness provides the basis for indicator species analysis and ecological site description (Pielou 1990). These parameters then serve forest managers by improving their abilities to monitor the effects of current practices and evaluate future management options (Burton et al. 1992). The functional, structural and compositional characteristics of protected ecosystems can also serve as baselines. By examining ecological processes and functional relationships in their natural states, land managers can gain improved insight into the workings of natural systems and through emulation improve their abilities of managing or restoring altered areas (Franklin 1981; Dixon and Sherman 1990). Examples of ecological parameters which are studied in the  40 field or modelled by computer simulations include biodiversity, succession, speciation, biomass, productivity, population levels, physical attributes and community structure, organization and dynamics (Jenkins and Bedford 1973; Mack et al. 1983; Pearsall 1984; B.C. Ministry of Forests 1992). Researched processes include watershed function, material cycling and energy and material use, exchange and loss (Jenkins and Bedford 1973). Such ecological information provide resource managers in general and forest managers in particular with invaluable comparative data which can be directly applied toward the development of management prescriptions that more closely resemble natural processes and ecological characteristics. "New Forestry" is an example of a sustainable ecosystem management strategy which incorporates stand and landscape level ecological research findings into its underlying principles. Developed by Jerry Franklin and his research team at the H.J. Andrews Experimental Forest in Oregon, the underlying premise of New Forestry is that before ecological and timber resource values can be sustained, managed forests must be maintained as complex ecosystems rather than simplified monocultures (Franklin 1989b, 1990a). Through their research on old-growth forests in the Pacific Northwest, the Andrews Group, among others, have concluded that the main objective of professional foresters when designing management prescriptions should be to insure that a viable level of old-growth forest ecosystem elements are perpetuated. The rationale for retaining old-growth characteristics is to maintain or rapidly redevelop complex forest ecosystems with a compositional, structural and functional level of diversity absent in monocultural plantations (Franklin 1989a). Franklin and others argue that the biological diversity passed on to the young stand through the retention of such "biological legacies" as coarse woody debris and snags is important for maintaining ecosystem resilience (the ability of a system to retain the integrity of its basic relationships and thereby resist the change associated with disturbance), which along with soil is the ultimate determinant of long-term site productivity (Maser 1988; Maser and Trappe 1984; Franklin et al.  41 1989; Franklin 1990b). In order for forests to be managed in a sustainable manner which preserves natural species and structural diversity as well as facilitates ecosystem recovery, New Foresters advocate harvesting regimes which more closely resemble natural disturbances such as windthrow, wildfire or insect outbreak (Franklin and Waring 1979; Spies, Franklin and Thomas 1988). At the stand level, this entails the retention of large rotting logs on the ground, the maintenance of snags and the retention of a minimum number of live green trees (Franklin et al. 1981; Hopwood 1991). At the landscape level, this entails harvesting patterns that mimic the conditions of natural landscapes, which tend to contain a heterogeneous mosaic of stands of different sizes, shapes, types, conditions and successional stages (chronosequences) due to unevenly exerted disturbance histories (Harris 1984; Franklin 1989b; Foss 1990). These and other strategies under the guise of New Forestry are being devised in an attempt to manage for functional stands and landscapes that maintain a more long-term productive capacity, sustain a more natural level of biodiversity and harbor a greater range of values. The above account reveals that the foremost scientific value of wilderness, as it pertains to resource production, are as models of sustainable productivity and concomitant baselines against which alternative silvicultural systems and management practices are assessed (Norse 1990). Other baseline studies also give direction to resource managers. For example, untouched whole drainages provide benchmark data on fisheries levels and health, natural rates of erosion and water quality and quantity to forest managers studying the impacts of logging and road construction activities on such attributes in managed watersheds (Franklin 1981). Improved ecological understanding of the relationship between native forest structure and vertebrate community composition has allowed researchers to discern to a greater degree which ,  structural elements of the plant community fulfill wildlife's variable habitat requirements, including microclimate, food and sheltering needs (Hansen et al. 1991). Studies of natural populations of larger herbivores and predators, along with their interactions, require large tracts of wildlands in particular because of the large migratory ranges normally involved (Franklin  42 1981, 1985). Such knowledge, as it more fully develops, will give forest managers greater guidance when designing species-sensitive harvesting plans, at both the stand and landscape levels. Unaltered ecosystems also serve as both a working model and a source of living components for habitat reconstruction efforts (Ehrenfeld 1976). Lastly, besides acting as reference sites for ecosystem analyses of silvicultural and management practices, wilderness and its complement of wild communities serve the interests of society by acting as loci for the monitoring of background levels of environmental pollutants. Because many wild species are sensitive to changes in environmental conditions and internalize pollutants readily, indicator species can be used to define normal or baseline environmental conditions. This is accomplished by assessing their population numbers and fluctuations and their health and by measuring toxicity levels in their tissues and organs (Ehrenfeld 1976; Bunnell and Williams 1980b; Root 1990). Direct surveillance of environmental parameters is also done by abiotic chemical analyses. Sampling programs have provided baseline data on pollution levels of air, water, soils and plants. Repeated samplings have enabled scientists to determine trends in the background levels of common pollutants (Cutler 1980; Franklin 1981; Mack et al. 1983). With much of the monitoring focussing in on particular problems, the primary purpose of environmental quality studies in ecosystems is to "act as a sentinel to warn of critical changes while there is time to head them off" (Jenkins and Bedford 1973, 170). The need to detect potentially dangerous changes and threshold effects in the biosphere is critically important in the modern age because of the potentially adverse global impacts of greenhouse gas and ozone-depleting chemical emissions. Scientific values can create effective arguments for setting aside wilderness areas as natural research facilities. As such, wilderness provides opportunities to acquire the baseline data necessary for understanding such complex issues as global warming, succession, nutrient cycling, pathogen levels, species evolution, etcetera (Ewert 1990b; Burton et al. 1992). Much of the ecological data is being collected in an effort to improve management practices. New  43 Forestry is an example of this. Others (such as pollution monitoring) are designed to detect environmental changes and correspondingly change behavior in order to put society on a more sustainable footing. All baseline informational values emanating from wilderness allow humans to better manage their civilized world by monitoring changes in or copying the features and processes of the natural world.  Environmental Information for Media Individuals can gain in abstentia benefits from wilderness as a result of someone else's visit. Because off-site users can attain the benefits vicariously through on-site users, these values are known as "vicarious values" (Pearsall 1984). Wilderness areas provide humans with substantial vicarious values by serving as the subject matter, inspiration and setting for artists, writers and film-makers. People extract educational, aesthetic and recreational value from wilderness through the reading of wilderness books, poems, magazines and articles, the watching of television specials devoted to wildlands and their plants and animals and the viewing of wilderness-inspired artwork (Pearsall 1984; Driver, Nash and Haas 1985). Hendee, Stankey and Lucas (1978), while conceding the difficulty of quantifying such values, illustrate their significance in the following manner: Any estimate of the amount of indirect use clearly would be difficult to make and none is available. Sales of wilderness-related books are substantial, and films with naturewilderness themes draw crowds. Several popular TV programmes are based on nature and wilderness. We would speculate that the number of vicarious and other indirect users is probably greater than the number of actual visitors. The public has exhibited an increase in its personal and economic expressions of preference for such themes in recent years by showing a greater interest in the wildernessbased topics covered in the media and by purchasing at burgeoning levels wilderness literature (Driver, Nash and Haas 1985; McNeely et al. 1990b). The media has responded to this increase in demand by placing greater emphasis on the natural environment as a subject matter of general and educational interest in both print and electronic mediums (McDaniels 1992). As  44 a result, wilderness as a topic has shown high commercial value for use in magazines, papers, books, films and mass media television, and as scenic backgrounds for drama, commercials and other promotions (Cordell and Reed 1990). Tourism Opportunities Numerous phrases are used to describe wilderness-based tourism; among them are ecotourism, green tourism, nature tourism, adventure tourism, wildlife tourism, educational tourism and scientific tourism (Haas 1990; CEAC 1991). Although each has slightly different connotations depending on the values sought, all share a reliance on wilderness as a tourism draw. Difficulties arise, however, when attempts are made to estimate the proportion of tourism values directly attributable to wilderness. Statistics on tourism in B.C. are primarily gathered by user surveys of resident and non-resident travellers. Typical measures of tourism values such as visitor days, expenditures and package tour prices are usually unrefined and consist mostly of raw data unattached to specific activities (Woodbay 1989; B.C. Ministry of Forests 1991b). Data deficiencies and other limitations must therefore be kept in mind when viewing the following estimates of expenditures, employment, participation levels and trends. Although the numbers are crude and do not fit neatly within the wilderness-based tourism category, they do provide insight into the importance of wilderness as a focal point for attracting tourism activity. Adventure travel, which is defined as "a leisure activity that takes place in an unusual, exotic, remote, or wilderness destination and tends to be associated with high levels of involvement and activity by participants, most of it outdoors", was worth approximately $55.7 million and supported 1590 jobs in 1986 (Outdoor Recreation Council 1988, 1). If supplierrelated and consumer impacts are taken into account, the Gross Domestic Product and wage income generated from adventure travel is about $70 million and $47 million respectively and provides employment to 2810 service workers (Woodbay 1989). The activities included in this estimate are backpacking, heli-hiking, trail riding, sailing, cross country skiing, hiking,  45 canoeing, river rafting, bicycling, ski touring, mountaineering, kayaking, nature observation, SCUBA diving and heli-skiing. Total direct expenditures climb to $133.7 million when hunting and sports fishing are included (Woodbay 1989). However, it must be noted that these figures understate the significance of adventure travel because they do not include all of the adventure tour operators and do not include individuals who do not purchase adventure travel packages (Outdoor Recreation Council 1988). Being the fastest growing tourism sector in B.C. (MacLaren and Ethos 1988), the adventure tourism industry is projected to generate $500 million in economic activity by the year 2000 (Outdoor Recreation Council 1988; Natural Resources Management Program 1990). B.C. has a relative advantage over most regions in North America due to its ecosystem diversity (Thompson 1987), and resultantly gains more economic benefits from adventure tourism than any other North American jurisdiction (Outdoor Recreation Council 1988; Natural Resources Management Program 1990). B.C. parks attract high numbers of visitors. In 1989, B.C. Parks recorded over 21 million visits, which contributed $400 million to the provincial economy. Although visitors do not spend all of their time and money within a particular park, it is usually the park that defines a region and attracts the visitors initially (B.C. Ministry of Parks 1990a). Moreover, growth in the level of visitation to B.C. parks reveals their increasing importance to residents and nonresidents alike; since 1991, the rate of visitation has increased by 23 percent (Hume 1993). Altogether, British Columbians alone spend about $615 million (1985 dollars) annually on wildlife recreation, which includes such items as transportation, food, accommodation and equipment. Approximately 85 percent of these expenditures are for non-hunting activities. These expenditures support 19,000 man-years of employment, generate wages and salaries totalling nearly $240 million, produce $360 million in GDP and yield $123 million in taxes and other government revenue (B.C. Ministry of Environment and Parks 1987). Future development of all segments of the wilderness-based tourism industry require preserved wilderness areas to supply scenic vistas, wildlife viewing opportunities, along with  46 the many other natural values which attract tourists (Natural Resources Management Program 1990). Retaining wilderness is particularly important when one considers that many tourists are attracted to the natural themes promoted through the "Super Natural British Columbia" marketing campaign, regardless of whether or not they plan to actually visit any natural areas. Wilderness, therefore, indirectly benefits all segments of the tourism industry, an industry which generated $3.5 billion in revenues and supported 83,700 direct jobs in 1987 (Woodbay 1989). Non-Market Based Human Demand Values Non-market values are not generally bought or sold in competitive markets. Lacking readily available market prices, these values must be measured by using non-market valuation techniques which elicit individual's preferences in some other way. Environmental economists and others have adopted procedures like contingent valuation, the travel cost approach, hedonic pricing, among others to assess such non-market values as those outlined in this section (see Chapters IV and V for methods). Outdoor recreation, aesthetic appeal, existence/option and cultural/historical are considered human demand-related because they all service some felt human preference (Norton 1987). Although distinctions have been made between these changeable and culturally or psychologically determined human preferences ("amenity values") and genetically or physiologically fixed values which fulfill "essential" human needs (Norton 1987, 98), amenity values remain powerful justifications for preserving wilderness. Outdoor Recreation Wilderness recreation, though not needed for life-supportive or economic reasons (though it has significant commercial dimensions as we have already seen), satisfies the well being of individuals participating in recreational activities in other ways. Rolston (1985, 1988) identifies beneficial use and pleasurable appreciation as the two primary recreational values stemming from wild lands. The first relates to the physical and social rewards which such goal-directed activities as  47 camping, fishing, hiking, mountain climbing, horseback riding, cross country skiing, swimming and kayaking instil in individuals or groups (Brockman and Merriam 1979). Achievement (e.g., practicing, challenging and enhancing skills, seeking excitement and adventure, testing endurance and stamina, feeling a sense of conquest), physical fitness, risktaking, self sufficiency and the building of social ties are examples of motives which are satisfied by using wilderness as a physical or social setting (Brown and Haas 1980; Driver, Nash and Haas 1985; Merigliano 1989). For this first set of participatory motives, wilderness provides an environment in which people can "show what they can do" (Rolston 1988, 7). The second set of motives, on the other hand, relate to the values which are reached as people are "let in on natures show" (Rolston 1988, 7). The focus here is on the intellectual, psychological, spiritual and cultural values which are achieved through a contemplative, introspective and reflective relationship with nature (Brockman and Merriam 1979; Manning  1989a). While such pursuits as nature study, photography, painting, archeological/historical research, meditation, aesthetic contemplation among others may also be physically rewarding, their main purpose is to enhance the inner perceptual and/or spiritual life of the people achieving closeness with wilderness (Brockman and Merriam 1979). The activities exemplified in the physical group of motives also provide emotional, cognitive and spiritual side benefits. Wilderness provides a particularly conducive experiential setting for interacting closely with nature and collaterally enhancing one's personal development. Unlike developed recreational areas, wilderness by definition is primitive and unconfined, allowing solitary retreat and escape from an increasingly fast-paced, crowded and stressful society (Reed and Merigliano 1990). Solitude in this context refers to the opportunities that wilderness provides "for temporary release from the rules and pressures of everyday life. Solitude in this situation refers to escape and isolation from certain social structures and environments, rather than isolation from individual people" (Hammitt 1982, 480). Solitude and privacy enable individuals and small, like-minded social groups to avoid societal intrusions and experience  48 wilderness in an unconstrained, free manner, where personal autonomy, emotional release, self-evaluation and inherent alertness are heightened and nurtured (Hammitt 1982). With wilderness affording a rich array of opportunities for enhancing one's perceptions of the environment and oneself, the literature is replete with examples of the many psychological and spiritual benefits arising from wilderness recreational activities and contemplative pursuits. Drawing from studies on the impacts on personal values from wilderness experiences gained through such programs as the Outdoor Challenge Program, researchers have identified numerous interrelated positive perceptual responses to wilderness, including (1) feeling at one with or a part of the environment (Rossman and Ulehla 1977; Talbot and Kaplan 1986; Taylor 1990); (2) feeling awe and wonder from the mysteries of nature (Kaplan and Talbot 1983; Kaplan 1984b; Talbot and Kaplan 1986); (3) experiencing, learning about, appreciating and observing the beauty and complexity of nature (Brown and Haas 1980; Kaplan and Talbot 1983; Merigliano 1989); (4) experiencing tranquility, peace and spiritual rejuvenation (Merigliano 1989; Taylor 1990); (5) expressing, actualizing and affirming one's self-identity and improving one's self-concept through reflection on personal values (Brown and Haas 1980; Hendee and Brown 1988; Williams, Haggard and Schreyer 1989; Driver, Tmsley and Manfredo 1991); (6) relieving stress and feeling mentally, emotionally and physically renewed, relaxed, refreshed and restored (Scott 1974; Talbot and Kaplan 1986; Henning 1987; Hartig, Mang and Evans 1990; Driver, Tinsley and Manfredo 1991); and (7) the therapeutic/healing values for individuals with psychological, social or physiological disorders (Bernstein 1972; Gibson 1979; Levitt 1989, 1990; Ewert 1990a; Greenway 1990). Both motives collectively provide an overriding view of the benefits of recreation, which can be defined as a "refreshment of one's mind and body after labor through a diverting activity" (Reed and Merigliano 1990, 96, quoting the 1978 American Heritage Dictionary). Wilderness harbors a range of biological, physical and cultural features and conditions which bestow value to an area for consumptive and non-consumptive recreational purposes. As a result, it offers an exceptionally rich setting for individuals and social groups searching for  49 physically, intellectually, emotionally, culturally, socially or spiritually enriching outdoor recreational experiences. Aesthetic Recreational experiences in wilderness settings, as mentioned in the preceding section, often involve positive perceptual responses on the part of the recreational participant. One of the most commonly cited perceptual experiences which motivate many recreationists to seek close contact with wilderness is the aesthetic enjoyment derived from such a direct encounter. Aesthetic means "to perceive by the senses" (Bunnell and Williams 1980b; Bunnell 1990). As such, aesthetic enjoyment of wilderness involves the satisfaction or pleasure derived from the direct sensory experience of wilderness systems and their complement of wild species and physical features. This conceptualization of aesthetic value goes beyond the traditional view of aesthetics as scenic beauty by embracing nature's impacts on all of our sensory faculties, including thought, visual, auditory, olfactory and tactical, as well as the lingering, residual effects of memory (Pearsall 1984; B.C. Ministry of Forests 1991b). At a deeper level, natural aesthetic value also includes the transformative affect of nature's sublimity (awe-inspiring grandeur, majesty, richness, vastness, etc.) on the cognitions, emotions and spirit of people aesthetically immersed in a wilderness experience (Driver, Nash and Haas 1985; Vest 1987; Manning 1989a, 1989b; Driver, Easley and Passineau 1990). Natural settings and wild species are often likened to great "works of art" that can instil pleasure or satisfaction in the valuer (Bunnell 1990). Wilderness has in fact served as a source of creative inspiration to those contributing to art, literature and music and has influenced cultural development in the process (McCloskey 1989, 1990) Like art, ecological systems and the species they contain are often perceived to be more aesthetically appealing when they are rare or unique (Kellert 1986; Sober 1986). Moreover, the magnitude of aesthetic preferences for and pleasures derived from natural and man-made art alike depend to a relative degree on the culturally-determined subjective perceptions of the observer (Norton 1987).13  50 However, aesthetic experiences of wilderness differ from aesthetic experiences of art objects in a number of important ways. Firstly, wilderness normally surrounds an observer within a totally encompassing frame of experience, where all senses are involved in apprehending the total environment in which the observer is immersed. The aesthetic awareness of art objects, on the other hand, often involves a more detached, sensorialyrestricted view (Hepburn 1968). Secondly, unlike art objects which are constrained within physical and temporal reference frames, wilderness is frameless in both time and space. Wilderness is aesthetically unique in this regard in that it is detached from all temporal relationships (not rooted in any one period of human history) and is dynamic in character and thus always changing. This provides a wider and richer scope for perceptual appreciation (Hepburn 1968; Manning 1989a, 1989b). Thirdly, wilderness represents reality firsthand. Because nature is not filtered through an artist's imagination, skills or tools, an observer of wilderness can directly encounter and admire its beauty without wondering if he interpreted the artist's intentions correctly. This allows for a freer form of interpretation and a greater scope for individual imagination (Hepburn 1968; Rolston 1991). Lastly, as earlier stated, the sublimity of wilderness endows it with a transformative power not commonly associated with human artefacts. In its deeper, dynamic dimensions, the aesthetic experience can act to alter emotional reactions to objects and affects felt and expressed values and preferences in the process (Norton 1987; Rolston 1988). Wilderness thus possesses unique aesthetic characteristics which can gratify all of the senses and can inspire awe in ways which cannot be duplicated by the works of humans (Flamm 1989). Existence I Option Some values connected to wilderness are identified as being independent of present use. Unlike economic, recreation and aesthetic values which are expressed either through 13 However, the cultural influences on aesthetic preference for natural environments are found to be less pronounced than previously thought. Evidence from aesthetic surveys suggests that high regard for the aesthetic appeal of nature is consistent across cultures (Ulrich 1983).  51 consumptive or non-consumptive use of the wilderness resource, existence and option values are expressed without its current use. These "non-use" or "preservation" values instead reflect the values that people place on intact wilderness for the purpose of simply knowing that it exists or retaining the possibility of visiting it or exercising other options in the future. Because wilderness resource economics has traditionally focussed on the costs and benefits of recreation use, these non-use demands for wilderness have normally been excluded from conventional benefit-cost analyses (Walsh, Gillman and Loomis 1982; Walsh, Loomis and Gillman 1984). As a result, conventional site valuation studies capture only the behavior of people visiting a site and miss the social benefits of protecting a natural area from irreversible loss, seriously underestimating demand (Brown 1985; Walsh and Loomis 1989). Recognizing this omission, the environmental economic literature has since fleshed out more fully these concepts and has devised alternative approaches for incorporating them into more fully developed economic frameworks. These total economic valuation exercises share the goal of attempting to include both the preservation values of option and existence demands and the use values of recreation demands among others within their analytical structures (Walsh, Gillman and Loomis 1982; Randall 1987b; Swanson and Peterson 1988). Although inclusion of preservation values has met with variable success, existence and option values remain important values to consider in land use decisions. Existence  Existence value is derived from the knowledge that wilderness continues to exist and, as such, has been defined generally by Bishop (1978, 15) as "the utility that people receive from simply knowing that something exists." In economic terms, it can be defined as "the willingness to pay (or to accept compensation) for the existence of a resource without the prospect of using the resource" (McConnell 1983, 255). Recognition of the importance of existence value occurred first to Krutilla (1967), who noticed that people, through their subscriptions to the World Wildlife Fund among other organizations, were gaining satisfaction from the knowledge that parts of North America were  52 remaining as wilderness (Bennett 1984). Others also pointed out that people need never visit an area or come in contact with a species directly in order to ascribe value to it. For example, Barkley and Seckler (1972, 131) illustrated existence demand for blue whales when they mentioned that "...countless millions of people derive pleasure from simply knowing that the whale exists." Equivalently, "Consider wild polar bears for example. Few people ever expect to see a wild polar bear or to otherwise use one, yet many apparently value the fact that they exist" (Talhelm 1983, 276). Countless other examples of and empirical evidence for the expression of existence value for wilderness and wild species can be found in the literature (see Brookshire, Eubanks and Sorg 1986, 1987). The utility of the existence (or disutility of disappearance or extinction) of wilderness and species provides the source of existence value for individuals aware of such areas and organisms (Randall 1987a). Since existence values are independent of present use, expected future use and the avoidance of risks associated with future use, authors such as Randall (1986, 1987a) and Sen (1977) have suggested that they must stem from a motivation of altruism, caring or sympathy. Several motivations have been suggested as components of existence value, including vicarious consumption, bequest, stewardship and the intrinsic characteristics of the valued entity. Existence value as represented by vicarious consumption relates to the notion that individuals experience philanthropically-driven gratification from, and thus should be willing to pay for, the knowledge that others are using wilderness (Pearsall 1984; Brookshire, Eubanks and Sorg 1986, 1987). Non-use values may also be motivated from a desire to allow future generations to use the wilderness resource. Satisfaction is thus gained through the bequest value of the natural area (Krutilla 1967; Randall 1986, 1987a, 1987b). Moreover, people may value the existence of wilderness for stewardship purposes; this relates to the idea that some individuals assign personal value to the knowledge that the ecological integrity and diversity of natural systems are being cared for (Brown 1985; Brookshire, Eubanks and Sorg 1987). Lastly, preserved wilderness represents for some a  53 refuge for species which are intrinsically valuable. Satisfaction here arises from the contented belief that wild species themselves are benefitting from being undisturbed (Randall and Stoll 1983).  Option Uncertainty about the future is implicit in many arguments stressing the importance of wilderness preservation (Bunnell and Williams 1980b). Preferences, climatic changes, resource supply levels, new uses from unstudied species/chemicals etcetera are examples of unknown future decision variables which collectively imbue unpredictability in biological and socioeconomic outcomes and make resource demand assessments risky. Many have argued that since environmental decisions are uncertain and possibly irreversible (e.g., ecosystem/species losses, which cannot be replicated), a value can be placed on the preservation of natural areas for the purpose of exercising potential future uses (Fisher and Hanemann 1985). Wilderness advocates view such a value as a means of maintaining options and flexibility in the face of unforeseen events. Recognizing the strength of the "hedging our bets" argument, economists have responded by developing the "option value" concept. Option value can be viewed generally as the value of retaining the possibility of using a resource in the future (Talhelm 1983). Under conditions of uncertainty (unknown demand and resource availability in particular), individuals may be willing to pay for the continued existence of a wilderness in order to secure an option to use such an area in the potential event that such a future use as recreation is demanded (Randall and Stoll 1983; Barrick 1985). At a societal level, preserved natural areas may serve a resource-banking function to provide for future generations (Pearsall 1984). Option value is often conceptualized as an insurance (risk aversion) premium, since it reflects the adjustments that should be made to present welfare in response to future uncertainties (Swanson and Peterson 1988). The risk aversion premium notion of option value represents the premium individuals are willing to pay to avoid the risks associated with a reduction in or an elimination of a valued natural good or service (Bishop  54 1982; Barrick 1985). It is thus a "means of assigning value to risk aversion in the face of uncertainty" (McNeely 1988, 22). Another take on option value has since emerged in the literature. Called "quasi-option value", it represents the value of sustaining options under conditions of expanding or deepening knowledge. Randall (1987a, 412) describes it as "...the value of the emerging information conditioned on having made the first period choice (preservation, in this case) that maximizes the second-period array of alternatives." The concept is most often used in reference to irreversible policy decisions which can either be undertaken now or delayed for some period of time. If implementation of the policy is delayed and relevant new information becomes available for use at the beginning of the second time period, then there is value connected with the delay (Swanson and Peterson 1988). An example of quasi-option value is the value of new knowledge gained through the accumulation of research results on pharmaceutically useful compounds. Randall (1986, 84) describes the importance of preserving species which as yet contain no demand value: For any species there is some positive, if unknown, probability that new uses generating positive value will eventually be discovered. With extinction, however, the probability drops to zero, as does the expected value of the resulting benefits. Cultural 1 Historical Intact wilderness can preserve cultural and historic values in two ways. Firstly, its physical existence protects features of historic/cultural significance which lie within its boundaries. Secondly, its continued presence symbolically defines and strengthens individual, group and national identities. Many cultural and historic features existing in natural areas have heritage value. The B.C. Ministry of Parks (1990b) identifies several native Indian cultural features and historic features deserving of protective consideration. Archaeological sites (middens, kickwillee holes, village sites and legend sites (areas of indian religious and spiritual importance)), rock paintings (pictographs), rock carvings (petroglyphs) and archaeological routes (grease trails)  55 are recognized examples of native Indian artefacts and sites which embody cultural value. In B.C., areas which are known to represent examples of these include Adams River, Roderick Haig-Brown Park (kickwillee holes), Ninstints, South Moresby National Park Reserve (village sites), Tahltan Eagle, Spatsizi Plateau Wilderness Park (legend sites) and Petroglyph Park (petroglyphs). Historic features, representing different major eras (such as discovery, early exploration, fur trade period, early gold rush, beginnings of forest, mining, fisheries and agriculture industries and early settlement) have also been determined by the B.C. Ministry of Parks (1990b) as meriting protection. Included here are historic sites (areas linked to early European exploration, settlement and pioneering), historic routes, monuments and historic cultural features (features associated with early industry). B.C. examples include Mackenzie Rock, Sir Alexander Mackenzie Park, Cape Scott Park, Fort Rodd Hill National Park, Newcastle Island Park, Alexandra Bridge Park (historic sites), the Mackenzie Trail (historic route), Ruckle Park, Barkerville Park and Wells Gray Park (historic cultural features). Although a wilderness area may not envelop any of these heritage resources, many do. Such resources when residing in wilderness settings (whether designated or de facto) are not only physically protected from obtrusive forms of development but are also often protected contextually. Cultural or historic features are often fully imbued with character and meaning only when integrated within the larger context of their natural surroundings. The integrity of the relationship between wilderness and the heritage resources it contains is thus an important one to maintain because wilderness contextually defined the nature of such sites and features in the first place by acting as their original source of inspiration (e.g., native artefacts) and by physically imposing constraints on and giving shape to their development (village sites and structures, historic routes and sites, etc.). The values associated with a heritage resource often relate to its ability to act as a tourism draw, an aesthetically appealing recreation destination, a focal point for cultural and spiritual development and an informational source for educators and  56 scientists interested in native and non-native societies.  14  Heritage value is also represented symbolically in wilderness through its defining influence on the identities of cultural groups and nations. Fox (1985, 164) provides context to the importance of the natural heritage value of wilderness to Canada's identity in the following passage: Wilderness is an important feature of Canada's national, as well as natural heritage. It affected the lives and development of the people who created what is Canada today, Indian, Inuit and European settler alike, and influenced our recorded history. A people's heritage is profoundly important. It provides a symbolic rallying point for people's pride and fosters a sense of identity and belonging. Although much of the literature on this subject is devoted to the importance of wilderness to the American identity 15 , wilderness shaped and continues to shape Canada's character as well. For example, many features of old-growth forests, such as large trees, are recognized as being part of the fabric of Canadian and British Columbian identities (B.C. Ministry of Forests 1991b). Wilderness provides a protective refuge for such species as beaver, grizzly bear and moose, which function as symbolic expressions of our national identity. It also inspires artists, poets, musicians, writers and sculptors, which further strengthens Canada's personality and fosters unity among Canadians (CEAC 1991). The cultural identities of North American indigenous peoples are especially integrated with wilderness. As Laidlaw (1991, 12) points out, "While Native (North) American traditional cultures differ substantially in the specific, they share common tenets which hold custom, culture, practice and religion as inexorably tied to relationships with the land." He goes on to say that many native cultures view nature and wild species "as distinct but integrated components of a natural and spiritual system. Each element of this system possesses a spirituality akin to the human soul." Tribal religions make little or no distinction between the 14 See Neumann and Reinburg (1989a, 1989b), Fischer (1989), Hanks (1989), Lipe (1989), and Corrigall, Schneider and Reed (1990) for overviews on cultural resource values and archaeological research in wilderness. 15 See Nash (1969, 1982), Manning (1989a, 1989b), Williams, Haggard and Schreyer (1989) and LePage and Ranney (1990).  57 spirituality of humankind and the spirituality of the natural world which sustains human existence (McDonald 1990; Laidlaw 1991). Moreover, subsistence uses of fish, wildlife and plant species support the physical, social, economic and cultural well-being of many aboriginal cultures. Wilderness, "by maintaining conditions in which resource populations can be naturally sustained, have the potential to serve as a repository of many traditional subsistence values, activities, and lifestyles, as well as preserve cultural diversity within modern societies" (Muth and Glass 1989, 142). Wilderness has recently provided experiential settings for native youth groups learning native customs, religions and traditional land use practices first hand through the teachings and council of elders. "Rediscovery" is an example of a program which runs youth camps for natives and non-natives in wilderness settings. Focusing on personal, cultural and environmental awareness, and drawing on the traditions of North American indigenous cultures, the program has shown great success in instilling pride, confidence and belonging in individuals, and concomitantly strengthening traditional values and cross-cultural understanding (Henley 1989). Defining and reinforcing such native values has been an important trend in recent years. Laidlaw (1991, 13) puts this movement's significance into perspective: "For many native people the traditional values which define their relationship with the land have gained increasing importance as a focal point for maintaining and recovering cultural identity." Wilderness, therefore, helps to define the cultural self, group and nation by acting as an historic symbol, harboring species which are steeped in symbolism, providing a setting in which traditional practices can be carried out and protecting culturally and historically significant artefacts. It also provides a setting in which spiritual values can be encouraged, an overlapping wilderness value which will be covered next.  Human Spiritual Values The previous section has intimated that the basis of many indigenous groups' myths,  58 legends and lifestyles is supported by the reverential or sacred character of wilderness, where spirituality is personified in the land, plants and animals. Native Indian spirituality, along with the transcendental insights of early environmentalists, have helped to lay the philosophical foundation for modern North America's understanding of and concern for the spiritual values of wilderness (Driver, Nash and Haas 1985; McDonald 1990). However, spirituality need not be linked to the religious practices of a specific cultural group to be experienced by the individual. The literature is replete with examples revealing the importance of the spiritual dimension of a wilderness experience to native and non-native wilderness users alike. Norton (1987) uses the term "transformative value" to describe the spiritual aspect of human preferences. Brown (1988, 339) gives introduction to transformative values by offering the following description of them: It represents a complete change of being and a shift to a higher mode of operating. It implies the awakening of new levels of awareness; a fundamental resolution of the internal causes of stress; the discovery and clarification of essential values in the world; and the redirection of life energies toward a higher and more fulfilling purpose. Norton (1987) distinguishes these values from the human demand-related values outlined earlier by stating that such spiritual values are not directly linked to the satisfaction of human wants, but are instead instrumental in transforming human desires through the rebuilding of spiritual health or the reshaping of preferences. These two transformative values of wilderness will be sketched next.  Rebuild Spiritual Health Man may yet restore himself to health if he will learn to understand himself in relation to the world of nature in which he evolved as an integral part, and to appreciate the nature of his relationship to the world of nature. This short account by Montagu (1966, 227) provides insight into the restorative value of a wilderness experience. The ability of wilderness to restore, rebuild or promote the growth of an individual's spiritual health relates back to the discussions on the psychologicallemotional  59 benefits of recreation and aesthetics. As was noted there, intimate contact with nature may catalyze within a person engaged in a wilderness-based activity or aesthetically-based contemplative pursuit certain positive cognitive and emotional responses. For example, the possible emotive benefits of solitude include the enhancement of personal autonomy, emotional release, self-evaluation and inherent alertness. It was also shown that there are mental and emotional stress-relief and therapeutic benefits of wilderness. The qualities of awe, wonder, peace, tranquility, harmony, self-affirmation, among others can be instilled in wilderness users who are directly exposed and open to the sublime nature of natural systems and wild species (McDonald 1990). Evidence suggests that wilderness environments are particularly rich in characteristics that foster such states (Scott 1974; Hammitt 1982; Stankey and Schreyer 1985). The wilderness experience thus has cognitive and affective components that interact and may produce a transcendent or spiritual experience (Stankey and Schreyer 1985; McDonald and Schreyer 1991). Wilderness has the capacity to contribute to the personal rejuvenation of spiritual health by offering an optimal setting in which one can achieve peak experiences of higher states of consciousness or self-awareness (Scott 1974; Young and Crandall 1984). It offers an environment in which one can contemplate and gain insight into the natural creative process of evolution and one's role within it (Berry 1988; McDonald 1990). Quiet introspection and contemplation away from the distracting influences of civilization allow receptive individuals to focus their awareness on the power of wonder and mystery emanating from wilderness and its interdependent relationships and dynamic processes, unleashing an enlightened perceptual perspective (Graber 1976; McDonald 1989). Olson (1966, 218) describes such an enhanced state of awareness as a "sense of cosmic purpose". Enhanced awareness can also orientate people spiritually by illuminating within them the feeling that they are in unity with or related to this untouched natural environment and its rich, untainted flow of evolutionary history (Montagu 1966; Rossman and Ulehla 1977; Talbot and Kaplan 1986; Taylor 1990). Spiritual connectedness to and interaction with the natural world in turn inspires a deep sense of tranquility and peace in many wilderness participants,  60 restoring spiritual health further (Graber 1976; Kaplan and Talbot 1983). The heightened mental acuity and renewed spiritual health which is specifically linked to a wilderness experience provide an underlying foundation to the general concept of spirituality, which McDonald (1989, 19) defines as "a process of becoming aware of the life present in all forms." Driver, Nash and Haas (1985, 302) summarize succinctly the importance of wilderness for promoting, cleansing and restoring spiritual health by describing wilderness as a "setting for answering the deepest questions of human existence, for celebrating the creative power behind life and things, and for understanding the unity of them all." Reed and Merigliano (1990) assert that the importance of wilderness as an untouched environment which permits humans to regain perspective and sense of place is increasing as the civilized world becomes more hectic and stressful. Wilderness for some people, groups, cultures or societies are so significant as settings for spiritual improvement that they consider them sacred (Graber 1976; McDonald, Guldin and Wetherhill 1989). The transformative value of wilderness also has the capacity to reshape the preferences of people by influencing what they know and believe to be important (McDaniels 1992). The preference reshaping function of wilderness will be described next.  Reshape Preferences Such transcendentalists as Ralph Waldo Emerson and Henry David Thoreau recognized that the experience of wilderness both elevates human values and informs humans of their humble place in the cosmos. 16 They had a concern, which is shared by present-day environmentalists, that many human values are overly consumptive and materialistic and, therefore, unworthy of satisfaction. Their common belief is that these values are maladaptive  and can be recognized as such while experiencing wilderness. The wilderness experience allows one to contemplate one's place in the natural system and its encompassing great expanses of space and time, while simultaneously permitting one to shun values which are in 16 Thoreau (1962) and Emerson (1982) are good sources for gaining insight into the environmental philosophies of transcendentalists.  61 disharmony with these "higher thoughts" and incorporate or elevate harmonious values. Out of this reflective soul searching emerges a new sense of value (Norton 1987). Wilderness thus provides an inspirational and experiential setting in which to examine and reshape felt preferences as opposed to just satisfying them. This allows for the expression and elevation of superior preferences and their related value systems and the rejection and curtailment of inferior ones. Norton (1987, 189) stresses the transformative value of wilderness and its strength in reciprocally promoting the interests of those supporting and being supported by preservation in the following quotation: A value system that includes transformative as well as demand values can limit and sort demand values according to their legitimacy within a rational world view....Experience of nature can promote questioning and rejection of overly materialistic and consumptive felt preferences. Appeals to the transformative value of wild species and undisturbed ecosystems thereby provide the means to criticize and limit demand values that threaten to destroy those species and ecosystems while at the same time introducing an important value that humans should place upon them. The natural environment in this context is used as a source of inspiration for the development of human and environmental ethics, which helps lead individuals toward more austere, environmentally-sensitive modes of living (Douglas 1983). If such a vision or set of values are supported and adopted by the larger society through their incorporation into public policy, many would argue that the interests of society itself are supported and improved. It has been asserted that societies which protect natural systems and the interests of non-human species are ones which carry the attitude of respect for nature into human affairs. A changed consciousness in an individual's relationship with nature can presumably through extension improve the welfare of society by infusing such attitudes as respect, nurturance, caring, concern, understanding, among others into the relationships of people (Nash 1982; Berry 1988). Understanding the importance of wilderness and society's dependency on it also instills humility into the ways in which resources are used and managed, enabling stewardship practices, development activities and the economic system to be redirected onto a more  62 sustainable pathway (Brown 1981; Daly and Cobb 1989; Milbrath 1989). Wilderness may also generate spin-off benefits for society by offering character-building recreation or aesthetic pursuits, thereby producing more emotionally, physically, intellectually and spiritually satisfied, and thus productive, citizens (Driver, Nash and Haas 1985).  Ecological Values The discussion so far has concentrated on the instrumental values of wilderness in servicing or transforming the preferences of humans. Wilderness also fosters a number of important ecological values which relate to their functioning as integral ecosystems. An ecosystem can be generally thought of as a biotic community made up of interdependent populations interacting with their physical environment (Tansley 1935; Dasmann 1972). It is above all a complex system which has both structure and function. The multitude of organisms, along with the abiotic environment with and within which they interact, make up the structural component. The functions of ecosystems are embodied in the ecological processes (e.g., photosynthesis, herbivory, carnivory, decomposition) that link and sustain the various interrelated structural components via the unidirectional flow of energy and the cyclical movement of nutrients and matter (Smith 1980; Ehrlich and Ehrlich 1981; Lee 1989). Ecosystems also function over time, as such evolutionary processes as succession, disturbances and speciation exert changes to the physical organization, nature or pattern of natural systems and their biotic components (Zimmermann and Noss 1990). The spatial dimensions of such systems pertain largely to the artificial boundaries which humans place around what are perceived to be homogenous functional units. Scales can range from narrowly limited small ponds to the entire biosphere (Davis 1989; Noss 1990). Terrestrial ecosystems and their association of biotic communities are normally defined by their vegetation according to some standard of homogeneity and based on dominant and/or characteristic plant species. Animal communities often are associated with specific plant communities, although habitat structure is commonly thought to be more meaningful than floristics (Noss 1990).  63 Many contemporary ecologists conceive ecosystems as having permeable and gradational boundaries, where transfers of energy and materials between ecosystems occur at various scales (Bailey 1982; Dawson and Davis 1989). Such a conception often includes the idea of hierarchical ecological units of various scales, with smaller systems nesting within larger units along a blurred continuum of ecosystems (Bailey 1982; Davis 1989; Dawson and Davis 1989). Ecosystems themselves can be thought of as one continuum embedded within a biological hierarchy encompassing various levels of complexity, from organic molecules to the planet Earth. Noss and Harris (1986) describe these levels, including the ecosystem level, as holons. Each holon is an autonomous composite of lower-order component parts that in turn form an interactively dependent part of a larger system. This interconnected hierarchy of assemblages suggests a degree of dependency between levels which has implications to the system as a whole when human-caused interference occur. As Noss and Harris (1986, 301) illustrate, "The hierarchy concept implies that components at all scales move through the spacetime mosaic that characterizes our world, and disruption at one level can cascade through any or all of the others." The holistic representation of ecosystems is brought up here simply to point out that such systems are largely simplified scientific artefacts designed to organize and facilitate human understanding. Also, the view that ecosystems are an integral thread of a larger life support system that sustains all life, including human, highlights their importance and fragility. The life support value of ecosystems is a salient and pervasive quality, yet because of its intangibility and tenuous link to human preferences remains often overlooked. Although they are largely  unquantifiable and indirectly attached to human preferences, the dependency of life on the ecological values of wilderness make them an important one to emphasize in this discussion. The free goods and services that ecosystems supply to humans can only be assured if the integrity of a critical level of systems and their functional relationships are maintained (Farnworth et al. 1981). The entire discussion on the ecological values of wilderness ties back to this requirement  64 The ecological values (often referred to as "ecosystem services") stemming from wilderness included here are the recolonization of damaged areas, the intensification of diversity, the contributory support to other species and the circulation/cleansing of air and water.  Recolonize Damaged Areas Natural disturbances play pivotal roles in the evolutionary development of ecological systems. Disturbances occur at a multitude of spatial and temporal scales, from frequent but spatially limited canopy gaps caused by tree falls to infrequent but potentially extensive and/or intensive wildfires. Natural disturbances at any scale revitalize forest ecosystems by breaking the dominance of established species or individuals, procuring a flush of resources such as sunlight, moisture or nutrients, and promoting growth and regeneration of new individuals (Noss 1990). They play a necessary part of an ecosystem's function by creating conditions which foster the processes of renewal (Solbrig 1991). Natural landscapes tend to be dominated by patches in various stages of recovery from disturbance. Because disturbances exert their influences unevenly across the landscape in both space and time, the landscape can be viewed as a "space-time mosaic" (Watt 1947) or "shifting-mosaic steady state" (Bormann and Likens 1979). Unmodified forested landscapes tend to be dominated by a heterogeneous mosaic of stands of different sizes, shapes, types, conditions and successional stages (Harris 1984; Franklin 1989b; Foss 1990). These spacetime mosaic patterns imply that at a landscape scale species composition remains in an approximate steady state despite drastic local fluctuations (Noss and Harris 1986). The degenerative properties of disturbance are thus counterbalanced at most times by the regenerational processes of succession, unless irreversibly modified by such longer-term influences as climatic change, one-way species migrations or evolution (White 1979). Modern ecology has begun to recognize that moderate levels of natural disturbance enhance landscape complexity and species diversity (Pielou 1975; Pickett and White 1985;  65 Petraitis, Latham and Niesenbaum 1989). Natural disturbances potentially help to support an increased level of species diversity by producing a diverse landscape with many seral stages. Landscape heterogeneity helps fulfill multiple habitat requirements and reduces the likelihood that a few species will exert long-term dominance (Hunter 1990). Less intense alterations to ecosystems typically leave the soil, soil fauna and seed bank intact, allowing rapid release of suppressed trees and swift invasion by early succession species (Hansen et al. 1991; Burton et al. 1992). Severe disturbances, on the other hand, result in the immediate "destruction of many populations, with future recovery dependent on gradual re-invasion, establishment, growth and (sometimes) microsite amelioration in the process of primary succession" (Burton et al. 1992, 230). High-severity disturbances, therefore, push back biotic communities to a much earlier, less diverse successional stage (relative to less severe disturbances), where successional development builds biodiversity and structural complexity over longer periods of ecological time (Hansen et al. 1991; Burton et al. 1992). 17 At the landscape level, all intensities of natural disturbance likely increase species diversity over variable periods of time, providing they are contained within ecosystems which are sufficiently large to maintain internal recolonization sources. An ecosystem which has what Pickett and Thompson (1978) call a "minimum dynamic area" provides enough spatial latitude to allow for a shifting mosaic of its variably disturbed patches, while containing and maintaining its characteristic level of species diversity and species composition (Noss and Harris 1986). Landscape heterogeneity and species diversity at large scales confer resilience (the ability of a system to return to a state that existed prior to a natural disturbance) to a region  (Department of Natural Resources 1991). "Natural disturbance" is emphasized here because native species in a particular area 17 As the "New Forestry" discussion pointed out, however, structural legacies left behind following disturbance of old-growth may help promote more rapid development of some regenerating stands. Because it is rare for even severe natural disturbances to consume logs, snags and trees fully, succession rarely starts from a completely denuded or lifeless state.  66 have evolutionarily adapted to a specific disturbance history, which may not be accurately copied by human-caused disturbances (Noss and Harris 1986; Noss 1990). However, the recolonization value of wilderness benefits human-altered areas as well, providing the natural area's range of influence (as a seed source for example) extends into the modified environment. The ability of an area to naturally recover from disturbance, whether from natural or human forces, depends in part on whether or not seeds, propagules and migrating animals are available nearby to recolonize a disturbed area (Norton 1987). Island Biogeography has taught us that a more rapid rate of recovery and an improved chance of developing to biologically diverse late seral successional stages are often associated with a surrounding, proximate refugia which harbor a diversity of colonizing species (MacArthur and Wilson 1967; Harris 1984). This accentuates the importance of maintaining natural systems with a community structure containing a large and diverse pool of nearby species which are available for colonization (Norton 1986, 1987). By supplying colonists from a range of successional stages, wilderness maintains, regulates and diversifies both its internal mosaic of naturally shifting disturbance regimes (and thus its biotic communities) and any contiguous human-damaged regions of land left untouched following disturbance. Intensify Diversification The previous section revealed the importance of retaining diverse arrays of species within healthily functioning ecosystems as agents for encouraging recolonization of disturbed areas. Recolonization sets the stage for and pushes the process of succession. Succession is the directional change in species composition and community structure over time (Smith 1980). Although biotic communities follow variable patterns of successive occupation and replacement, unimpeded succession always moves a system towards greater diversity. 18 The sequential change in biotic communities generally starts with an early stage characterized by an 18The  specific character of ecological succession is governed largely by site conditions (moisture and nutrient status of soil, macro- and micro-climate, etc.) and community characteristics (Smith 1980).  67 abundance of organic production, short food chains and simplified food webs and progresses towards a highly stable community with a highly evolved structure, complex food webs, and a portioning of energy among many units within the system (Smith 1976). The variety of species residing in a geographical area thus advances the opportunity for further diversification of disturbed ecological systems by supplying a range of colonizers through ecological and evolutionary time (Norton 1986). Norton (1986, 1987) describes the tendency of diversity to beget diversity in ecosystems through time as "self-augmenting diversity spirals". In ecological time, the interrelated processes of colonization and succession drive the system towards a heightened level of diversity by actuating a series of sequential events and stages. Following disturbance, the first invaders occupying a site are pioneer plant species. Pioneers tend to be opportunists and generalists, thriving on any site where disturbance has exposed a range of unoccupied niches and eliminated most forms of competition. Through high dispersion rates, fast growth and short life spans, early invaders have an adaptive strategy suited for rapid colonization of open spaces. Such adaptive characteristics, however, make them ill-suited for competing in the long run. Stronger competitors, which tend to put less energy into dispersal and have longer life spans and greater biomass development, slowly dominate a site by tapping into more resources and shading out the shade-intolerant early invaders (Harper 1977). As more and more species invade and hold territory in time and competition builds, selective pressures will promote the emergence of a community made up of a greater number of specialists (with increasingly sophisticated mechanisms to exclude competitors) occupying a greater number of spatially confined microhabitats (Whittaker 1972). MacArthur and Wilson (1967) explain that such temporal diversification results from the packing of more species with narrower niches along resource gradients (alpha species diversity 19 ). The alpha species diversity which occupy the resource gradients set by and contained within local ecosystems 19 Norton (1986, 1987) uses the term "within habitat diversity" instead. Both describe the number of species coexisting within a uniform habitat (Cody 1986).  68 thus expands in ecological time (Norton 1987). Superimposed on and intensifying the diversification taking place over these relatively shorter time periods are processes promoting the augmentation of diversity over longer periods of evolutionary time. Beta species diversity 20 broadens and multiplies over these expanded time frames, as more niche spaces are created along what MacArthur and Wilson (1967) describe as "niche axes". Norton (1987, 57) describes an axis as "one particular determinant of survivability within a niche, such as a source of food, availability of shelter from predators, et cetera." Niche spaces increase in evolutionary time as an expanding number of species enter a biotic community, fit themselves in between other species and exploit a particular combination of resources to which they are singularly adaptively suited. Some species create new resource options, providing resources and suitable niches for other species. The net result is an expansion in the packing of species along axes, and consequently an increase in the beta diversity of species and community structure (Whittaker 1970). Gamma species diversity 21 , which is a function of the other two types of species diversity (Norton 1986), therefore increases as new niche axes (beta diversity) are added and lengthened and more species are packed into narrower niches (alpha diversity) along these axes (MacArthur and Wilson 1967; Whittaker 1970, 1977; Norton 1986, 1987; Burton et al. 1992). The intensification of diversity is self-augmenting over ecological and evolutionary time because a greater level of total diversity across a heterogeneous and dynamically shifting landscape supplies increasing levels of colonizing species. Invaders of all successional dependencies promote the process of diversification further by intensifying competition for niche spaces in habitats and forcing habitat differentiation. This allows greater numbers of species to occupy greater numbers of narrowly specialized niches. These species in turn create 20Norton  (1986, 1987) uses the term "between habitat diversity" instead. Both describe the degree to which species lists vary between habitats within the same locality. Changing habitat is determined by climatic, soil and topographic gradients (Cody 1986). 21 Norro -^(1986, 1987) uses the term "total diversity" instead. This is a measure of the variety of species existing across the landscape (Whittaker 1977). It reflects the diversity that can form in similar habitat at different geographical locations (Cody 1986).  69 more heterogeneity across the landscape and provide a diversity of colonizers to repeat another loop of the self-augmenting diversity spirals (Norton 1986, 1987).  Contribute Support to Other Species A related value stemming from the interrelationships that develop during successional recovery is the interdependent support provided by species to each other. Dependencies range from complete dependencies (obligate mutualists), such as many coevolutionary fungi-plant symbionts, to lesser dependencies. The contributory value of species can be observed at different interactive levels, from one species providing specialized support to another species to the collective sustenance of the entire biotic community. Species also sustain one another at various successional stages, interdependently pushing the process of succession along to greater levels of diversity and stability. The previous sections emphasized the importance of proximate, diverse colonizers for accelerating the rate of ecosystem recovery. It was noted that each phase is dominated by a particular assemblage of species which harbor characteristics suited to exploiting particular conditions. Successional recovery and diversification is helped along by the contributory support of species interactions. In other words, species diversity contributes to species diversity because various ecological relationships develop which lend support to species and, through the totality of linkages, to the ecosystem as a whole. Pioneer plant species, for example, play important roles in maintaining the structural integrity of disturbed soils and retaining nutrients and key soil organisms during the critical tree establishment period (Perry 1988; Perry and Maghembe 1989). The importance of soil to forest health and productivity underscores the crucial function of pioneer plants. Perry (1988, 9) highlights soil's significance by stating that "Soil has much more than an immediate influence on tree growth, it is the connecting link between a forest that is destroyed by harvest or natural processes (e.g. fire) and the succeeding stand of trees." Pioneers help bridge the gap and hasten recovery by occupying a site rapidly and fulfilling integral ecological functions. Natural  70 vegetation minimizes the losses from nutrient leaching, soil erosion and moisture runoff, which are commonly associated with bare ground, by intercepting and evapotranspiring moisture via canopies and developing root systems. Root penetration anchors and stabilizes soil and organic matter, improving soil structure, promoting deeper water percolation and enhancing soil's nutrient and moisture holding capacities (Ehrlich and Ehrlich 1981; Norse et al. 1986; McNeely et al. 1990a). By forming fundamental associations with key soil organisms, pioneers enhance the nutrient retention and cycling characteristics of a soil, improving the habitability of a site for other plant species. Nitrogen-fixing species (e.g., Alnus spp.), for example, house a family of bacteria in their root nodules which are capable of converting atmospheric nitrogen into a form which is usable by all plants (Perry 1988). Symbiotic relationships are also forged between the roots of many plants and mycorrhizal fungi. Besides reciprocally benefitting each other, mycorrhizal symbionts benefit the ecosystem by supplying essential minerals and adding stability to the soil biological community and concomitantly enhancing tree seedling and other vegetation survival (Pirozynski and Hawksworth 1988; Perry and Maghembe 1989; Perry et al. 1989). Although these interdependent linkages are relatively less complex in pioneer and early seral communities, such species collectively produce conditions which support a greater array of more competitive species, propelling successional development forward. The value of early and mid-seral stage colonizers in diversifying ecological systems is demonstrated by Norton (1987, 61) in the following paragraph: Even species that fail in the competition for niche space in a particular ecosystem have an important contribution to the structure of that system. By intensifying the competition for niche space, they have, even in the process of being extirpated by superior competitors, exercised adaptational pressures on those species, caused further specializations, and developed more tightly packed and more highly integrated species relationships. Consequently, even the losers in such competitions are important. The ecosystem development and total diversity of an area are served, both directly and indirectly, by the existence of every species. That is, even if a species ultimately disappears because of its inability to compete, it still may have contributory value.  71  Increments in diversity over time eventually lead, barring further disturbance, to a highly complex and diverse biotic community and food web. Climax communities, such as the old-growth temperate forests of B.C., tend to be structurally variable with multiple canopy layers contributing to vertical complexity and patchy gap distributions contributing to horizontal diversity. Structural diversity is also contributed by large live trees, snags and logs (Meslow, Maser and Verner 1981; Maser et al. 1988; Thomas et al. 1988). Such morphological variability provides expanded levels of microhabitats and niche spaces to increasing numbers of specialized species and provides more pathways for fixing nitrogen and retaining and recycling nutrients (Franklin et al. 1981; Maser and Trappe 1984). As a result, complex ecological relationships have evolved within old-growth forests, interdependently linking community members (Forman and Gordon 1986; Franklin and Forman 1987). Such interdependencies not only provide individual support to species pairs through coevolved adaptations but provide collective support to all community members through the promotion of ecological stability. 22 It is asserted, therefore, that the complicated patterns of relationships that have evolved among the diversity of interacting species confer stability to natural systems (Smith 1976; Franklin and Forman 1987; Perry 1988). Burton et al. (1992, 231) exemplifies this assertion by pointing out that "The more pathways via which energy can flow to the top from the bottom of a food web, the less severe will be the consequences of failure of any one pathway." Complex ecosystems have many functional duplications and compensatory interactions which provide back-ups when an element or linkage is disabled 22 Stability  refers either to constancy (the ability of a system to resist change caused by disturbance) or resilience (the ability of a system to return to a predisturbance state). Climax communities are often considered stable because they are diverse. Various conceptions of stability and various views on its links to diversity have been forwarded. This paper holds the view that diverse, highly evolved systems are dynamically stable (autogenic). This implies that they have reached a stage where they are able to buffer, within equilibratory limits, the impacts of normal disturbances that originate from within the system (indefinitely replacing community members). Their intricate connectivity, however, makes them more sensitive to abnormal disturbances that are outside of their normal experience. This latter trait is particularly evident when keystone species (species which have a disproportionate influence on the characteristics of an ecosystem) are eliminated from a system (see Norton 1987 for full discussion).  72 (Department of Natural Resources 1991; Burton et al. 1992). That diversity contributes to stability reinforces the essential roles that individual species play in maintaining the life cycles of other species and the integrity of the system as a whole. Most species contribute to the survival of other species, with the importance of their ecological roles varying by matters of degree (Leitzell 1986) and with differences in levels of support depending on a species' numbers, biomass, status in the food pyramid, among other attributes (Myers 1979). This collective support implies that any loss in diversity may result in further losses when anomalous disturbances occur (Myers 1979; Norton 1987). With scientific understanding of ecosystems and their species interdependencies being limited, it is impossible to determine which species is cardinal to the functionality of a system (keystone species) and whether or not its extirpation will result in a cascading wave of species losses and biological simplifications (Norton 1987). Consequently, the contributory value of species, though difficult to assess, extends through all levels of dependencies and relationships as well as over great expanses of successional time.  Circulation 1 Cleansing of Air and Water Wilderness as ecosystem, besides maintaining the functional integrity of biotic components and relationships, also contributes to the effective functioning of abiotic systems. Its integral influence on the integrated terrestrial, aquatic and atmospheric cycles encompasses large dimensions of space and time, from local to biospheric and daily to millennial. This section will concentrate on a few of the more significant and conspicuous abiotic processes which are driven in part by wilderness systems. These revolve around the circulation and cleansing of air and water. Specifically, this discussion includes ecosystems' effects on air quality, water quality/quantity and climatic/atmospheric regulation. The first two impacts provide obvious anthropocentric value on a local or regional scale (i.e., breathable air and drinkable water). The moderating influences of wilderness on such climatic processes as the carbon and hydrological cycles are less directly tied to human values  73 and are as extensive in scope as the biosphere as a whole. Nonetheless, they contribute, in fundamental and pervasive ways, to the health and survival of people. All three are interrelated and are integrated within the larger biospheric system, whose aggregation of cycles function most efficiently in the presence of relatively undisturbed ecosystems (Pearsall 1984; Wolf 1985). Air Quality Forested ecosystems have the ability to cleanse the air of human-made and natural atmospheric pollutants and particulates (Ehrlich and Ehrlich 1981; Wolf 1984; Norse 1990; USNRC 1992). The amount of airborne wastes that can be intercepted by vegetation depends primarily on the canopy leaf area existing in a region along a local weather system's path. Oldgrowth forests are particularly adept at acting as pollutant traps, as their uneven, large canopies create turbulence, effectively promoting chemical precipitation on needles (Norse 1990). Providing that levels of human-made emissions remain below toxic levels and are not dominated by heavy metals, many chemicals form part of a tree's biochemical cycle (and thus act as fertilizers) and a region's biogeochemical nutrient cycle. Among these are nitrogen and sulphur, byproducts of the burning of fossil fuels, which are taken up and metabolized by plants and form fundamental constituents of plant and animal proteins (Norse 1990). Water Quality /Quantity Related to the hydrological cycle and closely linked to ecosystems' functions in  moderating climates and forming and maintaining soils is their role in regulating freshwater supplies. Natural vegetation cover and organic litter stabilize hydrological functions by helping to control erosion, reducing downstream sedimentation and flooding and regulating strearnflows (Dixon and Sherman 1990). Forest canopies, plants and duff protect the soil by shielding the ground from the force of rain. Needles and leaves of tall trees provide increased levels of moisture to a system by intercepting rain and condensing fog onto their surfaces (Ehrlich and Ehrlich 1981; McNeely et  74 al. 1990). Large canopies and boughs also support large quantities of snow, allowing it to melt slowly rather than piling up. Rain on snow events are less likely to lead to catastrophic flooding under conditions of gradual melting (Zuckerman 1991). Deep penetration by tree or other vegetation roots aid in holding soil in place and make the soil more permeable to rainwater, particularly following root decay when small channels are left behind. Soil organisms and small burrowing animals also create similar drainage channels (McNeely 1988; Ehrlich 1990; Zuckerman 1991). Decomposers contribute to subsurface water percolation by creating organic matter which binds soil particles, improving such structural properties as increased pore size (Norse et al. 1986). As a consequence, run-off is slower and more uniform across the landscape as the forested system absorbs and retains more water over time, gradually releasing it through springs and streams (Ehrlich and Ehrlich 1981; Norse 1990). Uniformly flowing run-off and gradually released sub-surface water help modulate streamflow fluctuations by serving steady, year-round flows (McNeely 1988; Ehrlich 1990). Reductions in erosion and flooding in turn reduce incidences of stream siltation and fish habitat destruction (Maser and Trappe 1984; Norse et al. 1986; Flamm 1989). Watershed protection is also advanced when water quality is maintained. The affinity of soil particles and leaf surfaces to various pollutants (acid, heavy metals, etc.) helps scrub harmful elements from rainwater. The surface and subsurface water, therefore, may be of higher quality than the original precipitation (Bormann 1976; Ehrlich and Ehrlich 1981; Norse 1990). Climatic/Atmospheric Regulation  The earth's ecosystems have a considerable influence on regional and global atmospheric conditions. Their regulatory control over atmospheric conditions takes place over a range of time periods and spatial scales. Their influence on climate is also exerted both physically and chemically.  75 Physical Influence  In the short-term, terrestrial systems exert their influence on weather patterns by physically altering surface and atmospheric conditions (such as temperature and the hydrological cycle). Differential heating and evaporation of the earth's surface drives the circulation of the atmosphere. Forested ecosystems regulate such weather systems by affecting the albedo (reflectivity) of the land surface and by affecting the production of clouds, which also have reflective properties. For example, in temperate or tropical rainforests, much of the intercepted fog and rain is held and then released to the atmosphere by the transpirational pumping action of plants. This released moisture helps form clouds during times when adiabatic conditions permit condensation. Increased incidences of cloud cover promote heightened levels of precipitation and reflectivity, helping to produce relatively moister and cooler conditions (Ehrlich and Ehrlich 1981). Undisturbed forests also help to maintain rainfall in their immediate vicinity by promoting atmospheric turbulence by way of their uneven canopy surfaces (McNeely 1988). In the absence of vegetation, there would  likely  be increased runoff,  less transpiration, smoother air flow, as well as altered albedos at the surface and in the atmosphere. If land clearing is extensive enough, the likely effect from these physical and functional alterations would be a regional climate which is hotter and drier than the predisturbed state (Ehrlich and Ehrlich 1981; McNeely 1988; Ehrlich and Wilson 1991). Chemical Influence  Natural systems also control and ameliorate global, long-term climatic conditions by helping to maintain and regulate the concentrations and cycles of key atmospheric gases. In general, ecosystems help maintain the gaseous composition of the atmosphere, preventing the relative mix of gases from straying outside of tolerable limits or changing too rapidly and thereby allowing the biota to persist and adjust (Ehrlich and Wilson 1991). By being intricate components of the carbon cycle, forests play particularly pivotal roles in keeping the climate in balance. With concerns being voiced over rising levels in carbon  76 dioxide and global temperatures (and their associated impacts on climate, ecosystems 23 , species, sea level rises, etc.), attention is increasingly given to the "carbon sink" function of forests. Although there is disagreement in the scientific community over how much of the global temperature rise is attributable to human-caused carbon dioxide (CO 2 ) emissions, most scientists agree that the average temperature of the Earth's surface has increased by about 0.5 °C since 1880 (Houghton 1991). Also, the atmospheric concentration of CO 2 is approximately 25% higher than preindustrial levels (Flavin 1990; Houghton 1991). Other gases which have been linked to the greenhouse effect (warming associated with gases which allow sunlight through but trap radiated heat) have also built up steadily as a result of industrial activity (e.g., fossil fuel burning, deforestation, etc.). These include nitrous oxide, methane and the synthetic chemical class of chloroflouro-carbons (Arrhenius and Waltz 1990; Flavin 1990). The evidence that the warming that is occurring is the result of the greenhouse effect comes from strong historical correlations between increased concentrations in gases such as CO 2 and increased temperatures. The integration of the two is consistent with the positive feedback mechanisms which are observed in the earth's climate. Moreover, global climate models (such as General Circulation Models (GCM's)), which simulate the physics of the atmosphere and the complicated linkages and feedbacks taking place there, verify a tight coupling which is consistent with historic temperature changes. Such models, which are also used for future projections, predict a 1.5 to 4.5 °C average global temperature rise for every doubling of the CO 2 concentration (Houghton and Woodwell 1989; Houghton 1991). If current emission trends continue, CO 2 and other greenhouse gases are predicted to push the climate into the doubled CO2 temperature regime in less than 50 years (Perry and Borchers 1990). Calculations suggest that warming over the earth's surface will be uneven, with some areas experiencing significantly hotter climates, others only slightly warmer, and still other 23 See Peters and Darling (1985), Franklin et al. (1991), Perry and Borchers (1990), Peters  (1991, 1992) and McAllister and Dalton (1992) for analyses on the potential impacts of the greenhouse effect on ecosystems and species.  77 regions registering cooler climates (Jones and Wigley 1990). Such predictions are tinged with uncertainty because the models do not adequately capture the complexity of the earth's climate system. However, enough consistency exists in the temperature change predictions of the various GCM's, as well as in the opinions of scientists on the disruptions such changes could incur in natural systems, to merit caution and concern (Peters and Darling 1985; Peters 1991). Forested ecosystems are pivotal variables in GCM's. This is because they are important components in the global carbon cycle. Trees and other plants absorb CO 2 during photosynthesis and store the carbon in their biomass (leaves, wood, roots, etc.). The photosynthetic process also involves the release of oxygen. Vegetation gives off a proportion of this CO 2 and extracts atmospheric oxygen (0 2) when they respire at night. Decomposition and fire also result in the release of CO 2 and the absorption of 02 . Through an elaborate and interdependent feedback system, forested ecosystems help maintain relatively constant carbon and oxygen levels in the atmosphere, oceans and land (Lyman 1990). In an unperturbed world, the carbon cycle remains balanced because inputs and outputs of CO 2 to and from the atmosphere remain equal. Terrestrial ecosystems contribute to this climate stabilization by balancing its sources and sinks of CO 2 . Terrestrial ecosystems contain approximately 2000 billion tons of carbon, which is almost three times the amount held by the atmosphere (Melillo et al. 1990). Each year about 100 billion tons of carbon is photosynthetically removed from the atmosphere, representing 5% of the terrestrial carbon and 14% of the atmospheric carbon content. 40 billion tons of carbon is released during this annual period through respiration. The 60 billion ton difference between the total amount of carbon fixed during photosynthesis and the respiratory release is net primary production (Houghton and Woodwell 1989; Melillo et al. 1990). Net primary production represents the amount of carbon that is used by terrestrial plants to build and maintain biomass and reproduce. It also represents the amount of energy added to the first trophic level, which provides the underlying  78 supportive basis for the maintenance, growth and reproduction of all terrestrial heterotrophs, including humans (Vitousek et al. 1986). The atmospheric carbon budget is maintained within limits in the absence of human emissions because net primary production is approximately balanced by the release of carbon from soils through microbial respiration (Melillo et al. 1990). Any measures to stabilize atmospheric CO 2 would therefore require not only a reduction in fossil fuel combustion, but preservation of ecological systems and expanded reforestation efforts as well. Approximately 5.5 billion tons of CO 2 are released annually to the atmosphere through fossil fuel combustion (Goreau 1990). Goldenberg et al. (1988) estimate that the industrialized world could reduce emissions by 25 billion tons by increasing energy efficiency. Halting deforestation would soak up 1.5 to 3.0 billion tons of this industrial CO 2 flux. A massive reforestation program might extract as much as 2.0 to 3.0 billion tons from the atmosphere annually for a few years (Houghton 1991). Goreau (1990, 230) highlights the importance of sustaining ecosystems for the purpose of maintaining a balanced carbon budget in the following sentences: CO2 stabilization is primarily a biological problem, not a technological or geophysical one, because the bulk of carbon flows through biotic processes. Increased photosynthesis and carbon storage in vegetation, soils, and sediments is therefore needed along with emissions reductions to stabilize CO 2 . This can be accomplished by halting destruction of highly productive ecosystems and increasing productivity of areas already degraded. Arguments in recent years have ensued over whether or not CO 2 content in the atmosphere could be lowered if slow growing, "decadent" old-growth forests were cut down and replaced with fast growing, intensively managed young forests. A study conducted on forests in the U.S. Pacific Northwest indicated that although young forests have a higher net primary productivity (and thus sequester more atmospheric CO 2), conversion to plantations results in a net flux of CO 2 to the atmosphere. Using three lines of evidence (current reductions in carbon from old-growth harvesting; a model of carbon dynamics in both successional stages; and a simulated comparison of carbon storage in both), Harmon, Ferrell and Franklin (1990)  79 found that harvesting, removal of logs from a site and eventual degradation of wood products result in a greater net release of CO 2 in the long run than intact old-growth forests. Although not conclusive, their results indicate that the high, long-term carbon storage capacity associated with old-growth may be more important in evening out atmospheric CO 2 levels than the high, but short-term productivity associated with young stands. Over very long time frames this reasoning may prove to be particularly valid; as has been previously stated, studies on the ecological characteristics of old-growth forests have found that untouched, complex ecological systems have a higher long-term productivity than plantations (particularly genetically-restricted monocultures) on some sites, due in part to the former sustaining stable soils and biodiverse, resilient biotic communities. A diversity of such systems across a heterogeneous landscape may prove to be the best means of ameliorating and mitigating the ecological disruptions associated with climate change. Maintaining a diversity of species with large populations (to permit successful colonization of emerging habitat) and large geographic ranges with no physical barriers (to permit escape from inhospitable, changing habitat) will help species adjust to changing conditions (Peters 1991). These and other strategies linked to preservation efforts can be thought of as forms of insurance, designed to provide an ecological safety net in order to protect the full range of wilderness values in a changing, uncertain world.  Summary Conflicts between different stakeholders have arisen in recent years which have brought the issue of wilderness preservation to the forefront of public concern. These concerns have centered largely on the overall questions of what to protect and how much to protect. However, before wilderness allocation decisions can be effectively made, as answering such questions implies, it is important for all involved in the debate to have a common understanding of the concepts involved and the various categories of values associated with natural areas.  80 This chapter attempted to help the debate along in this regard by defining the concepts of benefit, wilderness and wilderness value and by outlining and organizing within a structured hierarchy the fundamental values of wilderness. The taxonomy included what are considered to be the major wilderness values, namely market based and non-market based human demandrelated values, human spiritual values and ecological values. It was asserted that developing a rationally organized and relatively complete wilderness value taxonomy helps advance the allocation decision making process by providing a basis for a clearer and more developed understanding of such values. This in turn helps in the forging of sounder decisions by improving communication, raising the level of debate, facilitating research, garnering greater levels of support for the decision processes in place and providing valuable information for and validating specific evaluation procedures. In order for the wilderness values outlined here to be of any use in the decision-making process, however, the quantitative or qualitative significance of these values to the people involved should be ascertained using a viable elicitation technique; comparisons can then be made between the preference measures of wilderness values and preference measures of the conflicting values which should also be weighed in any specific decision process. The nonmarket value elicitation procedures themselves are outlined and critiqued in the following two chapters, with examples of familiar indirect approaches covered in Chapter IV and direct approaches in Chapter V. The following chapter will also describe the nature of value tradeoffs, a necessary consideration for understanding the decisional balancing which is required in allocation conflicts.  81 Chapter IV INDIRECT APPROACHES TO ELICITING VALUES AND THE NATURE OF VALUE TRADEOFFS  Introduction The previous chapter unveiled the rich array of wilderness values which are prized by many people. However, scarcity of supply within a decision milieu which includes many conflicting interests restrain decision makers' abilities to preserve it all. Wilderness, in such a constrained decision environment, is thus one of many land use options. Competing claims on a limited forest resource impel decision makers to allocate land-use on the basis of choice criteria which weigh the benefits and costs associated with alternative sets of actions. In British Columbia, the most intractable, polarized and high-profile land-use conflicts have involved forested areas which are jointly valued by environmentalists and logging advocates. Although only a subset of the total land-use allocation conflicts in which the government, industry and various interest groups are engaged, the value-laden nature of these disputes brings into focus the central role that values play in imbuing an allocation decision with discord. Resolution of these and other value conflicts over forest resource issues depend in part on the explicit elicitation of values within a framework which balances costs and benefits. Because many wilderness benefits are exogenous to the supply and demand functions of competitive markets, non-market based demand values, spiritual values and ecological values have no revealed prices to signal the quantitative strength of peoples' preferences for them. Other approaches must be used in such cases. This and the next chapter will outline the more widely documented non-market valuation 24 methods. The techniques classified here are arranged in decreasing order of 24 "Valuation"  refers to the process and procedure of estimating the value of non-market goods and amenities, so as to provide a basis of comparison with the value of market items. "Value  82 dependence on market data, and are organized into two broad categories: indirect methods (which are based on individuals' revealed preferences) and direct methods (which are based on individuals' expressed preferences). The former set, which makes up the better part of this chapter, valuates benefits using surrogate markets, and includes the travel cost method and hedonic pricing approach. The latter set, which is reviewed in the next chapter, hypothetically valuates benefits, and includes willingness to pay surveys and multi-attribute utility approaches. The strengths and weaknesses of each approach in terms of their potential in clarifying the values associated with preservation decisions will also be included in the discussion, as will be the decisional situations in which they have been applied. Their efficacy in eliciting specific wilderness values will help determine the particular usefulness of each in pushing the decision-making process forward. However, a broader context which links all of the evaluative procedures will first be delineated; the discussion thus begins with a descriptive overview of the nature of the costs, benefits and tradeoffs which accompany any allocation decision which attempts to weigh the respective values of wilderness and timber resources.  Value Tradeoffs in Preservation 1 Harvesting Conflicts  Wilderness land-use issues have provided a focal point for many of the high-profile disputes in B.C. in recent years. The B.C. Wilderness Advisory Committee (1986, 6), which was formed in part to examine some of the broader issues relating to wilderness allocation in the province, recognized the impediments to resolving disputes between forest industry and wilderness interests: The difficulty of resolving conflicts between logging and wilderness preservation was raised frequently. The problem appears to arise from having overcommitted our timber resources in the past. This makes it difficult to provide a forestry company with replacement timber when a proposal is made to remove some timber from the elicitation" and "evaluation" are used synonymously with valuation here. Comprehensive reviews of valuation methods are provided by Hufschmidt et al. (1983), Hyman and Stiftel (1987) and Braden and Kolstad (1991).  83 commercial forest. Conflicts over resource use, therefore, arise out of expanding demands being placed on an increasingly scarce resource. Exacerbating the difficulties in settling such disputes are the complexities and uncertainties surrounding the natural and socio-economic systems making up the decision-making environment. The conflicts themselves have several causal dimensions which further complicate the task of resolution. Dorcey (1986) identifies four types of conflict: (1) cognitive conflict (conflict arising out of different understandings of the nature of the situation under dispute, such as differences in technical judgments on factual information); (2)  interest conflict (conflict stemming from disagreements in opinions over the distribution of costs and benefits); (3) behavioral conflict (conflict which develops as a result of clashes in personalities and temperaments); and (4) value conflict (conflict which stems from differences in preferences over the outcome of the dispute). Although all four conflict types may influence the nature of a dispute, this discussion concentrates on value conflicts. However, it must be noted that all forms of conflict provoke tensions and thereby complicate the task of finding a mutually satisfactory solution. In addition, conflict resolution occurs within the wider sphere of decision-making, where all aspects of conflict must be tackled. This may imply that various methodological strategies and institutional frameworks will need to be employed to handle various facets of conflict. For example, SussIdnd and Cruikshank (1987) and Wood (1989) identify interest conflict as one which lends itself in particular to the process of negotiation. Also, the direct approaches reviewed in Chapter V must also deal to some degree with conflicts arising from personality clashes and disagreements over facts, as these can factor quite significantly into the success of the outcome and the validity of the elicitation results. Value conflicts stem from different judgments about the objectives to be accomplished in a specific decision (Dorcey 1986; Wood 1989). Differences in judgments about the desirability of the tradeoffs between such opposing objectives as environmental quality and  84 economic growth create particularly intense value conflicts. Helping to promote efficient, equitable and mutually agreeable tradeoffs lies at the heart of successful valuations. Resource use tradeoffs focus on the gains (benefits) and losses (costs) which are connected to each use alternative (Phillips 1985). The scarcity of the forest resource implies that it is both valued and limited and that one benefit (wilderness) can only be actualized at the cost of forgoing another valued use (timber). Consequently, the benefits of wilderness protection must be weighed against the opportunity costs of protection, including the economic returns which could have been earned had the area been converted to a commercial forest. The ultimate question, from an anthropocentric standpoint, is how to best satisfy the collective well being (preferences) of all individuals associated with a resource decision, given that choices are constrained. Biotic resource allocation issues, including wilderness/harvesting conflicts, must be resolved within this broader context (Randall 1986). The process of comparing costs to benefits in any resource allocation decision which stows conflicting values is an unavoidable undertaking (Clawson 1980). However, decisions involving wilderness and timber resources present numerous impediments to the efficient and equitable enumeration of costs and benefits. As previously mentioned, complexity encompasses every forest resource allocation decision. The nature of biotic resources are particularly complex. One stream of complexity relates to those characteristics of the goods and services supplied by natural areas which contribute to market failure. Many goods and services coming from wilderness are nonrival; this means that the total amount of the goods and services available can be enjoyed by anyone without diminishing their supply. In other words, once provided to one they are effectively provided to all. Many also tend to be non-exclusive in nature; it is not possible, in such cases, to exclude anyone from consuming or appreciating the wild resources. Natural areas also contribute to market failure by producing off-site effects. Such benefits as the regulating and cleansing of water and the stabilizing of climate are examples of positive externalities. Another economic hindrance is the tendency of development  85 decisions (e.g., timber extraction) to disrupt the highly evolved, intricate structures and functions of natural systems for extended periods of time. Because technology by and large cannot substitute for or recreate lost life forms within an acceptable time frame, such decisions are often considered irreversible. These public good-type properties introduce impediments to the proper functioning of markets and obstacles to the valuing of wilderness benefits (Randall 1983, 1986, 1987a; Anderson and Bishop 1986; Dixon and Sherman 1990). Uncertainty also complicates the valuation exercise. Bishop (1978) defines two major types of uncertainty: "natural uncertainty" and "social uncertainty". The first relates to the incomplete knowledge of what the future supply and potential uses of various known and unknown species will be. Our fractional and deficient understanding of biological diversity exacerbates natural uncertainty, as represented by Ehrenfeld (1988, 214): "We do not know enough about any gene, species or ecosystem to be able to calculate its ecological and economic worth in the larger scheme of things." Norton (1988) adds that the complex interdependencies existing in natural communities hamper valuation by presenting indivisible values which encompass both the independent and contributory value of species, the latter of which is particularly difficult to assess (see Chapter III, this volume). The second form of uncertainty relates to the incomplete knowledge of what future demand will be. Both types of uncertainty concern the difficulty of dealing with futurity in inter-temporal allocation decisions (which demand the consideration of equity) and value elicitation procedures (which demand the efficient assignment of present value). As Hanemann (1988, 197/98) summarizes, "So many consequences involve future events, which are extremely difficult to predict - long-run ecosystem impacts, economic variables, and even the preferences of future generations, who are not around now for one to observe their behavior." The collective effect of the non-market characteristics of wilderness on value elicitation efforts is illustrated by Dixon and Sherman (1990, 30) when they state that "The market failures mentioned previously add to the information problem - their presence makes it hard to assign accurate values to many benefits of protected areas. Predicting the future value of these  86 benefits simply adds to the uncertainty problem." The price mechanism does not capture, in the presence of such externalities, the full social costs associated with the goods and services produced in an economy. Markets thus fail to allocate efficiently in such cases (Stanbury, Vertinsky and Thille 1990). The concept of economic efficiency lies at the core of tradeoff analysis. Generally speaking, the greater the net benefits arising out of a resource allocation decision the greater the economic efficiency of that decision. As such, choosing the resource use pattern that renders the highest net benefit to society may be an appropriate criterion in resource use decisions (Phillips 1985). However, resource values must be known in order to quantify benefits and costs and, as we have seen, many wilderness values are not amenable to traditional forms of economic quantification (i.e., price). Because wilderness contains many extra-market values and because the information base remains extremely weak, significant restrictions have been placed on the registration of wilderness advocates' choices, preferences and values in the marketplace. As a result, timber values, which are easily quantified, have often been weighted more heavily in past B.C. land use decisions. Justifications for ignoring or undervaluing non-timber values in allocation decisions of the past have also been strengthened by the fact that the forest resource is so important to the economy of B.C. and its industries, workers and communities. Estimated provincial revenue coming from the forest products industry for 1993/94 will be $913 million (B.C. Ministry of Finance 1993). Government revenue earned from such timber products as lumber, plywood, related wood products and pulp and paper is generated from stumpage, rents, various fees, royalties, the small business forest enterprise program and interest payments (B.C. Ministry of Forests 1991a). According to an input-output model which utilized 1979 data, 8.4% of B.C.'s direct employment and 13.8% of the province's gross provincial product (GPP) (at factor cost) were generated by the forest products sector (Jacques 1988). Direct forest employment in B.C. in  87 the late 1980's stood at around 100,000 (B.C. Ministry of Finance 1991; B.C. Round Table 1991). When all direct, indirect and induced economic impacts are included, the timber industry was estimated to contribute 175% of total provincial employment and 25.5% of provincial GPP (Jacques 1988). All told, the value of forest products produced in B.C. in 1990 was $11.2 billion. This represents approximately one-third of Canada's total (B.C. Round Table 1991). Community stability is another important benefit derived from the extractive value of the forest resource. In B.C., 124 communities were totally dependent on the forest resource in 1981 (White et al. 1986). Including only incorporated B.C. communities, 32% of 112 communities are considered forestry specialized. An additional 14% have the forest sector as one of two dominant sectors and another 19% have forestry as one of three sectors dominating the economic base (Simons 1991). Stanbury, Vertinsky and Thille (1990, 11) summarize nicely the socio-economic significance of the B.C. forest industry: Since the forest industry in B.C. is the most important resource in sustaining the provincial economy and the social structure in many communities, its use in achieving a variety of socio-economic benefits cannot be overlooked. Indeed, the role of the forest as the economic mainstay of certain communities and as a source of "higher quality" employment has, to a large extent, shaped the allocation of forest land and investments in it over the past decade. Stanbury et al. go on to point out that while the traditional emphasis of forest resource allocation issues was the attempt at reconciling conflicting uses, the focus has since shifted to identifying all benefits and costs, including the difficult to measure benefits of wilderness. This shift in focus has levelled increasing attention on various methods of weighing benefits and costs when evaluating alternatives that involve commercial values of the timber resource and market and non-market values of wilderness. Benefit-cost analysis (BCA), which involves the evaluation of a stream of benefits and costs over some chosen time period, is the most widely used analytical framework. The conventional BCA addresses the single objective of economic efficiency in allocating scarce resources, with the resource use option which provides the  88 maximum net present value being adopted in most cases (Hufschmidt 1982; Irland 1988; Tisdell 1990). Social welfare broadly construed, however, includes not only the economic efficiency criterion (i.e., Pareto improvement) but a range of extra-market effects as well. Recognizing that the central problem of incorporating environmental values into BCA results from the incommensurability of such values, economists have devised various approaches in an effort to measure extra-market values within an extended BCA. 25 Economic techniques for placing dollar values on natural amenities are often divided into two types: (1) economic surrogates based on preferences revealed through behavior, and (2) hypothetical techniques which measure demand through surveys (Hufschmidt 1982; Hufschmidt and Hyman 1982; Hufschmidt et al. 1983). Although alternative approaches 26 have been developed in response to many of the shortcomings of BCA, the remainder of this chapter and the following chapter will outline and critique three prevalent non-market value elicitation methods which were formulated and elaborated under the guise of an extended BCA: the travel cost method and hedonic price method, which are the focus of this chapter, and the contingent valuation method, which is covered in the next chapter. Also included in the next chapter is multiattribute utility 25 Farnworth  et al. (1981) identify a class of intangible wilderness values which have yet to be incorporated into any valuation system; the authors contend that their complexities preclude any meaningful assignment of value. They include such benefits as global carbon balance and atmospheric stability maintenance, life support systems, natural laboratories for the study of nutrient cycling, evolution, natural selection, etc. and support of indigenous cultures. 26 Among the most cited alternative approaches are the safe minimum standard (SMS) and an approach devised by the Resources for the Future people (RFF). SMS concentrates on the irreversible social costs of species extinctions. It adopts a modified version of the "minimax" criterion, which is the choosing of an alternative that minimizes the maximum possible loss of species, up until the point where the opportunity costs of protection are prohibitively high (Ciriacy-Wantrup 1968; Bishop 1978). RFF, on the other hand, uses the conceptual framework of BCA. Its adherents, however, recognize the elusiveness of many preservation benefits to quantification. RFF thus combines a detailed BCA with the presumption that the benefit-cost calculus underrepresents wilderness benefits. RFF attempts to redress this bias by giving preservation the benefit of the doubt, should doubts remain at the end of a BCA (Smith and Krutilla 1979; Fisher 1981).  89 technology; although forged under behavioral decision theory, it, like contingent valuation, measures preferences directly. The travel cost and hedonic price methods, on the other hand, elicit values indirectly. In the broadest, idealized sense, each method can be evaluated in terms of how accurately its derivative, assigned values express the underlying welfare or held values of the participants involved in a valuation exercise. However, demonstrating whether the true preferences of individuals have been accurately assessed or not cannot be achieved because held values lay hidden in evasive, abstract value systems; assigned values thus cannot be objectively validated with them. Also, for many reasons 27 all non-market elicitation methods, at most times, fail to provide results which accurately portray normatively-held values. Such a normative assessment, therefore, remains an extremely daunting task. Consequently, the various methods are assessed in more modest terms, where the strengths and weaknesses of each hinge on whether or not they provide contributory information to the decision-making process by furnishing usable results which approximate the collective values of those observed or involved. As such, their strengths and weaknesses are assessed on the basis of whether or not they provide normative guidance for decision makers in allocation decision contexts which invariably include such complicating factors as uncertainties, long time horizons, interrelated decisions, complicated institutional environments, poorly conceived and complex values and multiple stakeholder groups with conflicting objectives (McDaniels 1990). It must be kept in mind, however, that all the procedures outlined and critiqued below and in the following chapter share the tendency of providing a simplified and tentative snapshot of preferences and judgments. As new knowledge regarding wilderness benefits are "Brown (1984, 237) states that "The appropriateness of an assigned value for use in a resource allocation decision depends on the degree to which its use in the decision enhances the resource owners welfare." He goes on to say that most assigned values do not reflect accurately welfare because most individuals are ignorant of their held values and the objects they are valuing and because they often improperly identify the constituency they belong to: "The most appropriate assigned value results from perfect perception, strict correspondence of held values to welfare, and the lack of any constituency problems. Such conditions are unlikely to occur" (p. 238).  90 incorporated into an individual's value system, the underlying values (which form the basis for preferences and judgments) evolve and become more reflective of an individual's or a society's welfare. The relative worth of evaluated objects thus tends to shift in time as a person's perception of such objects becomes more refined (Brown 1984).  Indirect Non-Market Value Elicitation Techniques  In the absence of established market prices, it may be possible to estimate the value of a wilderness good or service indirectly by examining the price paid for a closely associated good or service that is traded in a market. In such cases, the value of these comparable commodities act as surrogates for their non-market counterparts. Because people's preferences are revealed through observation of their transactional behavior in markets, approaches based on surrogate prices are often referred to as revealed preference techniques. The most widely employed revealed preference measures of environmental extra-market values are based on travel costs and hedonic prices.  Travel Cost Method Procedure The most commonly used technique for inferring recreational demand is the travel cost method (TCM). 28 The basic TCM developed by Clawson and Knetsch (1966) proceeds from the postulation that a recreationist would react equally to an increase in an entrance fee as to an increase in travel cost. The travel costs associated with a trip to and from a recreation site can therefore be used as a proxy for price in the derivation of a recreational site demand curve 29 (Sorg and Loomis 1984). The fees for park use are in most cases free (or nominal) and thus 28For  thorough reviews of the TCM see Mendelsohn and Brown (1983), Rosenthal, Loomis and Peterson (1984), Anderson and Bishop (1986), Bowes and Krutilla (1989) and Forster (1989), as well as those listed under footnote 24. 29 The demand curve refers to quantity demanded (i.e., number of visits) as a function of price (i.e., added cost per visit).  91 much lower than what an individual would be willing to pay. The difference between actual and maximum willingness to pay (WTP) payments is the net benefit (or net WTP) derived from a site, and is known as consumer's surplus. The travel cost approach uses the information on the pattern of recreational use of a park to derive a demand curve to estimate the total amount of consumer's surplus (Dixon and Sherman 1990). The procedure begins with the collecting of data on the number of visits to a site and the origins of such visits. Visitor origins are then grouped into concentric zones 30 of varying distances from a site, with incurred travel costs increasing (and demand decreasing) as one moves from one zone to another away from the site. The visitation rate (visits per capita) and travel cost estimates to and from a site are calculated for each zone using the origin data (Forster 1989; Gunton 1991); the relationship between the two is a demand curve for the recreation experience (Hufschmidt and Hyman 1982). The demand for the recreation site itself is a derived relationship between the total number of visits and a hypothetical set of entrance fees. The visitation rate in conjunction with socioeconomic information for each origin is utilized in a regression analysis to statistically estimate a first stage site demand curve. Regression analysis involves coalescing visits per capita as a function of trip cost or distance and socioeconomic data (Sorg and Loomis 1984). More sophisticated models can also reflect the value of travel time and be expanded to include substitute sites and quality indexes (Knetsch and Davis 1965; Braden, Kolstad and Miltz 1991). Total visitation from all relevant origins reflects demand at current travel costs and represents one point in the second stage demand curve at the prevailing admission fee. That is, it defines the intersection between the second stage demand curve and the nominal (should fees exist) or zero (should fees be absent) price line (Freeman 1979a; Greenley, Walsh and Young 30The following approach is often called the zonal TCM (ZTCM) to differentiate it from the less common individual TCM (ITCM). In the ZTCM, recreationists are grouped into origin zones around a site, and a demand function is derived by estimating the statistical relationship between aggregate trips and travel costs from each zone. In contrast, the ITCM derives a demand function for an individual by estimating the statistical relationship between an individual's total trips in a given time period and distance travelled for such trips. Walsh (1986) and Bergstrom and Cordell (1991) elaborate on both.  92 1982; Hufschmidt et al. 1983; Forster 1989). The remainder of the second stage demand curve is generated by successively adding hypothetical entry fees to each origin's travel costs, in the assumption that visitors react to increased fees according to an estimated visitation response to increases in travel costs. Fees are increased until demand for park use, and thus the estimated number of trips by all individuals or from all distance zones, falls to zero. The per capita aggregate site demand curve is then used to estimate visits from each origin at each fee increment. The area under the resulting second stage demand curve plus any existent entry fees measure the recreation use value (i.e., total benefit or total WI?) attributed to the site. The difference in area between the total demand curve and fees represents the net benefit measure of total consumer's surplus (Greenley, Walsh and Young 1982; Walsh, Gillman and Loomis 1982; Hufschmidt et al. 1983). Applications The TCM was developed and applied for the valuation of recreation demand. Since its inception and period of development the TCM has been applied almost exclusively to outdoor recreation (Anderson and Bishop 1986). The procedure has been the preferred approach for estimating the benefits from recreation activities at specific sites, and has been found to work best for recreation areas of intermediate distances (100 to 150 miles) from the homes of most users (Walsh, Johnson and McKean 1990). The TCM can also be an acceptable way of estimating the recreation benefits which derive from specific activities or from changes in the quality of a site's resources (Rosenthal, Loomis and Peterson 1984; Ward and Loomis 1986). However, numerous stringent assumptions must be met and additional information taken before these latter computations can be successfully taken. As a result, the technique has been most widely used to infer a demand function for visits to a single-purpose, unitary site (Smith 1975; Bockstael and McConnell 1981; Mendelsohn and Brown 1983).  93 Strengths The main advantage of the TCM is that the secondary data required to calculate recreation demand is usually readily available. Consequently, such information as the number of visitors from each zone, length of stay, distance travelled and population per zone can be obtained relatively quickly and cheaply. Another strength of the TCM relates to its theoretical closeness to the market-based economic model. Because it uses actual consumer expenditures to impute non-market values it avoids some of the hypothetical value pitfalls attached to the contingent valuation method (Gunton 1991). Lastly, existing travel cost demand curves can normally be easily applied to a comparable site with minimal additional data collection. As such, developing a new travel cost model for each new site is not necessary in many cases (Dwyer, Kelly and Bowes 1977; Kaiser and Marchetta 1981). Weaknesses The literature cites numerous limitations and weaknesses in the use of the TCM. Principle among them are the difficulties in handling the cost of time, multi-destination trips, substitutes, congestion, quality changes and non-use values.  Time Cost A major empirical problem with the TCM is the difficulty of accurately determining the true costs of travel. Besides out-of-pocket travel expenses, Knetsch and Davis (1965), Cesario and Knetsch (1970), Walsh, Gillman and Loomis (1982), Bowes and Krutilla (1989), among others emphasize the importance of including the opportunity cost of time taken to travel to and from a site. Exclusion of time cost will result in an underestimate of site benefits (i.e., lower consumer surplus). However, due to numerous complexities 31 surrounding the travel time concept, the opportunity cost of time cannot be easily measured. Correcting for this bias is therefore difficult, and at this time no universally accepted formulation exists to accurately 31 See Cesario (1976), Bishop and Heberlein (1979), Hufschmidt et al. (1983) and Hyman and Stiftel (1988) for the problems in and possible means of incorporating the costs of travel time into the TCM.  94 account for it in a TCM framework (Smith, Desvousges and Fisher 1986; Gunton 1991). Multiple Destination/Purpose Trips Many recreationists visit more than one site when travelling. This is particularly true for long trips where numerous recreational sites exist between the zone of origin and the site under study (Hufschmidt et al. 1983). Allocation of the full costs associated with a multi-destination trip exclusively to the examined recreation site biases the measured demand for a site's benefits upwards (Haspel and Johnson 1982; Gunton 1991). The travel costs should under such circumstances be divided between the various sites visited 32 (Smith and Kopp 1980). Relatedly, multi-purpose trips (where more than one recreational activity is sought at a site) poses computational complexity and increases the likelihood of bias. Data collection is also problematic for sites which have widely separated, multiple entry points. In addition, the journey itself may provide travellers utility or disutility beyond those attributed to the costs of travel and time. When the underlying TCM assumption of pure visitation (site-specific, single purpose travel) is not met, much more extensive data collection 33 will be required to overcome these problems (Kaiser and Marchetta 1981; Hufschmidt et al. 1983). Substitutes Alternative sites with comparable attributes are often available to recreationists as substitutes. However, substitute sites are often ignored in travel cost evaluations, which tends to result in an overestimate of the benefits attached to the primary site (Swanson and Peterson 1988; Forster 1989). A more complex, data-intensive multi-site travel cost model must be employed in regions with complementary substitutes in order to provide more realistic benefit estimates. 34 32 Beardsley (1971) and Haspel and Johnson (1982) provide alternative ways of divvying up costs between two or more visited sites, with the former allocating costs in proportion to the relative amounts of time spent at each and the latter dividing costs evenly between each. 33 This may include in some cases visitor surveys. As Gunton (1991, 17) points out, however, "...if surveys are used, willingness to pay questions might as well be employed." 34 See Cesario and Knetsch (1976), Anderson and Bishop (1986), Rosenthal (1987) and Forster (1989) for examples of regional TCM's which incorporate substitution effects.  95 Site Congestion and Quality Increased density of recreation use may crowd a site to a point where the quality of the recreation experience diminishes. The conventional TC demand curve, however, implicitly assumes that recreational quality remains constant over the full range of use levels (Bishop and Heberlein 1979). Refinements in the model are necessary for evaluating the effects of congestion on site benefits (Newberry 1975). However, there is little agreement in the literature on how to account for the costs of congestion in the TCM and modelling its effects with aggregate data is not easily accomplished (Forster 1989). Other qualitative and quantitative attributes of a site are also difficult to model. For example, extracting information on site characteristics with multiple attributes adds increasing levels of difficulty and cost to an assessment. Although such adjustments to the simple TCM as the hedonic TCM 35 , devised by Mendelsohn and Brown, have been used to reveal users' WTP for individual recreational site characteristics, measurement difficulties persist. Aesthetic and spiritual values for example, although dominant features in many people's total recreational experience, remain extremely difficult to measure inferentially. Adjustments are also required to improve the TCM's application in the presence of qualitative and quantitative changes to a site (see Smith, Desvousges and Fisher 1986). A composite value representing all of the activities on a site may fail to appropriately reflect the different responses of each activity to changes in various site characteristics (Forster 1989). 36 Non-Use Values Because the TCM measures costs associated with actual visits to a site, it is limited to measuring use values only. It therefore can understate the value of a site by excluding non-user 35 In addition to Mendelsohn (1983) and Brown and Mendelsohn (1984), see Smith and  Kaoru (1987) and Mendelsohn and Markstrom (1988) for overviews and assessments of the hedonic TCM. Brown and Mendelsohn (1984, 427) describe in general terms how site characteristics are derived in the HTCM: "The prices of recreation attributes are estimated by regressing travel costs on the bundles of characteristics associated with each of several potential destination sites. The demand for site characteristics on site quality is then revealed by comparing the site selection of users facing different attribute prices." 36 Sorg and Loomis (1984) provide empirical estimates of various individual recreation activities using, along with other methods, the TCM.  96 benefits such as existence, bequest, option and ecological values (Swanson and Peterson 1988; Gunton 1991; McCollum and Bergstrom 1992). Other Weaknesses A number of other drawbacks have been cited for the TCM. Along with being an inappropriate approach for sites which heavily attract very distant travellers who make numerous stops, it has been found to underestimate demand for sites which are surrounded by • a dense population of proximate users. Although travel costs tend to be very low under such circumstances, people may highly value the natural amenities supplied by a recreation area; in fact, such amenities may have attracted people to locate in a scenic area in the first place, even if it means increasing their commuting costs to work. Also, a densely populated region which has residents with relatively uniform recreational travel expenses obstructs an analyst's ability of generating a demand curve because not enough data points can be induced (Hufschmidt and Hyman 1982; Hyman and Stiftel 1988). The assumption that people treat travel costs and entrance fees identically has been questioned in the literature as well. The effects of the multiple destination bias on this assumption has already been alluded to. In addition, Bishop and Heberlein (1979, 926-7) view travel costs as an inaccurate proxy of site fees in many cases, as indicated in the following quotation: Travel costs represent an aggregation of many smaller costs, some of which (e.g., tire wear) may not be obvious to the recreationist and which are not actually imposed on the recreationist at the time when recreation is demanded. Admission fees are paid immediately, usually in cash. Particularly in a world of satisficing, travel costs may not be perceived as equivalent to admission fees. Consequently, people often underestimate the costs of automobile travel (Common 1973). Income effects may also bias valuation results by undermining the travel cost proxy assumption. The TCM assumption that people in all distance zones monetarily value the same quantity of recreation equally was criticized by Seckler (1966), who argued that systematic  97 variations in incomes and utility functions between zones may exist. In such cases, individuals within groupings based on shared distances may have less in common in terms of tastes and preferences than individuals within similar income brackets (Pearse 1968). In the absence of systematic income differences between zones, socioeconomic heterogeneity within zones alone violates the uniform preference assumption. Although socioeconomic data have been included in numerous TC frameworks to help abridge the problems associated with these assumptive breaches, Hufschmidt et al. (1983) conclude that such inclusions have only met with varying degrees of success. Other fixed assumptions in the TC model which may be at odds with actual conditions are that (1) the singular purpose of the trip is for recreation 37 , (2) the amount of time spent at the site is the same for all users, regardless of distance travelled and (3) the identical mode of transportation is utilized for all trips (Hyman and Stiftel 1987). To improve the likelihood that these assumptions (along with the single-destination assumption) actually exist, Smith and Kopp (1980) recommend that a cut-off distance should be set for the furthest zone. As stated earlier, one distance range which has been suggested as optimal for many TC studies is 100 to 150 miles (Walsh, Johnson and McKean 1990). However, for many sites, travellers originate further afield, and setting an arbitrary distance limit will only prevent the preferences of such recreationists from being revealed in a TC exercise (Johnson 1980). Overall, the TC approach has many shortcomings which diminish its applicability in valuing wilderness extra-market benefits. In the case of non-use values, its failings are  complete in that the technique provides no assistance. As well, unless a site has only one, identifiable wilderness value (or a composite measurement is satisfactory) the method fails to provide realistic measures of individual use values within a multi-attribute area. In principle, many problems surrounding the valuation of recreation benefits can be overcome by expanding 37 0r  at the very least a recreation site must be a necessary part of the visit, or is, as economists call it, a "weak complement". Otherwise, transportation demand does not capture all of the demand for recreation (Braden, Kolstad and Miltz 1991).  98 the TC analysis beyond the traditional Clawson approach through increased levels of data collection on visitor preferences and itineraries. However, in practice such adjustments have not provided adequate scope to iron out the more serious flaws in the approach. The TCM remains on firm methodological footing only when the stated assumptions of human behavior and measured variables are met.  Hedonic Price Method Procedure Another prevalent indirect elicitation procedure is the hedonic price method (HPM). 38 The HPM infers the demand for non-market values by estimating implicit prices for individual attributes of a market commodity. The approach rests on the notion that some environmental goods and services are attributes of market commodities (Braden, Koistad and Miltz 1991). Property value is the most notable example of a commodity whose price is determined in part by the aggregated bundle of non-market goods and flow of non-market services which help to form the character of a property. The property value method uses real estate price differences between similar properties as a measure for the value of the one environmental quality attribute which differs between them (OECD 1989; Pearce, Markandya and Barbier 1989). Any changes in property prices can reflect quantitative or qualitative changes in any one of the property's characteristics, including in the encroaching environmental attributes. The property method can also capture these changes. It has been most widely applied in determining the value of ambient air quality in residential areas, but has also been used for valuing scenic vistas, noise, proximate natural areas, among other extra-market variables (Hufschmidt et al. 1983). Although most hedonic evaluations have used property values as surrogates for nonmarket good and service prices, the HPM can presumably be used for any good or factor of production which contains environmental attributes as one or more of its numerous characteristics. For example, Gunton and Vertinsky (1990, 15) illustrate in general terms the 38 Good sources for the procedure, use and assessment of the HPM can be found in Rosen  (1974), Harrison and Rubinfeld (1978b), Freeman (1979b), Nelson (1978), Anderson and Bishop (1986) and Bowes and Krutilla (1989), as well as those listed under footnote 24.  99 possible use of hedonic pricing for valuing the natural features of a recreation facility: The price of a good is a function of a bundle of attributes. The value of a recreation site, for example, is a function of a number of factors such as distance, facilities, availability of recreation activities, environmental quality and scenic value. The difference in attributes allows for the valuation of specific features by using statistical methods to assess the impact of each individual characteristic. For example, a site with good fishing can be compared to other sites which are similar in all respects other than the availability of good fishing. The difference in price can then be used to estimate the value of fishing. Economists use hedonic studies to ascertain the underlying demand for such characteristics by statistically analyzing the effects of them on the price of a good or factor (Palmquist 1991). The implicit prices of the individual attributes themselves reflect similar supply and demand forces as the observable commodity prices. Such implicit prices must therefore be tied to consumer tastes and preferences in order to generate an attribute demand function that can be used to make singular attribute value measurements (Braden, Kolstad and Miltz 1991). The demand curve can also be used to calculate the benefits or losses that occur as a result of marginal changes in the quality or supply of an environmental attribute (Hufschmidt et al. 1983). Applications The HPM was initially developed by Griliches (1961, 1971) to assess the impacts of  quality improvements in consumer goods on their value. The first study which attempted to use residential property values to infer environmental values was undertaken by Ridker (1967) and Ridker and Henning (1967). The property value method has since been used to estimate the effects of air quality (Harrison and Rubinfeld 1978a, 1978b; Nelson 1978; Freeman 1979a, 1982), water quality (Lind 1973; Brown and Pullakowski 1977), noise pollution (Pearce and Edwards 1979), aesthetic values (Armstrong 1974; Sinden and Worrell 1979) and improvements and damage to local amenities (Armstrong 1974; Polinsky and Shavell 1976; Abelson 1979; Bartik 1988) on property values. As previously mentioned, the HPM has also been hybridized with the TCM to estimate  100  the benefits and costs of congestion, quality differences and changes and other recreation site attributes (see footnote 35 for references). Strengths  The main strength of the HPM is that, like the TCM, actual market transaction data are used to estimate non-market values. Using observed market behavior eliminates any confusion between the intentions and actions of consumers because only actual transactions are studied (Braden, Kolstad and Miltz 1991); it is thus less prone to the systematic biases that may plague hypothetical valuation measures (Hufschmidt et al. 1983; Hyman and Stiftel 1987). Unlike the simple TCM, the HPM is specifically designed to evaluate environmental quality and attributes. Consequently, it is theoretically possible using the technique to infer demand for non-marketed commodities from markets with related commodities (Braden and Kolstad 1991). Weaknesses Hedonic approaches share some serious assumptive limitations with travel cost approaches. Firstly, the weak complementarity relationship must hold between the attribute and its associated commodity; that is, marginal utility can only be derived from an extra-market good if a certain amount of a market good is also consumed. Secondly, any changes which occur in an attribute are assumed to be fully absorbed in the price or quantity of its weak complement. This assumption does not hold for many complementary goods. Lastly, like any revealed preference approach, the HPM cannot measure non-use values (Braden, Kolstad and Miltz 1991). The approach also has some unique problems. One very serious difficulty with the procedure is that it relies on the assumption that sufficient information exists for the various variables which affect commodity prices. Because sufficient market data often are not available to identify all significant variables, bias can be introduced into the analysis when relevant variables are excluded and unreliable estimates can occur when irrelevant variables are included. Though obtaining precise data on the price of a distinct commodity is problematic,  101 data are particularly difficult to procure and measurements difficult to take for the myriad of variables that determine price. This is especially true for the environmental variables that need to be captured in the procedure. All in all, measurement and data elicitation impediments abound in the HPM. (Pearce, Markandya and Barbier 1989). Relatedly, the underlying price relationships have a diversity of possible formulations, each of which give different results. In most cases, the real relationship between the dependent variable (commodity price: e.g., housing) and independent variable (attribute affecting commodity price: e.g., pollution level) is not known, and clear guidelines do not exist for choosing the functional form statistically relating the two (Freeman 1979b; Gunton 1991). Isolating the effect of a singular attribute on the immediate value of a dependent variable is compounded in complexity by the expectations of future trends, which are inextricably capitalized into existing prices (Gunton 1991). Another statistical pitfall relates to the highly correlated nature of environmental attributes. In practice, disaggregating every attribute from an integrated whole in the effort of measuring and regressionally linking each to the dependent variable is not possible in most cases. To ease the complexity of this step, proxy or aggregate variables are normally used, possibly concealing the attribute of concern. The accuracy of the hedonic method's results is very sensitive to this exercise (Pearce, Markandya and Barbier 1989). This stage of the procedure also depends on the knowledge base of the consumers making up a hedonic study. Because the HPM can only capture the attributes that are known to commodity users, values of characteristics that exist but are not perceived (or at least not perceived in reference to the dependent variable) remain unmeasured. Insufficient information and imperfect perception on the part of the consumer violates the HPM's assumption of a perfectly functioning market and promotes inaccurate attribute assessments (Mendelsohn and Markstrom 1988; McCollum and Bergstrom 1992). Most of the literature on hedonic prices has been orientated towards using property  102 prices as an inferential measure of marginal changes in environmental quality, mostly air pollution. As a result, property values, though often inaccurate surrogates, are the most fully fleshed out hedonic prices in theoretical and practical use. However, wilderness is more often than not remote, and as such has little effect on residential housing markets. Other commodity prices must be used in such instances to reveal how much users are willing to pay for the individual characteristics of natural sites. Other than hedonic travel costs, the literature on hedonic approaches provides few examples of alternative surrogates which can derive meaningful measures of non-market wilderness values. These and other weaknesses limit the HPM's usefulness for determining the value of wilderness resources.  Summary  This chapter characterized both the nature of the overall tradeoffs which influence land allocation decisions and two indirect approaches which can be employed to elicit non-market values and thus inform such decisions. Devising workable and endurable tradeoff decisions involves resolving value conflicts through the enlightened comparison of all the costs and benefits arising out of alternative courses of action. Although economic efficiency within a benefit-cost framework is a viable choice criterion in decisions involving alternatives with clearly commensurable objectives, it cannot be used in wilderness decisions. This is because wilderness goods and services tend to be non-commensurable, non-rival, non-exclusive, indivisible and intangible and the decisions affecting them irreversible, thus violating many of the principle assumptions of perfectly functioning markets. Because many wilderness values are far removed from the ideals of the market model and because the forest industry was the traditional mainstay of the B.C. economy, the primacy of the timber resource in most forest allocation decisions remained unquestioned until the recent constriction in resource supplies and shift in public values. Wilderness values were  103 traditionally neglected in part because they lacked clearly defined market prices to signify their importance and as such were considered "soft". Various valuation techniques have since been devised to provide measures for nonmarket goods and services so that they, along with market benefits, are represented in benefitcost analyses. The travel cost and hedonic price methods have been advanced as two indirect techniques which can elicit some wilderness values. Both attempt to address the valuation problem by observing the revealed preferences of consumers in markets with related commodities, with travel costs being used as surrogate prices for recreation values within the TCM and hedonic prices being used as surrogate prices for extra-market attributes within the HPM. Although they each have strengths which allow them to address different facets of the valuation problem, serious shortcomings limit their use in valuing wilderness commodities, the most serious one being their inability to handle the numerous non-use values which have been ascribed to wilderness. Various alternative valuation approaches have been promoted to correct for this omission and other problems and stringent assumptions arising from revealed preference-based surrogate market methods. Two such approaches, which are based upon the direct elicitation of values from people or groups with stakes in a decision, are outlined in the next chapter.  104 Chapter V DIRECT APPROACHES TO ELICITING VALUES  Introduction  Another way of measuring the extra-market values of wilderness is to involve participants directly in either a hypothetically-derived market or public involvement process. Because people's preferences are actively sought out and explicitly expressed in these forums, approaches based on the direct elicitation of values are often referred to as expressed preference techniques. The expressed preference-based procedures which are outlined in this chapter are the contingent valuation method, which (like the TCM and HPM) is a tool within benefit-cost analysis, and the multiattribute utility technique, which stems from normative and behavioral decision theory. As in the preceding chapter, the procedure, applications, strengths and weaknesses of each will be discussed.  Direct Non-Market Value Elicitation Techniques  Contingent Valuation Method Procedures The contingent valuation method (CVM) 39 has been proposed as a popular alternative when the assumptions of revealed preference techniques cannot be met. The CVM allows people to provide direct expressions of preferences for non-market resources by asking them what they would be willing to pay for a benefit or what they would be willing to receive by way of compensation to tolerate a loss for a hypothetical market scenario. Benefits and costs are respectively presented as increments and decrements in the quantity or quality, or both, of 39Cummings,  Brookshire and Schulze (1986) and Mitchell and Carson (1989) are two volumes which offer detailed procedural outlines and assessments of the CVM.  105 targeted resources. In most cases, willingness to pay (WTP) questions are framed in a manner which elicits the maximum an individual is willing to pay for specified improvements or increases in an unpriced good or service. Less commonly, they can ask for an individual's maximum WTP to prevent a deterioration or reduction in an unpriced resource. Willingness to accept (WTA) questions, on the other hand, ask respondents to give the minimum amount of compensation that they would demand in order to accept a given deterioration or diminishment in an extra-market resource (Randall, Hoehn and Brookshire 1983; Forster 1989; Pearce, Markandya and Barbier 1989; McCollum and Bergstrom 1992). The contingent values are obtained either through a direct questionnaire/survey or through the use of experimental techniques in which subjects respond to a variety of controlled stimuli in laboratory conditions, with the former being the more prevalent approach. What are sought are the personal valuations of the respondents for specific levels of or changes in a good or service, expressed in WTP or WTA dollar amounts (Pearce, Markandya and Barbier 1989). The elicitation procedure is called the contingent valuation method because the elicited WTP (or WTA) values are contingent upon the specific hypothetical market described to the respondent (Mitchell and Carson 1989). In general, the value of the good or service is estimated by multiplying the average of these expressions of value by the number of consumers (Gunton 1991). Mitchell and Carson (1989, 3) describe a typical CVM, which generally consists of three parts: 1. A detailed description of the good(s) being valued and the hypothetical circumstances under which it is made available to the respondent. The researcher  constructs a model market in considerable detail, which is communicated to the respondent in the form of a scenario that is read by the interviewer during the course of the interview. The market is designed to be as plausible as possible. It describes the good to be valued, the baseline level of provision, the structure under which the good is to be provided, the range of available substitutes, and the method of payment. In order to trace out a demand curve for the good, respondents are usually asked to value several levels of provision. 2. Questions which elicit the respondents' willingness to pay for the good(s) being valued. These questions are designed to facilitate the valuation process without  106 themselves biasing the respondent's WTP amounts. 3. Questions about respondents' characteristics (for example, age, income), their preferences relevant to the good(s) being valued, and their use of the good(s). This information, some of which is usually elicited preceding and some following reading of the scenario, is used in regression equations to estimate a valuation function for the good. Successful estimations using variables which theory identifies as predictive of people's willingness to pay are partial evidence for reliability and validity. Several survey approaches have been employed in contingent valuation studies to elicit values from a subject (stage 2 above). Among the most common ones are the open-ended method, the iterative bidding method, the payment card method and the dichotomous choice method. Each of these are described and assessed in general terms below. A more extensive assessment of the advantages and disadvantages of the CVM will follow the discussion on its applications.  Open Ended Method -  The simplest format in which to obtain contingent value data is the open-ended method. Following a description of the product and means of payment (often referred to as payment vehicle), respondents are asked directly to give their maximum WTP or minimum WTA 40 for an environmental good or service. Its simplicity makes the approach amenable to mail surveys. Another advantage is that it avoids the response influences which are associated with the provision of starting bids or increments (Anderson and Bishop 1986; Forster 1989). A couple of criticisms have been laid against this approach. First, because very little information or stimuli is provided to encourage thorough consideration of a resource's value, respondents with little experience or knowledge will not be able to accurately estimate the value of such a resource. Consequently, respondents often have very little incentive in such hypothetical markets to devote serious effort in formulating correct responses. Second, with no specified prices to guide consumer choice, this method does not simulate market behavior very well. The net result is that people often find it difficult to answer questions of this type. People facing such difficulties may potentially 4To  simplify the following discussion WTP will be used to represent both WTP and WTA.  107 provide no response at all or give answers which are implausibly high or low (Carson 1991).  Iterative Bidding Method The iterative bidding game is the oldest and most frequently used CVM. The first stage, like all CV studies, provides a description of the non-market item and a hypothetical market in which the item is traded to a respondent. Iterative bidding begins with a suggested initial starting bid. If the respondent is willing to pay the starting bid, the interviewer suggests a higher bid. A series of revisions upwards takes place until the respondent is unwilling to pay. If the the initial bid is unacceptably high for the respondent, the interviewer progressively revises the bid downwards until an acceptable dollar figure is found. The accepted final bid is the measure of the respondent's maximum WTP for the item being evaluated (Sorg and Loomis 1985; Boyle and Bishop 1988). The main advantage of the iterative process is that it is said to help respondents to more fully contemplate and evaluate their preferences and thus presumably provides a relatively more realistic maximum WTP measure (Randall, Ives and Eastman 1974; Cummings, Brookshire and Schulze 1986). However, this advantage may be partly offset by the potential of the initial bid provided by the interviewer to influence the final bid provided by the respondent (Boyle, Bishop and Welsh 1985).  Payment Card Method A recent approach developed by Mitchell and Carson (1981) is the payment card method. After the non-market good and hypothetical market are described, a subject is asked to provide information on his income. He is then presented with a payment card which portrays a range of dollar values starting at zero and increasing at fixed intervals. The card corresponds to the respondent's household annual income and shows estimates (anchored figures) of what people at his income level paid, through taxes, for selected publicly provided goods. The subject is then asked to state, after considering the information on the card and his income level, a maximum amount he would pay for the non-market good in question. With no bidding  108 involved, the response is considered final (Boyle and Bishop 1988; Mitchell and Carson 1989; McCollum and Bergstrom 1992). Anchored payment cards were initially developed by Mitchell and Carson in an attempt to generate results not tainted by starting bid biases. Proponents argue that the approach accomplishes this promise while retaining all the advantages of iterative bidding (Mitchell and Carson 1981, 1989). However, questions remain as to whether or not the range of anchor point values provided on the card, along with the other information, prejudices responses to CV questions (Boyle and Bishop 1988).  Dichotomous Choice Method The first phase of the dichotomous choice method begins as all other CV exercises do (as described under the preceding methods). Respondents are then asked whether or not they are willing to pay some specified amount for a non-market good rather than do without it altogether. This take-it-or-leave-it or close-ended strategy does not permit respondents to state their own specific dollar values. With different respondents being given different dollar amounts to respond to, a series of rejection and acceptance subsamples are used in a discrete choice model. The model in turn estimates the probability of acceptance as a function of the stated price and other variables. The estimated probability distribution is then used to calculate the mean or median value of the elicited good or service (Bishop and Heberlein 1979; Loomis 1988; McCollum and Bergstrom 1992). Advantages of this approach are that it can be readily applied in a mail survey format, encourages responses and is free of starting point bias (Duffield and Patterson 1991). It also more closely resembles market transactions in which people either purchase or don't purchase goods at presented selling prices (Bishop and Heberlein 1986). Unlike the intricacies of anchored payment cards and the complexities of bidding formats, survey subjects only have to respond with a yes or a no to a presented price. Offering a conceptually simpler fixed amount to respond to avoids the problem of having subjects attach specific dollar values to monetarily  109 ambiguous environmental commodities (Boyle and Bishop 1988; Cameron 1991). The approach has some downfalls however. Foremost among them is the high level of statistical sophistication which is required to analyze qualitative responses (yes/no) to valuation questions. Because a larger number of observations are needed to achieve a level of statistical reliability matching those of other techniques, it is also more data intensive and expensive (Mitchell and Carson 1989). In addition, qualitative answers convey less information on respondents actual preferences than the former two approaches (Boyle and Bishop 1988; Carson 1991). Applications The CVM has been used to elicit values in a broad range of studies for a broad range of resource values. 41 Davis (1963) was the first to apply the approach when he used it to estimate the demand for outdoor recreation in the United States. It has since been employed to measure many recreational activities 42 including a total recreation experience (Menz and Mullen 1981), camping (Walsh, Aukerman and Milton 1980), fishing (Walsh, Aukerman and Milton 1980; Sorg et al. 1985; Cameron and James 1987) and hunting (Cocheba and Langford 1978; Brookshire, Randall and Stoll 1980; Sorg and Nelson 1986; Bishop and Heberlein 1986). In addition, the CVM has been employed in numerous environmental (Greenley, Walsh and Young 1981; Jackson 1983) and a few aesthetic (Randall, Ives and Eastman 1974; Brookshire, Ives and Schulze 1976) quality studies. The literature also cites CV exercises which have measured non-use values of wilderness. Included here are valuations of wilderness preservation 43 (Walsh and Gilliam 1982; Walsh, Loomis and Gillman 1984, 1985; Willis 1989; Walsh et al. 1990), water quality 41 Carson  (1991, 124-6) and Randall (1991, 221-5) provide literature reviews on numerous CV studies, with the latter also reviewing those which compare the CVM with the TCM and HPM. 42 Sorg and Loomis (1984) review numerous CVM and TCM studies on a diversity of recreation activities, including hunting, camping, fishing, hiking, picnicking and wilderness. 43 Both the wilderness and water quality preservation studies included measurements of option, existence, bequest and use values.  110 preservation (Greenley, Walsh and Young 1981, 1982), existence values" of wilderness (Bennett 1984) and wildlife species (Brookshire, Eubanks and Randall 1983; Stevens et al. 1991) and option values of wilderness (Barrick 1985; Barrick and Beazley 1990) and wildlife species (Brookshire, Eubanks and Randall 1983). It has further been used to measure the nonconsumptive use values and non-use values people derive from preserving such rare or endangered species as the northern spotted owl (Rubin, Helfand and Loomis 1991), the bald eagle (Boyle and Bishop 1987), the grizzly bear (Brookshire, Eubanks and Randall 1983) and the whooping crane (Stoll and Johnson). 45 Strengths The fundamental advantage of the CVM over revealed preference approaches is that it is not confined to evaluating non-market goods and services which have related surrogates in the marketplace. This imbues the technique with a relatively high degree of flexibility. Not being reliant on observed market behavior or availability of actual resource conditions, CV studies can estimate benefits under a wide range of hypothetical scenarios. It can thus be used to estimate the value of numerous attributes of interest, such as specific types of outdoor recreation and aesthetic benefits of the natural environment. Unlike revealed preference measures, the CVM can be used to measure people's expressed preferences for the non-use, preservation values associated with the mere existence of a wilderness area and the options which are generated from retaining such areas (Greenley, Walsh and Young 1982; Anderson and Bishop 1986; McCollum and Bergstrom 1992). The CVM's flexibility also permits it to be modified to incorporate alternative situations, such as different payment vehicles, market structures, tradeoff scenarios and qualitative and quantitative changes (Gunton 1991). Furthermore, it can be adapted to measure the values associated with the impacts of specific proposed policies (Swanson and Peterson 44Brookshire, Eubanks and Sorg (1986, 1987) describe and assess numerous CV studies on the existence values of wilderness and species. 45 Gregory, Mendelsohn and Moore (1989) describe and assess some of these (as well as other) CV studies on endangered species.  111 1988). Lastly, the CVM is said to be consistent with individual choice and welfare measurement theory (Smith 1987). Weaknesses Numerous potential biases and other problems have been cited that may undermine the validity (i.e., accuracy) and reliability (i.e., reproducibility) of the CVM. Distortions arising from biased responses to questionnaires and interviews have received particular scrutiny in the literature. Distorted expressions of preferences are said to produce biased results as opposed to mere confusion when deviations from statistical predictions are systematic rather than random (Gregory 1982). This section will provide an overview of many of these biases as well as other difficulties, beginning with those that are specifically correlated with the CV procedure and ending with those that are generally linked to the relationship between the cognitive processes of the respondents and the context set by the scenario and questions in the CV procedure. Included in this discussion are WTP versus WTA disparities, design biases (which stem from the characteristics of the questions asked), motivational biases (which are intentionally injected into the measurement procedure) and cognitive biases (which are unintentionally introduced distortions). First, however, the overriding issues of validity and reliability concerns are raised.  Testing Validity and Reliability As contingent valuations have become more refined, the literature has reflected an increasing concern over the limits of its application. Such concerns have focused attention on the overall validity and reliability of elicited contingent values and the means of testing for their existence. Validity studies for example are designed to determine whether the statistical expectation of an estimated mean is equal to the true mean of the item being evaluated; these test for accuracy in the elicited values. Reliability studies, on the other hand, are designed to ascertain whether the estimated contingent values are equivalent between successive  112 applications of the same CV procedure across time; these test for consistency, or reproducibility, of elicited values (Kealy, Montgomery and Dovidio 1990; Reiling et al. 1990). Predictive validity, which refers to how closely the contingent value of the good matches the value of the good were it to be traded in an actual market, is tested by comparing the WTP measure of the public good with the value of a highly correlated private good. However, in the case of public goods such markets generally do not exist, making such comparisons difficult, if not impossible (Kealy, Montgomery and Dovidio 1990). As an alternative to actual market comparisons, testing for validity is normally inferred by comparing contingent values with values procured through procedures based on the observation of actual behavior, such as simulated markets, the TCM or the HPM (Kealy, Dovidio and Rockel 1988; Reiling et al. 1990). Because such inferential procedures assess the convergence between the results of the CVM and indirect methods, the validity which is tested for is often referred to as convergent validity, and many examples of this type of research can be found in the literature. 46 These studies have determined that reasonable convergence occurs between the results of the CVM and indirect approaches. However, Cummings, Brookshire and Schulze (1986) and Mitchell and Carson (1989) caution that such results are only suggestive and not definitive reflections of validity. The reason that validation of contingent values cannot be assessed in most cases is that non-market valuation techniques provide an abstract measure of an individual's true preference, not a benchmark market value to which a more concrete comparison can be made. In other words, each technique provides an alternative measure of the same underlying construct (Mitchell and Carson 1989; Gunton 1991). As a result, a contingent value, which in social psychological terms is a measure of behavioral or attitudinal intention (Heberlein and Bishop 1986), is a hypothetical value which 'See Knetsch and Davis (1966), Bishop and Heberlein (1979), Sellar, Stoll and Chavas (1985), Smith, Desvousges and Fisher (1986) (TCM comparisons), Brookshire et al. (1982) (HPM comparisons), Bishop, Heberlein and Kealy (1983), Heberlein and Bishop (1986) and Kealy, Dovidio and Rockel (1988) (simulated market comparisons).  113 in most non-market circumstances cannot be compared, and thus validated, with the true preferences of respondents, which lie buried in often poorly delineated value systems. As Carson (1991, 136) points out, " applicable body of theory exists by which validity can be assessed because there are no explanatory models of the cognitive processes that underlie respondents' verbal reports." 47 Confidence in the validity (and reliability) of estimated values tend to be particularly fragile and tentative when ill-defined, unfamiliar goods are being evaluated. Convergence, therefore, will be weaker for less tangible, more amorphous goods and services (Cummings, Brookshire and Schulze 1986; Kealy, Montgomery and Dovidio 1990), such as for those within the categories of "human spiritual values" and "ecological values" covered in Chapter III. Reliability refers to the variation in the estimated contingent values due to random sources, or 'noise'. Mitchell and Carson (1989) identify three potential sources of variation in contingent values: (1) actual variation in values across the population being sampled; (2) variation resulting from the specific contingent valuation procedure employed; and (3) variation arising as a result of the small size of the population being sampled. Refining the design features of the CVM can be undertaken in an effort to reduce the imprecision which stems from the latter two sources of variation (Reiling et al. 1990). Most of the reliability tests which have been undertaken have evaluated the stability in responses over time and whether or not response fluidity, if it exits, is dictated by such potential influences as the nature of the good (Kealy, Montgomery and Dovidio 1990) or the seasonal timing of application of the survey (Reiling et al. 1990). Assessing whether or not the findings in a specific CV study (source (2) above) are consistent is normally accomplished through the employment of test-retest or multiple replication procedures, where the same sample of individuals respond to the same valuation questions at two or more distinct time periods. A sampling of the literature shows that a few studies have been undertaken in this manner to assess the reliability of CV measures across time. 48 47The  "Cognitive Biases" section elaborates more fully on cognitive-based difficulties.  114 Results from these studies suggest that the CVM does indeed provide statistically reliable estimates of value. However, like the validity studies, results from the reliability studies are suggestive only, and thus can not be generalized to any great extent. Reiling et al. (1990), for example, state that past efforts at assessing the reliability of hypothetical values lack generalizable utility because they either value a market good only (e.g., Kealy, Dovidio and Rockel 1988) or because their survey response rates are low (e.g., Loomis 1989). They recommend that future tests concentrate on the reliability of specific CV survey components and non-use values, both of which may influence to a considerable degree the reliability of the overall contingent value estimates. Until a greater number of tests like these are undertaken and their results verified with each other and other studies of similar goods, it cannot be determined whether CV estimates are reliable or not (Hyman and Stifle' 1988). As a result of disagreements over whether validity and reliability concerns are justified or not, it is not known at this point if Cummings, Brookshire and Schulze's (1986) conclusion that CVM estimates are accurate with plus or minus 50 percent of market value is a believable assertion or not. Although no definitive determination has yet been made on the consistency and accuracy of the CVM, the literature cites many potential difficulties with the approach that accent the concerns which have been raised over validity and reliability issues. It is to these weaknesses that we now turn our attention. Willingness to Pay I Willingness to Accept Discrepancy The traditional approach used by researchers to measure benefits is to estimate some form of an ordinary demand curve from which the Marshallian measure of consumer surplus 49 can be obtained. Although simply derived in that it avoids the problem of assigning appropriate property rights, such an avoidance makes it a non-reflective measure of an agent's welfare change. CV researchers are thus forced to make a decision on which Hicksian measure 48Kealy,  Dovidio and Rockel (1988), Loomis (1989), Kealy, Montgomery and Dovidio (1990) and Reiling et al. (1990) provide good examples of how reliability can be tested. 49 Consumer surplus in this context is defined as the area under the ordinary (Marshallian) demand curve and above the price line.  115 of consumer surplus, WTP or WTA, to use for a given welfare change. The question of property rights is the overriding determinant of which formulation to choose, in that WTP represents the buyers perspective and WTA represents the sellers (owners) perspective. Although property rights from both perspectives are more perceived than actually legal within a CV framework, their real or imagined influence on the views of respondents and their corresponding WTP or WTA answers have promoted a wealth of debate in the economic literature on which or either is the appropriate measure (Mitchell and Carson 1989). Established theory in welfare economics suggests that both WTP and WTA are commensurably viable measures of consumer surplus because they are approximately equal. Developed by Willig (1976) for price changes and Randall and Stoll (1980) for quantity changes, the stated theoretical assertion is that individuals are willing to pay the same amount of money for marginal increases in consumption as they would be willing to accept by way of compensation for an identical decrement in consumption. Providing income effects 50 are negligible and compensation entitlements do not notably increase real wealth, the theory should hold and the choice of measure should not be of significant concern (Willig 1976). In contrast with the theoretical axioms which predict only small WTP-WTA disparities, empirical evidence from a wide range of CV studies 51 show significant differences between WTP and WTA, with WTA far exceeding WTP in most cases. The consistency of the empirical findings, along with verification from simulated market studies (Bishop and Heberlein 1979, 1986; Bishop, Heberlein and Kealy 1983; Heberlein and Bishop 1986) and innovative laboratory experiments (Knetsch and Sinden 1984; Gregory 1986), strongly suggest that these differences are not methodological artefacts attributable to survey design flaws or the 50 lncome  effect refers to the impact that an expenditure on the good in question has on a household's total income, with a small income effect reflecting a small expenditure relative to total income. 51 Hammack and Brown (1974), Sinclair (1976), Banford, Knetsch and Mauser (1979/80), Bishop and Heberlein (1979), Brookshire, Randall and Stoll (1980), Rowe, d'Arge and Brookshire (1980), Knetsch and Sinden (1984) and Coursey, Hovis and Schulze (1987) give experimental results showing WTA exceeding WTP.  116 hypothetical nature of CV questions 52 , but real divergences (Mitchell and Carson 1989). Although rationales for why WTA measures tend to be much higher than WTP measures have yet to be thoroughly developed, a number of explanatory hypotheses have been put forth. Prospect theory (Kahneman and Tversky 1979) 53 provides the first hypothesis for this disparity. Unlike utility theory, which emphasizes final asset positions, prospect theory analyzes preferences based on gains or losses from a neutral reference point. According to the theory, the value function is steeper for losses than for gains, implying that a decrease in the quantity of a good away from a status quo position is valued at a higher rate than a corresponding increase. This tendency for humans to subjectively experience losses more forcefully than equal gains is known as loss aversion. The main implication 54 for CV studies from this psychological phenomenon is that individuals assuming the perceptual position of resource owners (in a WTA format) will place more worth on a resource that they currently posses than they would if they were buyers about to purchase the same resource (in a WTP format) (Brookshire and Coursey 1987). The buyer-seller disparity rationale is corroborated by dissonance theory, which posits that individuals are highly resistant to relinquish goods that they already own. It is also supported by the "endowment effect" formulation; it denotes that a good which forms part of an individual's endowment would be more highly valued than a good not held if the former (which is an out-of-pocket cost) is perceived as a direct loss and the money given up to acquire 52 Although early speculation also centered on whether the WTP-WTA discrepancy could be explained in terms of income effects alone, such a theory has since been rejected (Gordon and Knetsch 1979). 53 See also Tversky and Kahneman (1981) and Kahneman and Tversky (1982). 54 Prospect theory has other implications for CV studies. Among them is the psychological influence of decision problem framing on the preferences of respondents. It is posited that changes in expressed preferences, or even preference reversal (see Tversky, Slovic and Kahneman 1990 for a description of this phenomenon), can occur as a result of changes in how questions are framed (Kahneman and Tversky 1979, 1982; Tversky and Kahneman 1981). Context and framing effects provide a basis to many of the biases that may plague CV studies, and are discussed more fully within the "Contextual Bias" subsection under the "Cognitive Biases" section.  117 the latter (which is an opportunity cost) is perceived as a forgone gain (Gregory and McDaniels 1987). Another explanation offers the view that people are motivated to give higher WTA values because they reject the assignment of property rights which the WTA format conveys. Many respondents appear to regard WTA property rights as implausible or illegitimate or both, as indicated by the large number of protest answers (implausibly high bids or outright rejections) 55 which have been consistently received in CV studies using WTA questions (Mitchell and Carson 1989). However, Bishop and Heberlein (1979) and others have found that manifest rejection of the WTA format is less likely when real cash under simulated market conditions is offered. Nevertheless, it cannot be ruled out that protest bids represent actual registration of moral indignation and that scenarios proposing that highly prized public resources be sold off in market-like transactions may be highly offensive to some people. A related premise views the other side of the motivational forces which may underlie responses by concentrating on moral responsibility as opposed to moral indignation; namely, it looks at how the assignment of WTA property rights forces individuals to respond in ways which reflect the responsibilities they have been allocated. Higher WTA values could reflect the seriousness with which respondents assume such ownership roles, believing that it is their moral responsibility to make a decision which is in the interests of the greater public good (Harris and Brown 1992). In the case of decisions involving species protection, the assignment of moral responsibility could extend, in the minds of respondents, to the intrinsic interests of the species under their stewardship. These adopted perceptual perspectives may cause individuals to err on the side of caution, obliging them to invest a heightened worth in a resource to protect such interests. WTP measures, on the other hand, make a far less clear assignment of such obligations. As a result, framing effects caused by implicit property right differences when shifting between the two measures may contribute to the WTP-WTA disparity (Boyce et al. forthcoming). Harris and Brown (1992) caution, however, that 55 See Edwards and Anderson (1987) for evidence and significance of nonresponse bias.  118 respondents often feel discomfort with such responsibilities, and if given a choice will provide a WTP measure instead. Empirical data provided by them suggests that individuals' unwillingness to assume ownership status is particularly evident in CV studies involving environmental quality problems which can be tied to industry or others. In such circumstances, respondents often pin the obligations on polluters, believing it is they who should pay to resoNe such problems. These and other interpretive models 56 provide partial explanations for the empirical asymmetry between WTP and WTA. It is highly plausible that all of the various behavioral facets explained by each model exists, and that the WTP-WTA disparity can only be accounted for by a combination of these identified factors. In addition, theoretical work by Hanemann (1986) has also added a firmer basis to the assertion that the disparity can occur and yet still be consistent with standard economic theory. He has shown that the magnitude of the difference between the two measures is determined not only by income effects but substitution elasticity as well. Substitution elasticity refers to the ease with which other market goods can be substituted for the market good under study while keeping an individual's utility level constant. When carried over to the public sphere, the theory indicates that the fewer the substitutes available for the public good and the larger the income effect the greater the WTA-WTP disparity. This reveals that for more unique public goods (which by definition have low substitution elasticities), the divergence between the two measures can be considerable. Hanemann's theoretical relationships thus suggest that the property right chosen has immense implications for the magnitude and reliability of the valuation response, particularly for unique environmental goods and services which form a part of many of the wilderness values characterized in Chapter III. 56For a full account on this subject, see Hammack and Brown (1974), Gordon and Knetsch  (1979), Bishop and Heberlein (1979), Brookshire, Randall and Stoll (1980), Rowe, d'Arge and Brookshire (1980), Schulze, d'Arge and Brookshire (1981), Bishop, Heberlein and Kealy (1983), Knetsch and Sinden (1984), Cummings, Brookshire and Schulze (1986), Gregory (1986), Coursey, Hovis and Schulze (1987), Gregory and McDaniels (1987) or Mitchell and Carson (1989).  119 Although the above explanations provide some intuitive conjecture as to why the disparity persists and why the compensation measure commonly exceeds the payment measure, no systematically derived model exists which can provide clear guidance to help determine which measure is more representative of people's values and in which situations one is superior to the other. Acceptance of a difference between the two welfare measures thus forces CV researchers to make the complicated but influential decision of which to choose. The literature is not clear on this point however, with some analysts favoring one, others favoring the other and many stating that the choice is situational. For example, many CV specialists agree that the WTP format is appropriate in situations where increases in the provisional level of a resource are being assessed (Mitchell and Carson 1989). However, lesser agreement is generated in the case of whether to use the WTA measure in CV surveys assessing decrements. Although WTA may be an appropriate measure of welfare in evaluations of resource diminishments or deteriorations, there is growing consensus that reliable measurements of WTA cannot be easily made using a CV survey at this time. The rationale for the increasing lack of confidence in the reliability of WTA measures follows similar lines of logic which run through many of the above models regarding the perspective of the seller. "The problem in a contingent valuation market", as Carson (1991, 129/30) explains it, "is creating either a plausible situation in which the implicit agent who will purchase the good is likely to convey the money to the participant who can sell the good so that the seller's rational response is to set the price so high that the good will not be sold or a situation in which the purchaser has no choice but to purchase the good so that the seller's rational response is to ask for the highest feasible amount and not the minimum WTA." Again, unique goods will tend to show particularly high WTA responses because of lack of any substitutes to dilute their values. Because the two measures elicit very different responses and represent distinct forms of welfare measurement, the predicament facing researchers setting up a hypothetical market  120 involving resource depletions is that they cannot simply replace a WTA measure with a WTP measure in efforts of eliminating WTA overestimations. Mitchell and Carson (1989, 37) emphasize the difficulty of choosing an appropriate measure in the face of such obstacles: The result is that contingent valuation researchers continue to be faced with a dilemma: asking people to accept payment for a degradation in the quantity or quality of a public good simply does not work in a CV survey under many conditions, yet substituting a WTP format where theory specifies a WTA format may grossly bias the findings. This in turn poses a quandary, since researchers frequently wish to value quantities on both sides of the current level of provision of an amenity, and it is generally agreed that the correct measure for a decrease is the Hicksian compensating surplus WTA measure. They go on to recommend the use of the WTP measure for valuing decreases in the level of a large class of public goods that formerly were thought to require a WTA measure, while adding that carefully designed WTA CV studies can be successfully implemented in some limited situations, particularly if a referendum type format is adopted. Others concur with the notion that WTP measures are the preferred option under most circumstances (e.g., Cummings, Brookshire and Schulze 1986), while others prefer WTA measures for most cases involving resource diminishments (e.g., Knetsch 1984). Many more have adopted positions which lie between these two. 57 The significance of context in the shaping of preferences underlines the need to resolve differences in professional opinions on this point. The WTP-WTA choice is a powerful example of a contextual influence. Responses will depend to a large degree on the valuation perspective chosen and the manner in which questions are framed. Further study on the different values that people ascribe to gains and losses and how context alters perceptions of such relationships is needed and findings applied to measurement selection and CV design efforts. Until then, professional judgment, proposed hypotheses and preliminary experimental results will need be relied upon to aid in the WTP-WTA selection process. 57 Kahneman (1986, 188), for example, concedes that WTA measures of losses likely generate useless results, but cautions that WTP measures will seriously underestimate the value that people place on many goods in a compensation structure because they do not take into account the reality of loss aversion and, therefore, respondents genuine feelings of loss for resource diminishments or deteriorations.  121  Design Biases Design biases are used here to refer to systematic errors arising in WTP responses due to the design features of the CVM questionnaire. These biases relate to how the manner in which the questions are posed or the information is given provide implied cues to respondents which covertly influence the direction of their valuation answers. The main design biases stem from the initiated bid and range, the presented payment vehicle, the sequence or manner in which related goods are presented and, in general, the type of and way in which information is given.  Starting Point Bias Starting point bias arises when the initial bid introduced in an iterative bidding framework influences respondents fmal bids. A number of studies have tested for the existence of starting point bias by, in general, statistically examining the effects of alternative starting bids. While the results have been somewhat mixed and at times inconclusive, more studies have uncovered evidence of its potential presence (Rowe, d'Arge and Brookshire 1980; Boyle, Bishop and Welsh 1985; Mitchell and Carson 1985; Roberts, Thompson and Pawlyk 1985; Cummings, Brookshire and Schulze 1986) than not (Brookshire, Randall and Stoll 1980; Brookshire et al. 1981 58 ; Thayer 1981). After reviewing all the evidence, Cummings, Brookshire and Schulze (1986, 207) and Mitchell and Carson (1989, 241) conclude that starting point bias can present difficulties when employing bidding games. A few explanations have been advanced as to why an initial bid has the potential to influence a respondents final value. One possible source may arise from the suggestiveness of the introductory bid; the bid may convey to the respondent the approximate range of an appropriate final bid, impelling a respondent to anchor his WTP around the suggested final value. Secondly, if the starting bid is significantly different from a respondent's actual WTP 58Mitchell  and Carson (1989, 241) reject Brookshire, Randall and Stoll's (1980) and Brookshire et al.'s (1981) conclusions that no starting point bias is evident because the tests "have no power to statistically detect the very large difference they observed." Consequently, they view Thayer's (1981) study as the only convincing test to show its absence.  122 and the respondent values his time highly, he may become bored or irritated with a lengthy iterative process and settle on a bid before his true preference is expressed (Schulze, d'Arge and Brookshire 1981; Boyle, Bishop and Welsh 1985; Cummings, Brookshire and Schulze 1986). As a result, starting bids which are well above a subject's true WTP may bias the final bid upwards and starting bids which are well below a subject's true WTP may bias the final bid downwards; this precise pattern occurred in a study run by Roberts, Thompson and Pawlyk (1985). The influence of the starting bid is probably accentuated when respondents have poorly defined or developed values regarding a good; respondents, in such cases, tend to be more highly impressionable and agreeable, using the starting bid as an evocative source of information about what their true values should be. The tendency of uncertain respondents to make only modest adjustments away from the starting bid and thus be prone to anchor onto a value which is biased towards the initial value is supported by many in the behavioral science field (e.g., Tversky and Kahneman 1988; Slovic, Fischhoff and Lichtenstein 1988). The net effect from these contributory sources of starting point bias may be to undermine the accuracy of the CV survey (Schulze, d'Arge and Brookshire 1981). Moreover, Mitchell and Carson (1989) and Carson (1991) contend that there is no acceptable way of compensating for the effect of the starting bid in bidding games. They suggest, however, that in general the more plausible and realistic the situation, the easier it will be for respondents to provide accurate answers. 59 They also recommend the use of payment cards as an alternative elicitation technique. However, as is revealed next, the range of values presented in a payment card format has been implicated in bias as well.  59 As  Carson (1991, 137) explains, "The uncertainty induced by implausible scenarios promotes bias because the respondents are susceptible to treating supposedly neutral elements of the scenario, such as starting points, as clues to what the value of the amenity should be." Lack of realism and plausibility, therefore, promote many other CV biases as well.  123 Range Bias  As expressed earlier, the vulnerability of the iterative bidding procedure to starting point bias prompted Mitchell and Carson (1981) to develop the payment card method. While the range of values on a payment card reduces, if not eliminates, starting point bias, the featured scales may influence the magnitude of a respondent's WTP. Range bias occurs when the respondent believes that the value range on the payment card reflects the true distribution of values, inducing him to use them as a frame of reference to estimate and evaluate his preferences (Mitchell and Carson 1989). Except for Schwarz et al. (1985), who found a bias of this kind, and Mitchell and Carson (1981), who did not, few systematic studies have been done to verify or dismiss the significance of this bias. While empirical evidence is lacking, Mitchell and Carson (1989) identify three potential sources of range bias which could stem from the design configurations of a payment card: (1) the upper most value on the scale may be lower than a respondent's maximum WTP, thus constraining his or her expressed preference; (2) the upper value may impel a respondent to believe that it is a reasonable upper bound, leading him or her to give a higher amount than would have been given had the upper value been lower; and (3) the range may not encompass the amount that a respondent is willing to pay, inducing the respondent to choose a WTP amount that is either too high or too low. A related behavioral response may be added: a respondent with poorly defined preferences regarding the good in question may perceive the range, particularly if it is evenly spaced, as a normal distribution of values and thus be inclined to choose a middle value. Mitchell and Carson (1989) also relate how gaps between numbers, particularly if large, in the area of interest produces bias because people tend to choose either values on the list or values which are multiples of 5's or 10's. It is not known at this point how significant range bias is. A properly designed payment card which takes into account these various facets of behavioral influences may reduce the incidence and severity of such problems. For example, sources of the first kind may be  124 avoided by setting a sufficiently large upper bound (although this may stimulate influences of the second form). However, before widespread adoption of the payment card method occurs, further research should be done to assess its accuracy and the means of refining or reconfiguring its design characteristics to improve its accuracy.  Payment Vehicle Bias This form of bias occurs when the method of payment specified in a survey generates a reaction in a respondent which translates into a distorted response. For example, individuals generally have an aversion to taxes; subjects might as a result understate their WTP for a good if its mode of financing is to be made through increased taxes. On the other hand, an entrance fee, if the non-market good is recreation, may produce a less strident reaction (Cummings, Brookshire and Schulze 1986; Gunton 1991). Besides emotional reactions to payment methods potentially biasing results, familiarity with payment methods may also bias results. An example of this would be when hunters use the prevailing price of hunting permits, with which they are familiar, as a benchmark for their WTP calculations in a survey employing hunting licence fees as the payment vehicle. Consequently, instead of expressing a maximum WTP amount which resembles their preferences they may express a WTP amount which they regard as a reasonable price to pay for permits of this nature. 60 The payment schedule may also influence the magnitude of the WTP answer. People may, for example, be more comfortable (i.e., the payments may seem less 60Mitchell  and Carson (1989, 198) offer this line of reasoning as a possible explanation for Bishop and Heberlein's (1979) low WTP results for goose hunting permits. Other rationales are also forwarded (e.g., strategic bias). Such an effect also occurred in a study conducted by Sorg and Brookshire (1984) using an open-ended format to value the elk resource; half the respondents in this study gave the current cost of an elk hunting licence. Though this may be counter-intuitive to the notion that a presented scenario should be as realistic and familiar as possible to avoid biases, Mitchell and Carson (1989, 216) note that scenarios such as these can induce biases because they involve quasi-private goods which have prevailing prices to which respondents' are tempted to anchor WTP values. They also state that it may be compounded by providing too much information; in such situations, respondents may react to the information overload by ignoring important information and focussing excessive attention on less relevant, albeit familiar and persuasive, information. However, in general the position forwarded in footnote 59 holds.  125 onerous) making a series of payments (say, $/annum over next 5 years) rather than a lump sum payment, thereby arriving at a larger cumulative WTP for the former. Numerous researchers have compared elicitation results from alternative payment methods to detect for the presence of vehicle bias (Brookshire, Ives and Schulze 1977; Rowe, d'Arge and Brookshire 1980; Brookshire et al. 1981; Daubert and Young 1981; Greenley, Walsh and Young 1981). Except for Brookshire et al. (1981), whose results were inconclusive, all of the studies listed here empirically confirm that payment methods influence valuation. Moreover, the literature largely supports the vehicle bias proposition (Randall, Ives and Eastman 1974; Schulze, d'Arge and Brookshire 1981; Rowe and Chestnut 1983; Cummings, Brookshire and Schulze 1986). The literature, however, provides little in the way of guidance in how to minimize or eliminate it. Alternatively, Arrow (1986) and Kahneman (1986) point out that efforts at designing an unbiased payment vehicle are largely wasted because the method of payment forms an integral part of the evaluation; in other words, the evaluated commodity and the payment procedure cannot be separated. Hence, they maintain the assertion that there is nothing irrational about individuals responding differently to different payment vehicles; in fact, it is completely rational to have a different set of preferences for taxes than for user fees than for utility bills. Arrow (1986) concludes that WTP depends on the structure of the means of payment and that such a relationship is reasonable in most cases. 61 Based upon the strength of such arguments, Cummings, Brookshire and Schulze (1986) concur that the impact of the means of payment on the respondent's elicited value is not a problem. They add, in agreement with Mitchell and Carson's (1989) and Carson's (1990) emphasis on the need to build into scenarios realism and plausibility, that the mode of payment should be selected on the basis of what payment vehicle would realistically be employed were the non-market good provided. 61 However,  he adds the following caveat (p. 183): "Now for others, it may not be. You can get the framing problem. Say you get two methods of payment where every individual in fact is paying the same amount, or at least his or her random expected payment is about the same. Then if the responses differ, you may have a real vehicle bias."  126 Relational Biases  Relational biases are used here to signify a class of biases which spring from the linkages between the evaluated good and other related or interrelated goods. These biases are said to occur only when such relationships influence a respondent's formulation of his or her WTP in a way unintended by the researcher. A related good can give off distortionengendering cues by acting as a comparative benchmark, by functioning as a neighboring good within a sequence of goods and by serving as a larger good within which the evaluated good is embedded. Benchmarks Relational bias can occur when benchmark amounts listed on payment cards inadvertently prejudice respondents opinions on the good(s) being evaluated. As Mitchell and Carson (1989) point out, the benchmark values of the reference goods are supplied with the intention of informing respondents about the existence and magnitude of taxes which they are already paying for many public goods and prices which they are paying for regulated goods. However, impressionable subjects may rely on such prices to infer their WTP rather than constructing them from their own preferences. Mitchell and Carson (1981, 1984) tested for this kind of relational bias in their national water quality studies by systematically varying the dollar levels for the non-environmental benchmarks (e.g., police, fire protection, roads, highways, etc.) and the number of benchmarks on the payment card. They found no undue influence from these benchmark factors on respondents' WTP results in their study, but concede that potential still exists in other situations. Because of the threat of benchmark-based relational bias, Mitchell and Carson (1989) recommend that reference goods be chosen which are not directly related to the non-market good being valued. Also, to avoid inciting WTP responses which are framed only within the context of controversial items rather than within the larger framework which includes all items on the card, they warn that benchmarks should not be controversial or evocative. With few  127 studies having tested for it, it cannot be said how significant bias from benchmarks are. Sequencing The sequential evaluation of decision alternatives has been found to produce different preference expressions than simultaneous evaluations (Tversky 1969). Also, the position of an item in a sequence of valuation questions has been found to generate variable responses, with a higher WTP value commonly occurring for an item if it is first in the list rather than following others (Plott and Levine 1978). Tolley and Randall (1983), for example, found that the value of improved visibility in the Grand Canyon differed by a factor of three depending on whether this item appeared before or after questions dealing with air quality in the respondent's own city. Brookshire et al. (1981) also found a sequencing effect in their study on people's valuations of air quality-induced aesthetic and health effects. Although the severity of sequence bias is unknown at this time, its significance should not be understated. Kahneman and Knetsch (1992, 58) add that "Because the order in which goods are mentioned in a survey is purely arbitrary, any effect of this variable raises questions about the validity of responses? Embedding The embedding effect, which has been raised as a possible bias in the CVM by Kahneman and Knetsch (1992), refers to the difference between the value of a good when assessed on its own and the value of the same good when assessed in conjunction with other related goods. They found, as reported in Kahneman (1986), that Toronto residents' WTP was only slightly higher to prevent fish population losses in the entire province of Ontario than their WTP to preserve fish stocks in only a small area of the province. Based upon the results from this study, and others reported in Kahneman and Knetsch (1992), they conclude that the value of a good when inferred from the WTP for an inclusive good in which it is a component part will tend to be less than if it is evaluated on its own. They also conclude that although the effect is probably more pronounced when eliciting non-use values, use values are not immune to its influence.  128 They attribute the embedding effect to the fact that respondents perceive non-market goods in broader terms than previously ascribed; that is, they are viewed in terms of the sense of moral satisfaction that people receive when contributing to the provision of such goods, rather than the goods per se. Claiming that such a general hypothesis is consistent with economic analyses of altruistic behavior, Kahneman and Knetsch assert that moral satisfaction stemming from contributions to an inclusive cause can also be given to a subset of that cause with little loss in value. They add that all public goods conceptually far removed from the buying and selling functions of markets, whether they have use or non-use value, are susceptible to this effect because purchases contributing to such amenities as park expansions or species preservation have more in common with charity than with the purchase of consumption goods. As a result, the WTP value, like the WTA measure, may also be registering another form of moral concern (namely moral satisfaction), making inadequate the interpretation that WTP for the protection of public goods is equivalent to consumption good purchases. While conceding that embedding effects are not likely to be an ubiquitous presence in all CV studies, Kahneman and Knetsch (1992) point out that when it does occur it is not clear which measure (aggregate or disaggregate?) is the appropriate one. Principles that could guide the choice of embedding level are lacking in the literature, and they question researchers' abilities in finding one. Because the embedding level is arbitrarily chosen and because there is the possibility that peoples' true WTP measures are further concealed by their purchase of moral satisfaction, Kahneman and Knetsch conclude that the CVM is a very inaccurate technique for estimating economic values. Gregory et al. (forthcoming), while supporting its plausible presence, propose that the embedding effect's fundamental cause does not stem from expressed preferences not matching those being sought by CV analysts due to the implicit addition of unintended assumptive details (i.e., moral satisfaction gained from contributions) on the part of respondents. Instead, it is a measurement failure stemming from respondents' lack of clearly-defined monetary  129 representations of many non-market values, particularly in unfamiliar contexts. This clearly fits in with Kahneman and Knetsch's view that the factor that controls the severity of the embedding effect is the degree to which a good can be conceived of in purchasing terms. Another possible explanatory cause of embedding linked to measurement failure is the insensitivity of many participants to the precise meanings of the questions asked, possibly because relevant details are excluded (contributing to uninformed judgments), possibly because too many details are included (contributing to cognitive overload) (Fischhoff et al. 1993). Mitchell and Carson (1989) also looked at the notion of aggregation problems by reviewing the literature for the significance of "symbolic bias" and "part-whole bias". 62 Based upon their examinations, they came to the conclusion that such problems can bias results. However, they are more hopeful than Kahneman and Knetsch that such biases can be minimized through disaggregation strategies involving improved wording and clearer descriptions of the goods under evaluation. Fischhoff et al. (1993, 230), whose study found that a cognitively simpler paired comparison approach generated less of an embedding problem than a more demanding experimental design based on direct dollar estimates, support Mitchell and Carson's emphasis on the need for clarity to offset such effects when they state that there is a "need to ensure that subjects have heard, interpreted, and accepted a task's details as intended." They go on to stress, in reference to the level of detail to include and complexity of task to accept, the probable importance of infusing scenarios with realism and structural coherency: "One determining factor may be whether the full task creates a coherent whole, whose features can be chunked into smaller sensible units. That seems more likely when the CV scenario is derived from an actual problem rather than being composed of arbitrary details, patched around a focal good." 62Symbolic bias occurs when respondents base their valuations on the general symbolic  meaning of the amenity and not its specific level of provision. Part-whole bias, on the other hand, occurs when respondents are unable to differentiate between an amenity's provisional level and a larger geographic area, a larger aggregation of entities, a larger policy framework, etc. in which it is embedded (see Mitchell and Carson 1989, 249-52).  130 Although debate is still mixed and conclusions are premature on the question of the embedding effect, it has raised serious concerns about the ability of the CVM to uncover people's true preferences. Whether it be charitable inclinations, tenuous monetary representations, insensitivity to questions framed or elements of all three which are at the root of this bias, the commonality between them, namely that people's representations of nonmarket values are often poorly-defined and difficult to retrieve, seriously undermines confidence in the CVM, as traditionally designed. Mitchell and Carson's (1989) and Fischhoff et al.'s (1993) suggestions may point the way towards refinements in the CV design. However, some of the more conceptually elusive and indivisible wilderness values, such as non-use, aesthetic, cultural and ecological, may be far more intricate and entangled than even the most ingeniously designed contextual frames devised within a CV study could handle. The notion that values are multidimensional and that people thus cognitively define values in more complicated terms than can be captured by a unidimensional expression of value is being advanced by Gregory and McDaniels (1987) and Gregory, MacGregor and Lichtenstein (1992). As such, the true worth of these wilderness values could remain hidden, and the representative value of WTP could be orders of magnitude different. Closer monitoring of the significance and greater debate over the causes of embedding, as well as other effects related to the aggregation and multidimensionality of values, may shed more light on these recent challenges. Information Bias  Information bias has been generally regarded in the literature in excessively broad terms. Rowe, d'Arge and Brookshire (1980, 6), for example, define it as "a potential set of biases induced by the test instrument, interviewee, or process, and their effects on the individual's responses." By including many sub-categories of biases (many of which have already been described), such a conception fails to be specific enough to provide guiding principles for research efforts looking into its significance. Mitchell and Carson (1989) also  131 point out that, like the payment vehicle, information contained in a survey is a legitimate and inextricable part of the valuation exercise. Rational subjects should, therefore, vary their responses according to differences in the information provided. However, because the information provided has the potential to alter people's reported WTP values, genuine concerns are being raised over such issues as which information to disclose, how to disclose it and in what order it should be disclosed (Schulze, d'Arge and Brookshire 1981; Rowe and Chestnut 1983). If facets of information disclosure such as these produce varied valuations 63 when changed, the fundamental question is what value is closest to the respondents true preferences? It is of course impossible to answer this question because as we discovered earlier with validity issues, no understood abstract value exists to compare WTP values to. Also, because the detail, characteristics and quality of the information provided vary for each situation in a random rather than systematic way, "information bias" is a misnomer in that it affects the reliability rather than the validity of the contingent values. Nevertheless, this problem parallels the previous biases in that it belongs to the larger issue of changing contexts, and their influences on preferences. Gregory et al. (forthcoming), after studying the research looking into the nature of human preferences, conclude, for example, "that people are not just reporting their values or preferences. Instead, they are constructing them, with whatever help or cues the circumstances provide." This constructiveness view of human values implies that the quantity and quality of information provided may not only influence the magnitude of a person's WTP but his or her underlying preferences as well. If true, such a notion has significant design implications for the CVM as well as other expressed preference elicitation methods. As Gregory et al. (forthcoming) point out, "...value formation is intimately tied to the specifics of the elicitation procedure." 64 63 Little  in the way of strong empirical data exists to support the information bias proposition. For example, of the studies conducted by Rowe, d'Arge and Brookshire (1980), Brookshire et al. (1981), Thayer (1981) and Samples, Dixon and Gowan (1986) only Rowe et al. discovered evidence of it. 64 They propose as a solution to the inadequacies inherent in the holistic design of the CVM a hybrid approach, which integrates the CVM with the multiattribute utility technique.  132 Motivational Biases  Gregory (1982, 55) defines motivational biases as "intentionally introduced distortions of what the respondents feels are his or her true preferences? Misrepresented responses more often than not result from incentives or disincentives which are built into the interview situation in some way. It must be kept in mind, however, that although motivational biases have been raised as a set of possible problems and probably represent true distortions in some situations, a paucity of tests have been done on them; they thus by and large represent the intuitions of CV theorists. Three major classes of motivational biases are described here: interaction, strategic and hypothetical. 1met-action Bias -  This form of motivational bias is rooted in the dynamics of the interaction between the interviewer and respondent. Personality conflicts are examples of where such a bias can originate. For example, an arrogant, boring or discourteous interviewer may frustrate, tire or anger a respondent, inciting the respondent to retaliate by purposefully misstating his or her preferences or by making extreme judgments (Gregory 1982). Oppositely, respondents may engage in a form of compliance bias (Mitchell and Carson 1989) by attempting to give answers which please the interviewer or sponsor. Distorted responses of this kind may be particularly embellished when the respondent believes the interviewer represents a particular position on an issue (Gregory 1982), the respondent is not abreast of the issues raised in the interview (Schuman and Presser 1981) and/or the interviewer implicitly or explicitly sends out signals which, whether intentional or not, sways the respondent's answers in a particular direction (Gregory 1982). Evidence is mounting in the psychological literature that people are easily influenced by others and accept other, more influential persons' attitudes often without critical analysis (Harris, Driver and McLaughlin 1989). Research indicates that people easily accept presuppositions as facts (Loftus, Miller and Burns 1978), confuse inferences with direct  133 observations (Harris and Monaco 1978) and put greater degrees of belief into statements if they are repeated again and again (Hasher, Goldstein and Toppino 1977). Harris, Driver and McLaughlin (1989, 221), after reviewing such evidence, state: "Consequently, the normative signals given the respondent by interviewers or questionnaires applying the CVM may be particularly problematic, given the novel and potentially complex judgment problem that this method poses." Strategic Bias  Strategic bias occurs when respondents deliberately provide misleading answers in their attempts of influencing the study's outcome in a way which serves their own interests. It is proposed that the hypothetical structure of the CV setting may provide motivational incentives to act strategically, with the strategic posture assumed varying according to how the respondent perceives the structure of the contingent market. For example, if a respondent who desires a river to be cleaned up believes that he or she will be personally responsible for paying an amount in taxes dedicated to cleanup which is equivalent to his or her specified WTP amount, an incentive exists for the subject to free ride and understate his or her own true values, hoping that others will bid enough to have the program implemented. Alternatively, the average of all WTP amounts is taken as the tax level which all respondents must pay to produce a desired outcome, two major strategic postures could be taken. First, incentive to understate a true preference exists if a respondent believes that the average WTP level of all others is above his or her actual WTP. Second, if a respondent believes that implementation of the desired policy will not occur without a sufficient WTP amount given by him or her, incentive exists for the subject to provide a bid which is high enough to permit implementation but low enough to prevent the average WTP from slipping above the subject's actual WTP. Another example of a free-rider problem is when a respondent living in a specific region suspects that other nearby residents will bear control costs to clean up an airshed polluted by a power plant; a respondent who desires air quality improvements would  134 have an incentive in such a situation to overstate his or her maximum WTP (Schulze, d'Arge and Brookshire 1981; Hufschmidt et al. 1983; Randall, Hoehn and Brookshire 1983; Rowe and Chestnut 1983; Anderson and Bishop 1986). The extent of strategic bias depends on a number of factors. First, respondents must suspect that the study results will have an effect on program or policy implementation. In other words, no strategic incentive exists if respondents believe or know that the survey results will not influence a policy or an outcome (Schulze, d'Arge and Brookshire 1981). Second, respondents must believe that their responses will produce, or at least not hinder, a desired collective outcome while not being detrimental to their own personal cause. Third, they must be willing to be dishonest and understand their preferences sufficiently to provide a strategically dishonest answer (Gregory 1982). However, most hypothetical market scenarios are devised to elicit information which in most cases is not directly applied to policies or programs. Respondents often suspect this, thereby undermining the first assumption. Moreover, respondents normally have a very poor conception of the hypothetical mean and in general lack sufficient information to act strategically, contravening the second assumption (Schulze, d'Arge and Brookshire 1981; Randall, Hoehn and Brookshire 1983). Also, as we have seen, even if respondents are willing to provide dishonest answers, poorly developed values will make tenuous the third assumption. Much of the empirical evidence supports the notion that there is little scope for strategic behavior and in general has found little influence from strategic bias. 65 However, although the evidence is encouraging it is not definitive. Because such studies can neither confirm or deny its universal presence, strategic behavior cannot be ruled out in every situation. For example, 65 Bohm (1972), Scherr and Babb (1975) and Smith (1977b) are examples of studies which found no evidence of strategic bias. After reviewing these and other studies Mitchell and Carson (1989, 170) observe that although not conclusive, "the evidence...suggests that strategic bias is not a significant problem for CV studies under most conditions. Instead of being a fundamental, unavoidable threat to the CV method, strategic behavior is just one of many possible sources of bias which the designer of a CV study must take into account." Schulze, d'Arge and Brookshire (1981) and Cummings, Brookshire and Schulze (1986) also reviewed the findings and came to similar conclusions.  135 individuals may partake in strategic bias in polarized decision environments if they believe the results, if not affecting public policy, will at least have an influence on public opinion. In situations related to harvesting-wilderness tradeoff decisions, for example, forest industry personnel will have incentives to understate their wilderness valuations while environmentalists will tend to overstate theirs (Gunton 1991). So, while strategic bias is probably the exception rather than the rule, absence of a basis for categorical denial of the problem may lead some researchers to the recognition that they should at least consider it when designing and applying CV surveys in decision environments which encourage misleading answers. 66 However, Mitchell and Carson (1989, 170) point out that "The potential threat posed by respondents deliberately giving untruthful WTP values is likely to be much less serious than the possibility that they will give meaningless values." Hypothetical Bias  Even if incentives to manipulate responses strategically are minimal, the hypothetical nature of CV markets may provide disincentives to answer accurately. Hypothetical bias has been proposed as a validity problem which stems from the artificiality of the CV framework and the manner in which people speculate on their behavior in hypothetical market situations. It has been widely suggested that because people do not invest the same level of contemplation, research, time and care in arriving at a WTP amount as they would were they pricing and purchasing items in an actual market, they lack the same incentives required to give accurate responses (Bishop and Heberlein 1979; Freeman 1979a; Bishop, Heberlein and Kealy 1983; Rowe and Chestnut 1983). In other words, contingent values, which are measures of behavioral intentions or attitudes, may differ significantly from market values, which are measures of buying and selling behavior, because the former is more speculative and labile than the latter (Heberlein and Bishop 1986). The dissimilarity between behavior in actual and hypothetical market settings and the implications from such a disparity are illustrated by 66 See  Mitchell and Carson (1989, 162-5) for conditions that promote strategic behavior in CV studies.  136 Bishop, Heberlein and Kealy (1983, 627) in the following manner: Prior to being confronted by an interviewer or mail survey, subjects may never before have attempted to express how they feel about environmental assets in monetary terms. While constrained utility maximization is a useful construct, conversion of utility into monetary terms in the real world may involve repeated market transactions over time, consultation within peer groups, assessment of the markets for complements and substitutes, consultations within the household, and references to consumer information. It is questionable whether the interviewer or questionnaire designer can fully compensate for the lack of such experience and information in the limited time and space available. Hence, subjects are forced to deal with a situation which seems quite artificial from their point of view in comparison to situations where they normally arrive at monetary values. While the researcher hopes they will follow the same mental processes they would use in real markets, the social context within which contingent valuation occurs may be so artificial that people will be unwilling or unable to do so. However, Mitchell and Carson (1989) caution that such criticisms fail to distinguish between the systematic errors which underlie problems of validity and bias and the random errors which underlie reliability issues. They contend that hypotheticality is associated with random rather than systematic error. Lack of realism within a scenario therefore promotes random, directionless error not bias (Thayer 1981, 32). The failure of CV researchers to "translate posited sources for hypothetical bias into testable hypotheses and to test them..." (Cummings, Brookshire and Schulze 1986, 17) supports the proposition that the artificiality of the CV environment promotes reliability problems rather than validity problems. Nevertheless, although the severity of hypotheticality cannot be systematically determined, the problem can be significant and does form a structural basis to the biases and problems outlined earlier. For example, the embedding effect is either fundamentally caused (Gregory et al. forthcoming) or aggravated by (Kahneman and Knetsch 1992) a persons poor representation of contingent values, with monetary judgments being particularly occluded in hypothetical scenarios which are far removed from real market situations. Lack of realism probably also aggravates peoples' susceptibility to starting point and range biases (Boyle, Bishop and Welsh 1985). The WTP-WTA disparity may also result in part from the hypothetical nature of the questions asked (Bishop and Heberlein 1979; Rowe and Chestnut 1983).  137 This suggests that sources of error such as these can be minimized by increasing the realism, understandability, plausibility and meaningfulness of the scenario, as well as the perceived significance of each respondent's answers (Cummings, Brookshire and Schulze 1986; Mitchell and Carson 1989). There is a question, however, regarding how much realism can be reasonably injected into a CV exercise in order to diminish the divergence between the contingent market and a real one. For example, the most implausible and unreal hypothetical markets are often associated with those values which are poorly understood and monetarily represented, as is the case with many of the more elusive wilderness values. Also, adding realism may itself introduce bias. This conundrum is represented by Mitchell and Carson (1989, 216) in the following way: "The researcher who wishes to make a scenario more realistic faces a tricky problem: on the one hand, an insufficiently realistic scenario will be vulnerable to bias; on the other, the elements which add realism to a scenario may themselves cause bias." Thus, additional information can add realism when discerningly given, but can also add bias-inducing cues to vulnerable respondents when imprudently given. "Information overload", whereby respondents neglect important information and concentrate instead on unimportant information when forming their WTP answers, may also result when attempting to construct a realistic contingent market (Mitchell and Carson 1989, 216). Again, monetarily obscure and poorly perceived goods will require particularly high investments of information to make the market context and WTP questions realistic, provoking possible biases in the process. This issues back to the influences of context and question framing on elicited values. Such tensions will need to be accounted for when designing a survey which includes material which is intended to convey a credible market but which may unwittingly affect WTP responses. 67  Cognitive Biases The preceding sections revealed that potentially serious biases and random errors can 67 See  Mitchell and Carson (1989, 217-19) for suggestions on how to reduce random errors while minimizing biases.  138 occur out of subtle design features and elicitation stimuli. It also uncovered that explanations for the causes of these distortions more often than not broadly focus on the relationship between the contextual setting framed within a CV study and the manner in which humans process information, and how changes in the former influence the latter. The adoption of the psychological perspective was particularly pronounced when hypothesizing on how respondents react to agglomerated public goods and property rights for public goods. Psychological studies looking into how humans process information has received increasing attention from many CV theorists in recent years as they have begun to discern the overriding significance of the perceptual dimensions of contingent valuations and the relational tensions between attitudinal intentions and CV framing effects. Such attention has also highlighted the possible significance of cognitive biases. Cognitive biases are defined by Gregory (1982, 62) as "unintentional distortions of true preferences which result from limitations in an individual's ability to perceive, process, and evaluate uncertain or unfamiliar information." He goes on to say that they tend to be pervasive, resistant to change and systematic. They also tend to be most influential and extensive in studies involving complex and intellectually-challenging evaluation tasks, such as the case when evaluating environmental amenities and activities, "since both the identification and valuation of preferences will in most cases involve references to uncertain conditions or events, the assessment of information of limited validity and the weighing of competing or incommensurate factors" (pp. 62-3). The limits of human cognition and the corresponding dissonance between true preferences and expressed preferences highlight the significance of cognitive biases, particularly when respondents are required to process unfamiliar and uncertain, and thus cognitively-demanding, information. A number of prominent cognitive biases are identified by Gregory (1982), including anchoring bias, overconfidence bias and contextual bias.  139  Anchoring Bias An explanatory model for starting point bias centered on the tendency of individuals to be influenced by the initial bid when adjusting their expressed values away from the starting point. Anchoring bias occurs because people are prone to use the starting point as a first approximation or anchor for their judgments. Because subsequent adjustments, and the perceptions and valuations they represent, are still strongly influenced by the impressions received from the starting point value, total movement away from the starting point tends to be smaller than it should be considering the importance of new information (Tversky and Kahneman 1981; Slovic, Fischhoff and Lichtenstein 1988). The unintentional nature of the cognitive tendency to be highly impacted by first impressions implies that despite respondents' sincere efforts of not succumbing to the influence of a starting point or initial assumption, they are less insightful and adaptive than is needed in most cases (Gregory 1982).  Overconfidence Bias Gregory (1982) outlines four sources which boost respondents' confidence to the point where they overestimate their abilities to perceive and evaluate information, thereby biasing their valuations. Firstly, when evaluating unfamiliar or labile values, respondents may be more compelled to respond to an interviewer's question rapidly rather than admit their ignorance of the subject or their poorly-defined attitude. This bias tends to be carried through the evaluation even if commitment to an initial invalid position involves suppressing other inconsistent or competing views. A second related source stems from respondents' denial of uncertain outcomes, causing such outcomes to be viewed as known or safe entities, thereby skewing the value of their expressed preferences. Another source results from the tendency of respondents to underweigh outcomes which are probable in comparison to those which are more certain. Lastly, respondents may engage in a form of hindsight bias by believing, without realizing it, that they knew that a particular event would occur, even though they were just told about it for the first time within the valuation study (Fischhoff 1975, 1982). This last source can promote  140 overconfidence bias by instilling within respondents forecasting and predictive abilities they do not have.  Contextual Bias Context was alluded to earlier as an underlying factor which shapes peoples' impressions and assessments of particular non-market goods and services. Hence, the differing perspectives surrounding WTA and WTP measures of consumer surplus, the relational aspects of sequencing and embedding, the type and quantity of information provided, the hypothetical scenario presented, the elicitation method used, and so on all have contextual dimensions which when altered can have significant effects on people's expressed preferences. For example, information overload (Mitchell and Carson 1989) and inclusion of irrelevant information (Gaeth and Shanteau 1984) seem to impair judgment. Slovic and McPhillamy (1974) showed that an otherwise minor attribute is assessed more highly than normal if it is described more clearly and completely than other attributes. Moreover, with Prospect Theory revealing that losses are more highly valued than gains, expressed preferences and choice decisions can be significantly impacted by whether an outcome is framed as a gain or a loss and whether a negative outcome is evaluated as a cost or an uncompensated loss. For example, Tversky and Kahneman (1981) demonstrated that people can reverse their preferences if wording of two public health problems was framed in terms of "saving lives" versus "loss of life". Context effects such as these suggest that question wording and other "framing effects" (Tversky and Kahneman 1981; Hogarth 1982; Slovic, Fischhoff and Lichtenstein 1988) which build and shape the character of context can significantly affect the estimation of preferences. As a result, Fischhoff, Slovic and Lichtenstein (1980, 118) make the argument that "Subtle aspects of how problems are posed, questions are phrased, and responses are elicited can have substantial impact on judgments that supposedly express people's true values." It appears that people when faced with complex values unconsciously utilize  141 simplifying strategies to cognitively organize alternatives (Harris, Driver and McLaughlin 1989; Payne, Beaman and Johnson 1992), with such strategies being relied on more heavily as the complexity of the valuation task increases (Johnson, Meyer and Ghose 1989). This is often accomplished by, among other things, disregarding shared traits and integrating similar characteristics between alternative goods (Gregory 1982), relying on easily imaginable information or, as we have seen, anchoring on starting points (Gregory et al. forthcoming). Such judgmental rules of thumb, or heuristics, while being useful and at times necessary to simplify complex judgments, can produce biases (Kahneman, Slovic and Tversky 1982). Because CV studies typically require respondents to make holistic judgments about multidimensional values, they provoke respondents to take short-cuts and make use of fewer cues to construct their representations of expressed preferences than they would in cases of more detailed sets of tradeoffs (Slovic and Lichtenstein 1971; Harris, Driver and McLaughlin 1989). Holistic CV studies are thus vulnerable to promoting contextual bias, as people pick up on and incorporate into their valuations emphasized, cognitively simpler, better known, easily monetized, and/or other cues which simplify the task of preference development; at the same time, respondents tend to ignore many cues or information which may be important for an accurate portrayal of preferences, but which lay hidden or are obscure and thus complicate the mental effort involved in constructing a unitary measure of such preferences (Kahneman and Tversky 1979; Slovic, Fischhoff and Lichtenstein 1988; Tversky and Kahneman 1982, 1988). Based upon these and other findings and hypotheses arising from behavioral research, critics of the CVM assert that a unitary measure cannot capture with any accuracy the value of a good which is made up of complicated combinations of known, unknown and poorly conceived parts (Gregory et al. forthcoming). Instead, procedures which decompose wholes into their constituent parts are often recommended above holistic approaches for multidimensional values because they allow respondents to actively construct their values from the bottom up. One set of decomposition procedures which has received particular attention in recent years is multiattribute utility technology, a topic to which we now turn our attention.  142  Multiattribute Utility Technology Procedure Multiattribute utility technology (MAUT) 68 represents numerous models and techniques which provide "a formal basis for describing or prescribing choices between alternatives whose consequences are characterized by multiple value relevant attributes" (Fischer 1975, 7). It was originally conceived as a means of quantifying and analyzing complex individual preferences among decision alternatives that vary on multiple conflicting objectives (Edwards and von Winterfeldt 1987). Several MAU elicitation techniques have been designed to measure values of objects or outcomes which have several dimensions or attributes of value. In the same manner that the TCM, HPM and CVM are applications of the underlying conceptual assumptions of benefit-cost analysis, MAUT is linked to the assumptions of decision analysis (DA). 69 As such, many prescriptive MAU models and sets of assessment procedures have been developed to assist decision makers in their quest in forging more informed tradeoff decisions. Although it can be used as a descriptive tool to explain and predict the tradeoffs of decision makers who are left on their own (Fischer 1975), the techniques and models of MAUT are best applied in a normative manner to guide decisions (Kozielecki 1981; Edwards and Newman 1982). While numerous MAUT procedures exist in theory and practice, von Winterfeldt and Edwards (1986, 273) identify five steps which they all include: 1. 2. 3. 4.  Define alternatives and value-relevant attributes. Evaluate each alternative separately on each attribute. Assign relative weights to the attributes. Aggregate the weights of attributes and the single-attribute evaluations of  68 See Fischer (1975), von Winterfeldt and Fischer (1975), Edwards (1977), Johnson and Huber (1977) and Edwards and Newman (1982) for procedural overviews, assessments and applications of MAUT. 69 Decision analysis is a normative framework and set of systematic procedures and methods developed to rationally analyze complex and uncertain decision problems and provide guidance to decision makers. For comprehensive reviews see Raiffa (1968), Keeney and Raiffa (1976) and von Winterfeldt and Edwards (1986). Keeney (1982) provides a good procedural overview of DA, while Howard (1980, 1988) are good sources for assessments of DA.  143 alternatives to obtain an overall evaluation of alternatives. 5 . Perform sensitivity analyses and make recommendations. They point out that steps 1 and 5 are similar for all MAUT techniques, while procedures for single-attribute evaluations (step 2), assigning weights (step 3) and modelling for aggregation  (step 4) differ between them. The presented steps reveal that MAUT methodologies provide a means of decomposing the overall multifaceted evaluation task into a set of simpler subtasks, easing the judgmental burden of the decisionmaker (Eils and John 1980). Being based on the "decomposition paradigm" (Kozielecki 1981), MAUT facilitates the construction of singleattribute value and utility functions, which, after being recomposed using a suitable aggregation rule and weighting scheme, provide utility measures of respondent's expressed preferences for the various multidimensional alternatives under evaluation. To illustrate its use, a simplified version called SMART (simple multiattribute rating technique; see Edwards 1977; von Winterfeldt and Edwards 1986) will be briefly outlined as an example of MAUT. Edwards' SMART consists of ten steps, which Hogarth (1980, 228) has subdivided into the following four phases: (1) structuring the problem; (2) determining the importance of the dimensions of value; (3) measuring alternatives on the dimensions; and (4) choosing the optimal alternative. Although numerous aggregation rules and weighting procedures have been developed for MAUT, the methods illustrated here concentrate on taking a weighted linear average. Besides being more widely used, understandable and simpler than other methods, there is empirical and theoretical support in the literature (Dawes and Corrigan 1974; Einhorn and Hogarth 1975) to suggest that the linear model presented yields close approximations to much more complicated nonlinear and interactive utility functions (Edwards 1977).  Structuring the Problem The MAUT process begins with the structuring of the decision problem. This phase identifies the stakeholders, the decision problem or purpose of the evaluation, the entities to be  144 evaluated and the dimensions or attributes of value. Because an initially clear structure provides improved guidance for respondents and interviewers alike throughout all the remaining phases and because the best alternative may become evident simply by structuring the problem, the structuring exercise is considered by many as the most important MAUT phase (Edwards 1977; Hogarth 1980; von Winterfeldt and Edwards 1986). As McDaniels (1990, 14) puts it, "If decision making processes founder it is often because of a lack of a clear structure." Step 1: Identify the Stakeholders to be Involved  The structuring phase begins with the determination of all relevant individuals or interest groups who have a stake in the decision and who should be included in the process. Stakeholders provide the value attributes which are to be elicited. Attributes are the value dimensions which stakeholders care enough about that to ignore them in the decision framework would lead to a poorer decision than if they were considered (Edwards and Newman 1982). Step 2: Identify the Decision to be Made  Determining the decision to be made brings into focus the purpose and context of the value elicitation exercise. Along with the identified stakeholders, the context and purpose of the decision shapes the utility values to be elicited (Edwards 1977; Hogarth 1980). The decision context is defined by Keeney (1992, 30) as that which "defines the set of alternatives appropriate to consider for a specific decision situation." The purpose of the decision, on the other hand, represents the fundamental reasons for requiring clarification of the issues represented in the decision. It is thus extremely important to clearly define and identify the decision context and purpose from the onset. An example of a decision component which needs to be carefully considered is the relationships and implications arising from different but related decisions. Because decisions can be regarded at different hierarchical levels, from the higher order, fundamental goals to the lower order, contributory goals, the decision identification step should define the decision level  145 which is relevant for the values being elicited. Determining the hierarchy of decisions in which the evaluated decision is nested is also important because higher order issues provide insight into lower order issues. McDaniels (1990) exemplifies hierarchical identification within the context of the forest preservation issue in B.C. and the U.S. Pacific northwest: Various organizational processes designed to tackle different facets of the issue have broadened their frame of reference by shifting their focus away from lower order questions (e.g., "Should a particular forest be preserved or harvested?" (p. 15)) and smaller spatial scales (i.e., stand level) and towards more fundamental questions (e.g., "What kind of forest environment do we want to leave to future generations?" (p. 15)) and larger spatial scales (i.e., regional or provincial levels). Step 3: Identify the Alternatives to be Evaluated  Once the decision has been settled on, it directly follows that an appropriate set of alternatives which are defined by the decision context be chosen. The alternatives include all possible action outcomes or options which are to be evaluated within the context of the decision. For example, if the decision facing a decision maker is choosing an appropriate level of preservation, the alternatives would be the various levels of preservation, within a realistic range, to choose from. It is important to use imagination in order to generate as many and as clear a set of relevant alternative options for the decision at hand as possible. This initial list can be paired down later after considering the realism, relevancy, clarity and degree of independence of each and the various constraints which hobble the number of alternatives which can realistically be evaluated in a manageable and cost-effective manner. It is also important to consider at what level each outcome should be evaluated, as outcomes often represent opportunities for further action (Edwards 1977). If a decision strategy includes more than one action set level, with further actions being prescribed contingent on the outcomes of the initial set, descriptions of alternatives will need to include the simultaneous specification of decision points, events that could take place between them and any information that may be learned along the way. Complicated decision strategies  146 with dynamic decision processes can be conveniently represented as decision trees (Raiffa 1968; Keeney 1982), with the determination of where to prune (i.e., where to treat an action outcome as an end in itself) a matter of convenience in most cases (Edwards 1977). The level of detail chosen will also depend on the costs and benefits of different levels of analytical detail and corresponding budgetary constraints. Decision trees provide a visual representation of the connections between different acts and events, allowing one to organize the complicated set of outcomes and providing one with the means to see both the complexity of decision situations and the simplifying assumptions that were adopted to deal with them (Hogarth 1980). For complex decisions with uncertain, difficult to imagine alternatives, it may be necessary to develop scenarios, or simplified, hypothetical representations of future events. Such scenarios would be organized around a plausible set of decision outcomes, with each scenario's degree of certainty of occurrence being represented by expertly determined probabilities (Edwards 1977).  Step 4: Identify the Relevant Dimensions of Value Alternatives are normally evaluated on more than one value dimension. An important step in the structuring phase, therefore, is specifying and organizing the different dimensions of value. These dimensions are commonly referred to as objectives and are characterized by the decision context, object and the preferred direction of movement against which alternatives are evaluated (Keeney 1992). For example, in the preservation scenario outlined in the next chapter, one of the fundamental objectives against which the preservation level alternatives are evaluated is to maximize the spiritual values associated with a wilderness experience. For this objective, the decision context is wilderness preservation, the object is wilderness-derived spirituality and the direction of preference is more rather than less of the value. Devising a simple list of goals that are deemed important for the decision at hand may help to generate insight into the main objectives (Edwards 1977). Another strategy which is often used to initiate the objective identification process is to specify an unstructured list of  147 possible consequences of the alternatives and organize the generated list into a set of general concerns (Keeney 1982). The process of articulating objectives can be helped along by studying related topics in the literature, conducting analytical studies (for example, by modelling inputs, processes and outputs), and talking to and soliciting ideas from stakeholders, experts and decision makers (Buede 1986). Keeney (1988) suggests various cues which can stimulate respondents' thinking in this regard, including pointing out more specific objectives and alternative categories of objectives which have been missed. Also, having them separate objectives which are means to an end from fundamental objectives which are ends in themselves is an important undertaking. Both types of objectives are important to the decision: "Means objectives can be very useful for developing models to analyze decision problems and for creating alternatives. However, it is the fundamental objectives that are essential to guide all the effort in decision situations and in the evaluation of alternatives" (Keeney 1992, 35). The unstructured list is then normally organized into a value structure of some kind, typically hierarchical in nature. Objective hierarchies are structured in such a way that broad, general objectives are arranged at the top, with specific, detailed objectives that describe aspects of them arranged further down. Although means-ends objectives hierarchies are used in some decision contexts, most decision analysts prefer fundamental objectives hierarchies, where lower order objectives provide detailed descriptions of, rather than means of achieving, higher order objectives (Keeney 1982, 1992). The lowest level objectives require attributes, or performance measures, in order to evaluate the degree of achievement of the objectives (Keeney 1982; McDaniels 1990). For the preservation workshop, for example, two proxy attributes were used to simplify the elicitation process, with the percentage of the provincial land base preserved as wilderness being the performance measure for the benefit objectives and provincial revenue in present value dollar terms and employment in net permanent direct job terms representing the proxy measures for  148 the negative or cost objectives. The objective hierarchy or common value tree should represent a complete set of objectives spanning all issues of concern to stakeholders and decision makers and a complete set of attributes providing an adequate basis for assessing the degree to which all objectives are met. The attributes should also be operable; i.e., they should provide a meaningful measure of the objectives in order for the alternatives to be characterized within a realistic evaluation. Other considerations in the structuring exercise include ensuring independence between the preference for and uncertainty surrounding a given attribute (termed "decomposability" by Keeney and Raiffa (1976)) and ensuring that attributes are relatively non-redundant so that double counting does not occur because of overlapping measures. It is also suggested that the attribute set be kept to the smallest size possible while still fulfilling the above suggestions (Keeney and Raiffa 1976; Buede 1986). 7°  Determining the Importance of Dimensions The goal of this phase is to determine respondents' attitudes toward the relative importance of the various value dimensions by actively eliciting their judgments. This includes having them rank the dimensions in order of importance, translating the rankings to ratings and converting the ratings to numbers that sum to 1.  Step 5: Rank the Dimensions in Order of Importance This step simply asks respondents to rank order the attributes from the most important to the least important. The ranking task is done by individuals on the basis of their own judgments of the significance of each objective. However, Edwards (1977) recommends that before individuals give separate judgments, the ranking exercise can initially be done within group processes to help foster a common information base.  Steps 6 and 7: Translate the Rankings to Ratings and Normalize Elicited importance weights are then assigned to the attributes. Weights capture the 70Winterfeldt  (1980), Buede (1986), Brownlow and Watson (1987) and Keeney (1988) provide detailed descriptions of various structuring strategies.  149 value judgments of the respondents regarding the relative importance of the rated attributes. Various weighting schemes have been developed for the rating exercise, the most common among them being rank sum weighting, rank reciprocal weighting, ratio weighting and swing weighting. A rank sum weight is elicited by adopting an inverse ranking strategy: The largest number is assigned to the highest ranked attribute, the next largest number to the attribute ranked second, and so on down the ranked list of attributes until the least important attribute is given the weight of 1. The numbers are then added and each number divided by the sum to normalize the number set so they add up to 1. Rank reciprocal weighting, on the other hand, begins with the assignment of the numerical value of 1 to the most important attribute, 2 to the next important attribute, and so on down the attribute list until the least important attribute is assigned a number which matches its ranked position (i.e., if tenth in the list, it will be assigned a rating of 10). Each numerical value is then reciprocated or divided into 1 and then normalized, thereby ensuring that the most important attribute receives the highest numerical weight. Because the previous two weighting methods assign numbers to attributes on the basis of their ranks only without considering their relative degrees of importance, many decision analysts prefer procedures like ratio weighting which consider relative worth, despite their being more complicated and time consuming. Ratio weighting begins by assigning the value of 10 to the least important attribute. The attribute that is ranked just above the last ranked attribute is then assigned a number which depicts how much more important it is relative to the least important attribute. An assignment of 20 for example would signify that the respondent judges the second least important attribute to be twice as important as the least important attribute. The respondent then continues up the list until all attributes are assigned importance weights. If two or more attributes are judged to be equal in importance, they are given equal numerical weights. Ratio weighting is complete when each attribute's weight is normalized (Edwards and Newman  150 1982). Using the concept of importance as a basis for weighting attributes has been criticized by some in the literature (e.g., Keeney and Raiffa 1976). This is because such procedures ignore the dependence of the elicited units of the single attribute on the range of the scale over which the value function is defined. The assigned weights tend to be sensitive to the range of the scale changes, with weights typically increasing with increasing ranges and decreasing with decreasing ranges (von Winterfeldt and Edwards 1986). The relative weights assigned to all the values derived from wilderness, for example, would likely be different if they were elicited within the context of the specified ranges rather than without. For example, the weight spread between ecological values, if deemed most important, and spiritual values, if judged least important, could be greater if the respondent was asked to consider the gains that would occur in such values as a result of a 6% to 12% level increase in the land base allocated to wilderness than if he were to ignore such a consideration; this outcome would happen if the respondent believes that the preservation doubling enhances the benefits derived from the ecological values more than the benefits gained from the spiritual values, a likely judgment with the former value set being preferred. Weighting the attributes in the absence of the range in scale thus ignores the differential effects that such ranges have on different attributes. The swing weighting technique is designed to capture such effects by having respondents consider the weight of each attribute on the basis of the differential degree of improvement that occurs in each attribute from a change from the worst alternative (e.g., 6%) to the best alternative (e.g., 12%). von Winterfeldt and Edwards (1986) recommend assigning 100 to the most positively affected attribute, with the others being scaled downwards on the basis of their relative lesser improvements in value. The raw weights are then normalized to add to one as in all the previous weighting procedures. Measuring Alternatives on the Dimensions Step 8: Measure the Relative Value of the Alternatives on each Dimension  The next step is to determine the location measure for each attribute. A location measure  151 is an assessment of the desirability of each alternative in terms of the degree to which each contributes to specified improvements in single attributes. Because location measures are expressed as a number technically called an utility, the measure derived from this step is often referred to as a single attribute utility function. Determining the utility function for each attribute is complicated by the fact that attributes are not expressed in commensurable units. As a result, rescaling is required. The single attribute utility assessment procedure may involve simple mathematical transformations of objective measures or impressionistic judgments, with the former requiring transformation of what is considered to be true attribute measures into scales with comparable meaning; the alternative latter method is used when the attribute is inherently judgmental, making transformations unnecessary in most cases, as decision makers or impartial experts subjectively derive scales from the onset which allow comparable measures of desirability (Edwards and Newman 1982). A common means of deriving comparable scales involves using a value or worth scale which goes from 0 to 100, with the end points corresponding with the minimum and maximum plausible values of the given attribute. Experts or decision makers can be helped along in locating each alternative on each attribute's scale by graphically representing the relationship between the natural units of the attribute and its utility units. This is done by having the horizontal or X axis represent the range of an attribute's natural units, going from its lowest to its highest plausible value, and having the vertical or Y axis represent the range of its corresponding utility units, going from 0 to 100. For linear relationships, a straight line is then drawn from the point of intersection between the 0 utility measure on the Y axis and lowest plausible value on the X axis to the point of intersection between the 100 utility measure on the Y axis and the highest plausible value on the X axis. Assigning utility is simply accomplished by locating the alternative on the natural scale and reading off its utility on the Y axis. Simple linear relationships between utility scales and an attribute's natural units can also be represented in equational form. As such, if LA is the actual location measure for the alternative, L min is  152 the attribute's minimum value and L. is the attribute's maximum value, the single attribute utility for the alternative can be determined with the following equations: Location of LA = 100 (LA - Lmin) / (L nax - L . ), where more is preferable to less and Location of LA = 100 (Lmax - LA) / (Lmax - Lmin), where less is preferable to more (Edwards and Newman 1982, 66). Simple mathematical equations can also be developed to measure utilities for bilinear relationships, where an intermediate unit represents the maximum plausible value (see for examples Edwards and Newman 1982, 69). However, locating utilities poses more difficult challenges for respondents who are concerned about the nonlinearity of their preferences. One example of nonlinearity would be when respondents believe that specific attributes follow the economic law of diminishing returns. The shapes of respondents' utility curves also represent their behavior in the presence of risk. One of the simplest ways of overcoming such problems is to ask respondents to draw graphs representing their judgments of the utility curves for each attribute (Edwards 1977). Other direct techniques have also been devised to elicit values for nonlinear preferences, including those dependent on hypothetical indifference judgments, where utility functions are derived by constructing indifference curves for pairs of variables (see MacCrimmon and Siu 1974; Vertinsky and Wong 1975; von Winterfeldt and Edwards 1986). However, these methods are time-consuming, tedious and impractical when tradeoffs need to be made between many variables (Slovic, Fischhoff and Lichtenstein 1977). Indirect methods, which introduce risk attitudes and probabilities into otherwise riskless situations, have also been developed to assess both weights and utilities. These are justified within MAU theory but are exceedingly complex (see Vertinsky and Wong 1975; Keeney and Raiffa 1976; Slovic, Fischhoff and Lichtenstein 1977). Although difficult to obtain, uncertainty surrounding the magnitude and likelihood of consequences from alternatives can once incorporated, however, be conveniently represented to decision makers; experts can do this by laying out the sequence of connecting impacts with the probabilities of  153 intermediate events using a decision tree or visually showing how decision variables are connected in a causal or sequential manner using influence diagrams 71 (Gregory, Keeney and von Winterfeldt 1992).  Step 9: Calculate Overall Utilities for Alternatives The preceding steps of decomposition and isolation of the relevant attributes are followed by a recomposition step, in which the disaggregated utilities and importance weights are aggregated into a multiattribute utility measure for each alternative. For the SMART description provided, the most frequently used and simplest additive equation 72 will help illustrate the calculation of the composite utility for each alternative i, summed across the attributes j: MAU I = wji13 where wj is the normalized importance weight of the jth attribute and u jj is the rescaled utility of the ith alternative on the jth attribute (Edwards 1977; Kozielecki 1981; Edwards and Newman 1982). The wj measure represents the output of Step 7 and u 1.i the output of Step 8. If probabilities are explicitly incorporated into the model, this final utility calculation represents an expected utility (Gregory, Lichtenstein and Slovic forthcoming). Choice  Step 10: Choose the Alternative The normative rule is to choose the alternative with the largest assessed utility. Other considerations external to the MAUT model may also be factored into the overall decision and thus influence the final choice; however, the tools and outputs of the model will provide significant guidance to decision makers in their determination of the best overall alternative. 71 Although  the most familiar and widely used structuring tools are hierarchical-based (e.g., decision trees, value trees, etc.), von Winterfeldt (1980), von Winterfeldt and Edwards (1986) and Howard (1988, 1989) explore influence diagrams and other alternative decision organizers. 72 See Keeney and Raiffa (1976) and von Winterfeldt and Edwards (1986) for examples of multiplicative and other MAU aggregation rules.  154 Also, modifications to the procedures outlined above will allow other contributory factors, such as budgetary constraints, to be integrated into the model (Edwards 1977; Hogarth 1980). For SMART or other versions of MAUT to be used with confidence, the assumption of value independence should be checked. Value independence means that preferences for any  dimension of a specific alternative should not be affected by its measurements on the other dimensions; in other words, there must not be strong interactions between attributes in terms of the respondent's preferences (Hogarth 1980). Also, sensitivity analysis can help boost confidence in the evaluative results if such an analysis uncovers very little variability between them. This is done by observing the extent to which overall preferences for alternatives change when quantitative inputs (e.g., weighting schemes, outcome probabilities, etc.) to the problem are varied (Hogarth 1980). Sensitivity analyses can detect those inputs which have inordinate influences on the total utility measure and should, as a result, be subject to additional elicitations from stakeholders or the public at large. They can also help analysts identify and begin to diminish the major causes of disagreements between stakeholder groups over differences in utilities and weights (tradeoffs) by targeting those input variables which consistently bring preference measures into greater quantitative agreement and revealing to stakeholder groups how small alterations in their utilities and tradeoffs can result in significant changes toward the calculated values of other groups (Gregory, Lichtenstein and Slovic forthcoming). It is also important to note that not all of the steps outlined above need to be covered or slavishly followed as sequentially presented in order to act as an analytical tool for addressing complex decisions (Hogarth 1980). For example, the structuring phase may explicitly clarify and organize stakeholders' preferences in such a way that areas of conflict are clearly identified, making resolution through a negotiated compromise possible. The latter stages may also be ignored if an obvious option comes to light which all involved discern as mutually advantageous.  155  Applications MAUT within the normative framework of decision analysis has been used to evaluate complex decisions in a diverse variety of public and private contexts, including in the corporate, health and safety, social, educational, energy, environmental, developmental planning and resource management fields. 73 In terms of environment-related issues, decision analyses have been used as a decision aid for the siting for disposal (Merkhofer and Keeney 1987), transporting (Keeney 1988) and management (Lathrop and Watson 1982) of nuclear waste, the siting of energy facilities (Keeney 1980), the setting of long term energy policies (Keeney, von Winterfeldt and Eppel 1990), the setting of environmental standards (North and Merkhofer 1976; von Winterfeldt 1980, 1982) the constructing of objectives for climate change research (Keeney 1992) and the selecting of alternative development options which have environmental consequences (Keeney and Wood 1977; Rozelle 1982; Edwards and von Winterfeldt 1987). Despite being employed for such forestry problems as fire and pest protection (Holling 1978; Cohan, Haas and Roussopoulos 1983) and silvicultural prescription selection (Martell and Fullerton 1987; Brumelle et al. 1988; Pearce, McDaniels and Swoveland 1990), McDaniels (1990) points out that decision analysis has rarely if ever been used for land allocation decisions. Drawing on the literature, as well as a case study in which he interviewed four stakeholder representatives for the purposes of eliciting and structuring objectives which are relevant for old-growth forest decisions, McDaniels (1990) concludes that there is significant scope for the use of the techniques of decision analysis in clarifying decisions related to forest allocation. He goes onto say (p. 51) that "This clarity can come in a better representation and understanding of the value conflicts involved, a better representation of the implications of alternatives, and the creative design of new alternatives that are seen as more attractive by all parties", and recommends the sponsoring of selected case studies in order to determine the 73 See Keeney and Raiffa (1976), Keeney (1982) and von Winterfeldt and Edwards (1986) for  examples of experimental and real applications of decision analysis and MAUT.  156 potential uses of decision analysis for old-growth forestry decisions. 74 Strengths The primary strength of MAUT is that it conforms to the constructive nature of human preferences by openly encouraging participants to successively revise and refine their values while learning as they go and by accommodating the multidimensionality of such values (Gregory and McDaniels 1987). By decomposing multifaceted, complex decision problems into systematic, explicitly specified and simpler substeps, MAUT eases the cognitive task of making judgments in the presence of complexity (Eils and John 1980; Gregory et al. forthcoming). Value formation and clarification is facilitated by the use of such structuring aids as value trees and objective hierarchies. The process of developing a mutual set of structured objectives such as a common value tree, along with various cues to spur thinking about values, helps respondents uncover, clarify and organize important and sometimes hidden values and attributes, as well as identify irrelevant ones (Brownlow and Watson 1987). New ideas which form out of the process of deriving a clear set of objectives feed back to the option generation step, encouraging the creation of additional alternatives or exposing formerly conceived ones which should be improved or rejected. A complete and clear set of objectives also provides a guide for data collection and scenario selection; this supports the task of obtaining data that help evaluate the relative desirabilities of the consequences from various alternatives on the objectives and different scenarios which could improve such an evaluation (Edwards and von Winterfeldt 1987; Keeney 1988, 1992). Explicit elicitation of values and attributes, as well as the combined effort of structuring and refining generated objectives, should facilitate communication and mutual understanding between stakeholders and highlight areas of common interest and disagreement, thus providing a basis for resolving conflicts and making tradeoffs through compromise (Keeney and Raiffa 74Decision analysis has also been recommended as a decision tool for managing endangered  species (see Maguire 1991).  157 1976; Einhorn and McCoach 1977; Keeney 1988, 1992). Moreover, the mechanical, systematic nature of the process allows interviewers to teach the framework and its steps with relative ease, making it relatively transparent (i.e., readily understood) (Einhorn and McCoach 1977; Howard 1988). It is also suggested that simply providing a more systematic structure within which preferences and values can be openly aired helps reduce disagreement among members in group processes (Eils and John 1980). MAUT, besides encouraging explicit valuation judgments from respondents, also promotes explicit factual judgments from experts. Experts are allowed to concentrate on the activities that they are familiar with, such as assessing risks and judging probabilities, while stakeholders are allowed to concentrate on constructing the dimensions of value. Because the method provides a logical and systematic means of distinguishing facts from values, subjective, intuitive knowledge and objective data can be deeply examined and developed by each group and then later combined. This eases the task of each group and relatedly improves the quality of the information obtained (Keeney 1982; Gregory et al. forthcoming). It also helps to differentiate between disagreements based on values and those based on facts (Maguire 1991). Because tradeoff issues form an inherent part of decision problems, they must be addressed either explicitly or implicitly in the overall value model. By explicitly assessing value tradeoffs among attributes through their ranking and weighting and option tradeoffs through their relative utility measures, MAUT handles the issue in the preferred definitive and open manner. It forces stakeholders to make informed tradeoffs between objectives. Moreover, explicitly and clearly structured objectives and attributes help inform the rating exercise (Gregory, Keeney and von Winterfeldt 1992). Additional strengths of MAUT include its flexibility, agility in integrating market and non-market values, lessening of the embedding problem and lessening of problems deriving from people's beliefs about non-independence (Gregory et al. forthcoming). Firstly, MAUT is  158 flexible in changing circumstances. Having explicit and methodical stages in which to elicit value attributes and utilities permit researchers to redo calculations when additional new information arises. Secondly, neither market values nor non-market values are given an advantage in the MAUT model. Both economic models and simple non-market measures can be accommodated. Thirdly, because respondents are assisted in structuring their monetary values in a defensible manner, people's representation of such values are improved, thus reducing the embedding problem. If, on the other hand, the source of the problem arises from people's desire to "purchase moral satisfaction" (Kahneman and Knetsch 1992), embedding will also be lessened, as MAUT does not focus directly on the spending of money. MAUT would also help to reduce the cause of embedding that arises from people's inabilities to respond with sufficient refinement to design specifications (Gregory, Lichtenstein and Slovic forthcoming), or other complexity-inducing problems which diminish people's abilities to absorb details (see Fischhoff et al. 1993), because its elicitation methods are decompositional, thus allowing people to evaluate various dimensions of value sequentially as opposed to cognitively handling them all at the same time. Also, the utility for each attribute is elicited across the range (e.g., respondents could be asked to give swing weights for the various wilderness values when going from 6% to 12% or asked to give separate weights for each value at both levels). Lastly, holistic approaches such as the CVM are susceptible to eliciting measures of values from people which are biased because of people's perceptions of value dependencies; this occurs when an improvement in one value is thought to lead to an inevitable, similar improvement in another related value. Such beliefs about non-independence would be uncovered in MAUT's structuring phase, enabling the model to accommodate them through adjustments. Gunton and Vertinsky (1990, 9) provide an appropriate concluding assessment of the strengths of decision analysis in which MAUT is embedded in the following paragraph: The advantage of decision analysis is that it allows flexibility in terms of objectives,  159 definition and measurement of effectiveness. It allows an expression of subjective preferences among timings of consequences and the degree of uncertainty which is involved. The technique is transparent, the assumptions are minimal, and data bases can be used in conjunction with judgment to ensure comprehensiveness. The technique encourages a systematic approach to problem solving and provides a framework for sensitivity analysis. Weaknesses Although it is thought that some of the cognitive biases outlined earlier within the CVM section will be reduced using a well designed and applied MAUT process, it is not known whether the confidence in these potential improvements is well founded, nor whether such improvements are significant. Also, MAUT can theoretically suffer from the same motivational biases which afflict CVM, although, with the problem being decomposed down to greater levels of structured detail and with monetary measures being relied on less, hypothetical biases may be lessened. Again, motivational biases will present less of a problem if the interviewer possesses strong analytical and interactive skills and the process is well structured from the onset. As with the CVM, the type and detail of information provided and other contextual influences may introduce bias. However, again as with the CVM, the significance of information bias is not known and may be mitigated against. In any event, MAUT, by being a more detailed and structurally refined assessment approach, presumably possesses sufficient structural latitude to accommodate constructed scenarios with greater degrees of plausibility and realism, thereby retaining the capacity to reduce information and other biases and errors more successfully than the CVM. The high costs of time and money involved in preparing for, conducting and analyzing the results from a detailed MAUT workshop is often cited as a major disadvantage of the approach (Keeney, von Winterfeldt and Eppel 1990). Others have argued that the additional information provided may justify the added cost (Einhom 1974) and that simpler, less costly MAUT procedures can be designed to reduce the time and expense involved while still generating useful results (Einhorn and McCoach 1977; Edwards and Newman 1982). Gregory  160 et al. (forthcoming) add that the cost of CV processes may be increasing due to such recent trends as using focus groups and avoiding mail-based surveys, resulting in a reduction in the disparity between the costs associated with doing MAUT and CV assessments. Another criticism levelled against MAUI is the relatively high degree of expertise required of the analyst, with the assessment necessitating the full participation of the analyst throughout the elicitation procedure and the application of the MAUT techniques involving as much art as science. Gregory et al. (1992 forthcoming) counter that the skills required are no more demanding or subjective than conventional CV or benefit-cost analyses. Howard (1980, 1988) adds that use of such structuring tools as influence diagrams eases the demands on both the analyst and respondent. Other concerns have centered on problems which can arise when applying the various aspects of the steps involved in decision analysis. Van and Vecsenyi (1983) and von Winterfeldt (1983) provide examples of many of these as well as various strategies to minimize their effects. von Winterfeldt (1983), for example, identifies difficulties which can occur as a result of hidden agendas in the client-analyst relationship, poor definition of the decision problem, numerous institutional obstacles, structuring and elicitation complexities and obstructions in using and implementing the model. He adds that such problems can be reduced through improved and more insightful preparation, design and execution of the analysis (in the case of design problems) and through improved understanding, interaction and levels of trust between the players (in the case of motivational problems). Many of these have been alluded to earlier in the procedural outline of MAUT. Although convergent validity studies have found high correlations among a variety of models (risky and riskless, multiplicative and additive) and assessment methods (holistic and decomposed) it is not known how valid MAUTs results are. Finding agreement between the elicited values of MAUT and CVM (see, for example, Humphreys and Humphreys 1975), for example, reveals nothing about the validity of MAUT when the validity of CVM is unknown  161 (Slovic, Fischhoff and Lichtenstein 1977; Kozielecki 1981). Validating MAUT techniques by comparing alternative MAUT models, on the other hand, has shown that decomposition procedures are more valid than holistic approaches in some circumstances (see, for examples, Fischer 1976, 1977); how generalizable these results are are unknown at this point however. Summary  Though it cannot be stated with certainty that MAUT represents an improvement over holistic approaches, recent evidence coming from empirical studies and out of the psychological literature provide impetus to the notion that it is a superior approach in eliciting complex values in complex decision contexts. MAUT is a decomposition technique that appears to be in greater harmony with such cognitive strategies as those related to simplifying heuristics and preference construction. MAUT helps participants to identify and organize the various dimensions of their values and then uses these as a means for weighing value tradeoffs. Holistic WTP or WTA measures like those coming from most CV studies, on the other hand, ignore such cognitive realities and thus place too many cognitive demands on the people engaged in complex elicitation tasks (Gregory et al. forthcoming). The indirect approaches of the preceding chapter also have numerous shortcomings which limit their applicability in eliciting wilderness values, the most problematic of which include their inability to quantify non-use values and their rigorous assumptions. As a result, MAUT may be a simpler and more valid technique than the relatively more demanding approaches based on WTP (e.g., CVM) and surrogate prices (e.g., TCM and HPM) in decision environments that include incommensurable, complex and uncertain amenity resource values (Gregory and McDaniels 1987). As previously mentioned, although there is probably great scope for its use in such contexts, MAUT has rarely if ever been employed to inform land allocation issues. While  162 based on compelling theory, the above assertion regarding MAUT's superiority is thus only speculative at this point. The chapter to follow (Chapter VI) provides a description of a workshop which uses both the holistic-based CVM and decomposition-based MAUT approaches to elicit values related to wilderness allocation. It was conducted in part to determine the feasibility in using each to elicit valid values in the face of such complex decision problems.  163 Chapter VI WILDERNESS ALLOCATION DECISIONS WORKSHOP PROCEDURES  Introduction  Two workshops were held to elicit values related to forest preservation decisions from two different groups of Master's students attending the University of British Columbia. The first workshop sought values from students from the School of Community and Regional Planning; it was held at U.B.C. on Saturday, November 14, 1992 and was attended by fifteen students. The second took place on Saturday, January 16, 1993, with thirteen students from the Faculty of Forestry taking part in the elicitation exercise. Both took approximately 4 hours to complete. This chapter will provide a detailed description of the procedure format which was followed in the two workshops to elicit wilderness values from the two groups of respondents. An example of the package of question sheets on which responses were written by each participant during the course of each workshop is provided in Appendix II. However, the response results themselves are not presented in this volume; rather, Chapter VII presents the results from a feedback questionnaire which was filled in by respondents at the end of the workshop in order to gauge their impressions about the workshop and its two valuation approaches. The fundamental purposes for including this and the next empirical chapter are (1) to provide an example of one possible way of approaching the problem of eliciting wilderness values in allocation decision contexts, (2) to compare respondents' attitudes towards the evaluative techniques represented by CVM and those represented by MAUT and uncover respondents' feelings and thoughts on the utility of such approaches for clarifying wilderness values and advancing debate on allocation issues, and (3), based upon the findings in the  164 literature (Chapter V), what was learned from the workshop (Chapter VI) and participants' responses to the feedback questionnaire (Chapter VII), to choose the preferred technique, suggest possible refinements in the preferred procedure and describe the general usefulness of elicitation tools like this for gaining insight into complex land allocation decisions (Chapter VIII).  Workshop Procedures  Introduction I Context to Wilderness Issue Participants were initially given an introductory overview of the purpose, structure and tasks of the workshop and the general context of the B.C. wilderness-harvesting allocation issue. They were told that the main purpose of the workshop was to elicit value relevant information from them regarding the costs, benefits and tradeoffs associated with forest preservation decisions and generally test the feasibility of using the devised procedure to obtain public value inputs to the preservation debate. Their role would thus be to provide judgments about the relative importance of key factors that are useful in deciding about the best level of preservation. They were reminded that the answers that they provide will be subjective value judgments; they were thus encouraged to be reflective and thoughtful of their underlying values when providing answers on the question sheets. After the overall workshop format, agenda and objectives were outlined, the B.C. forest allocation decision context was briefly described. Included in the description of the context was a general overview of the nature of the value conflicts and resource constraints which shape and complicate the decision making process. Also impressed upon them was the related importance of using public value judgments in the decision making process and the corresponding tradeoffs which are required to be made when devising workable solutions in the face of a diminishing resource with multiple conflicting values placed upon it.  165 Just prior to handing out the booklet of worksheets, the principle decision alternatives were introduced: 6% (reflecting the approximate current level) and 12% (reflecting a doubling of the current level) of the provincial land base preserved as wilderness. It was explained to the participants that although 12% does not necessarily represent the requisite level of preservation from an ecological or other standpoint, it is a suitable option to evaluate because it is one which such bodies as the World Commission on Environment and Development (1987) recommend as an appropriate land base percentage to set aside as natural area. It was also explained that a doubling of the provincial area devoted to protected status represents a major goal of the B.C. government, as represented by the 12% target laid out in the B.C. Ministry of Environment, Lands and Parks' (1992, 1993) "Protected Areas Strategy" initiative. In order to deepen participants understanding of the possible areas to be preserved, their attention was drawn towards a map put up in the front of the room which was developed by the Valhalla Society, in cooperation with others, in 1988 (see Valhalla Society 1988) 75 . Although it suggested a set aside of around 13%, the map provided a graphic example of what an approximate doubling of the land base allocated to wilderness could look like on the ground. The introductory portion of the workshop ended with the fielding of questions and the handing out of the question sheet booklet (see Appendix II).  Elicitation of Wilderness Values Question Sheet 1: Contingent Valuation Household Willingness to Pay for Increased Preservation Participants were asked to read the first willingness to pay question on question sheet 1: 1. What is the maximum your household would be willing to pay in dollar terms in increased taxes in each year for the next 10 years in order to double the amount of the provincial land base preserved as wilderness within B.C.? What was specifically being asked was explained to them to help them answer the question. This included clarifying that what they were buying was approximately double the present 6% 75 A second edition (1992) map has since been put out; it suggests a preservation level in B.C. of just over 14%.  166 of the land base devoted to some form of protected status. The question of what constituted "wilderness" arose at this stage; they were informed that although various natural communities could be included in the defmition, the common link between the areas selected for preservation designation was that the areas' natural character be relatively free of disturbance arising from human-caused activities. To provide an example of what they were buying, they were asked to examine once again on the Valhalla Society map the present level of preservation and what such an area represents on a provincial basis and compare it to the proposed total area, which represents approximately twice the current level. The clarification stage also involved, besides indicating what they were buying, revealing why wilderness protection costs something in the first place. The major costs of preservation were brought up in general terms (i.e., without monetary figures) to illustrate this point, including the losses in provincial revenue and employment associated with reductions in the commodity forests available to the forest industry. Moreover, they were reminded that the relatively high standard of living which B.C. residents enjoy, including the government services which are supplied, results to a fairly significant degree from public revenue generated from the taxes, stumpage, royalties, etcetera paid by the forest sector and the taxes paid by its workers. After being given the approximate percentage of their earnings presently paid in taxes, they were asked to consider what their households' average annual earnings will likely be over the next 10 years; based upon these considerations, and their underlying preferences for acquiring double the current level of wilderness in the province, they were then read aloud the first willingness to pay question and asked to answer it. Provincial Willingness to Pay for Increased Preservation After completing the first question and being provided with an opportunity to have any  additional questions relating to it answered, respondents were asked to read the second willingness to pay question on question sheet 1:  167 2. How much do you believe the province should be willing to pay in foregone revenues in each year for the next 30 years to double the amount of preserved wilderness land within B.C.? 1992 budgetary figures from various provincial ministries were written on the board to help respondents derive an answer to this question. Such fiscal numbers as the Total and Forests budget revenues and expenditures, as well as expenditures from such ministries as Parks, Advanced Education and Health, were provided in an effort to build a general picture for the respondents of what the fiscal position of the government was, what the relative expenditures were between ministries as a percentage of total expenditures and how important Forest revenues were as a percentage of Total revenues. Respondents were then asked to weigh these considerations, along with their views on compensating timber companies for lost licenses, while determining what the province ought to pay in their judgment to obtain the preservation doubling. The second question was then read aloud and they were given time to contemplate and write down their answers. Question Sheets 2 to 13: Multiattribute Utility Technology Propose I Structure Wilderness Objectives Before proceeding to the MAUT question sheets, respondents were given an opportunity to propose as a group the benefits and costs associated with wilderness preservation which they judged to be important. Wilderness benefits were obtained first. As benefits were proposed by participants within the group discussion they were written on the board. To help them through the exercise, relationships between benefits were identified and structured accordingly on the board, distinctions between ends and means clarified, fundamental benefits identified, incidences of redundancies pointed out and additional benefits suggested. The benefit elicitation session also permitted individuals to explain why their proposed benefits were important and allowed them to comment on other participants' suggestions. They were also helped in drawing out and clarifying poorly conceived benefits, which were also added to the emerging list. When the group was satisfied that its structured list  168 of benefits were sufficiently refined and comprehensive, the list was compared to the benefit categories hierarchy constructed prior to the workshop. This hierarchy (which is displayed in Table 1, Chapter III) was presented to the participants (it was contained within their question booklet) and its categories and subcategories were discussed and compared to the product developed by the group. Once all the values in the hierarchy were described and explanatory examples provided for each, and participants were satisfied that it was complete and represented all the benefits proposed in their list, they were told that the benefit hierarchy would form the basis for many of the evaluations to follow. A similar process was followed to elicit the costs which respondents felt were associated with preserving wilderness. They were then asked to look at question sheet 2 in their booklet, on which the two cost categories to be used in the ranking and rating exercise were identified and defined. Costs Associated with Increased Preservation (Case 1) The costs of preservation were then explained in terms of the two cost categories: provincial revenue and employment. Provincial revenue was simply defined as "net forestryrelated income" and its performance measure or attribute was "measured in present value of 1992 dollars". Employment was defined as "net forestry-related employment" and its attribute was "measured in net permanent direct jobs". Respondents were told that although other costs, such as community stability, are also associated with increased wilderness protection, provincial revenue and employment represent the largest impacts on the province and its forestry workers. They also were informed that while this elicitation exercise spelled out the costs in terms of losses in direct revenue (stumpage and corporate income tax from the logging, wood products and paper and allied industries) and direct employment (jobs from these industries) only, inclusion of indirect and induced impacts would likely double the attribute estimates of those derived for the direct impacts associated with protecting twice the current  169 level of preserved areas in the province. Respondents were then asked to look at the table on question sheet 2, in which the estimates of the attributes for the two cost categories were supplied. The attribute estimates used represent the work of the Natural Resources Management Program at Simon Fraser University (see Natural Resources Management Program 1990), in which they undertook fiscal and employment impact assessments to estimate the costs associated with the Valhalla Society wilderness preservation proposal (see Valhalla Society 1988). Again, they were reminded that the numbers represent estimates of direct impacts only. They were also informed that the numbers include only the first year impacts from the wilderness protection proposal. To give respondents an idea of the magnitude of the costs to the provincial treasury over the long term, also specified in the table were estimates of in perpetuity provincial revenue losses at different interest rates. The ranges were then explained for the first case (question sheets 2 to 8). For case 1, the best case scenario (no preservation) from a cost perspective assumes that provincial revenue and employment will remain unchanged (i.e., status quo without consideration of fall down effect) and the worst case (double the preservation level) from a cost perspective assumes that losses in revenue and employment will be attributable to losses associated with preservation only. The ranges for case 1 thus represented the differences in each cost category with and without doubling the existing preservation level in the absence of any fall down effect. Respondents were also told that besides ignoring the compounding impacts stemming from the forest industry's increasing reliance on lower volumed second growth, the Natural Resources Management Program (1990) study's cost estimates used in our workshop also ignore mitigating factors which may offset to some degree the costs associated with preservation. These include increases in tourism, enhancing the timber supply through more intensive silvicultural management and rehabilitation of NSR (not satisfactorily re-stocked) land, increasing economic returns and employment from the timber volumes currently harvested  170 through value-added wood processing and easing the socio-economic impacts arising from implementation of the preservation proposal through gradual annual allowable cut reductions. After describing the compounding and offsetting factors, the relative cost functions and the ranking and weighting procedures designed to derive them were then explained. Assigning relative cost functions to each cost category was done using the swing weighting scheme. Respondents were first asked to carefully compare differences between the worst (losses of $25 million or 2554 jobs) and the best (no change) impact on the two cost categories. They were then asked to assign swing ranks to the cost categories, assigning 1 to that cost category for which the change from the best to the worst impact constituted the greatest detriment and a swing rank of 2 to that cost category which they judged to be of lesser significance. After providing the swing ranks, respondents were asked to assign 100 points to the cost category which they ranked 1. The cost category ranked 2 was then to be assigned a scaled down swing weight whose relative magnitude reflected the comparative degree of loss when moving from the best to the worst impact. For example 76 , if a respondent felt that a job loss of 2554 employees was twice as significant (i.e., twice as severe) as a provincial revenue loss of $25 million, he or she would assign a respective rank and rating of 1 and 100 to the cost category of employment and 2 and 50 to the cost category of provincial revenue. Once all the respondents were finished assigning their swing ranks and weights, they were asked to look at question sheet 3, as well as review the benefit categories hierarchy. Benefits Associated with Increased Preservation After defining the three fundamental benefit categories and reviewing the sub-categories underlying each, the common proxy attribute measure for the three was described. Deriving and employing different attributes for each benefit category is complicated in the case of wilderness attributes because disparate measures would need to be derived for the various distinct lower level values and because meaningful attributes can not be easily derived for the 76Respondents were given illustrative examples with numbers like this for assessments throughout the workshop.  171 more elusive values. The resealing procedure would also be a difficult undertaking in the face of such complexities. To simplify the derivation of relative benefit functions, therefore, the proxy attribute of percent of B.0 preserved as wilderness was used as a performance measure for human demand-related values, human spiritual values and ecological values alike. As such, the worst impact