Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessment of crop losses from ozone using biomonitor plants and risk estimates by experts Brown, Gordon Lindal 1990

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

Item Metadata

Download

Media
831-UBC_1990_A1 B76.pdf [ 14.56MB ]
Metadata
JSON: 831-1.0098765.json
JSON-LD: 831-1.0098765-ld.json
RDF/XML (Pretty): 831-1.0098765-rdf.xml
RDF/JSON: 831-1.0098765-rdf.json
Turtle: 831-1.0098765-turtle.txt
N-Triples: 831-1.0098765-rdf-ntriples.txt
Original Record: 831-1.0098765-source.json
Full Text
831-1.0098765-fulltext.txt
Citation
831-1.0098765.ris

Full Text

ASSESSMENT OF CROP LOSSES FROM OZONE USING BIOMONITOR PLANTS AND RISK ESTIMATES BY EXPERTS By G O R D O N L I N D A L B R O W N B.Sc, The University of Manitoba, 1973 M.N .R .M. , The University of Manitoba, 1977 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y in T H E F A C U L T Y O F G R A D U A T E STUDIES (Resource Management Science) We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F BRIT ISH C O L U M B I A June 1990 0 Gordon Lindal Brown, 1990 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. I n t e r d i s c i p l i n a r y S t u d i e s -Department of R e s o u r c e Management S c i e n c e The University of British Columbia Vancouver, Canada Date S e p t e m b e r 10, 1990 DE-6 (2/88) ABSTRACT Environmental policy makers are required to make decisions under uncertainty regarding the benefits and costs of specific regulatory action. Uncertainty is a phenomenon that cannot be avoided in the assessment of environmental impacts, due to the inherent stochasticity of environmental systems, as well as a lack of adequate empirical data related to specific cause and effect relationships. A primary constraint associated with generation of adequate data from experiments is that environmental research is expensive, and conclusive results may take several years to obtain. In the meantime, significant impacts could be occurring, virtually undetected. A high degree of uncertainty exists in the assessment of the potential effects of ozone ( 0 3 ) pollution on agricultural crop yield. Thus, the purpose of this research was to provide information related to the potential impacts of 0 3 pollution on crops in the Fraser Valley east of Vancouver, British Columbia, Canada. Two alternate methods were utilized: (i) biomon-itoring with Bel W-3 tobacco, a plant variety that is very sensitive to 0 3, and (ii) expert judgments of the risks of crop losses from 0 3. The biomonitor survey was conducted over three growing seasons (1985 - 1987), in which ambient 0 3 pollution conditions were atypically low, limiting the injury response data obtained. However, a correlation was established between biomonitor injury response and ambient 0 3 levels, demonstrating that phytotoxic pollution conditions occurred during these years. Calibration of biomonitor injury response with crop yield losses revealed the following: (i) yield losses due to 0 3 exposure are likely in the event that biomonitor plants exhibit C y induced injury symptoms, and (ii) the absence of biomonitor injury does not preclude the possibility of crop loss, since the 0 3 exposure threshold for biomonitor injury may exceed that for loss of certain crops. Although experts are commonly used to provide judgments of potential impacts under uncertainty, there is a paucity of information regarding the desirable attributes of expertise. Selection of experts is largely an ambiguous task, and choices of experts by different persons are likely to be inconsistent. Prior to selection of experts for this project, a comprehensive survey was conducted of over 200 environmental professionals to determine the characteristics of an expert in 0 3 effects on crops. It was shown that expertise in this area involves a considerable number of attributes. These were grouped, using factor analysis, into seven independent dimensions: education, type of career experience, length of career experience, cognitive skills, personal qualities related to credibility, scientific recognition and involvement in the scientific community. In general, there was agreement between d i f f e r e n t groups (e.g., Fesearch scientists and members of conservation groups) regarding the relative importance of the various dimensions of expertise. Nine crop loss experts were selected, based on nomination by a large group (166) of their scientific peers. It was demonstrated with regression analysis that nominated experts exhibited the attributes identified in the survey. Logit models were estimated that predict an individual's degree of expertise in 0 3 effects on crops, based on specific attributes possessed by that individual. Independent judgments were then obtained from the nine experts regarding probable crop losses under typical 0 3 pollution conditions in the Fraser Valley. Probabilistic judgments of crop losses were generally similar among experts and approximated the level of crop losses predicted from the biomonitor survey. Limited empirical exposure-response information for Fraser Valley crops indicated that some cultivars may be more sensitive than assumed by the experts. Additional exposure-response experiments will be required to determine the source of this inconsistency. i i i TABLE OF CONTENTS Page Abstract 1 1 Table of Contents iv List of Figures viii List of Tables x Acknowledgments xi 1. INTRODUCTION AND RESEARCH OBJECTIVES 1 1.1 Background to the Problem 1 1.2 Problem Statement 7 1.2.1 Assessment Using Biomonitor Plants 7 1.2.2 Risk Assessment by Experts 8 1.3 Research Objectives 10 1.4 Outline of the Thesis 11 2. SOURCES OF UNCERTAINTY ASSOCIATED WITH CROP LOSS ASSESSMENT 12 2.1 Introduction 12 2.2 Characteristics of Anthropogenic Ozone 15 2.3 Sources of Predictive Uncertainty in Crop Loss Assessment 16 2.3.1 Stochastic Uncertainty 18 2.3.2 Scientific Uncertainty 19 2.3.3 Analytical Uncertainty 44 2.4 Epilogue 52 3. CROP LOSS ASSESSMENT USING A BIOMONITOR PLANT 53 3.1 Role of Biological Systems in Crop Loss Assessment 53 3.2 Objectives and Rationale for the Biomonitor Project 56 iv TABLE OF CONTENTS (Continued) Page 3.2.1 Phytotoxic 0 3 Trends 56 3.2.2 Calibration of Biomonitor Response with Crop Response 58 3.3 Methods and Materials 61 3.3.1 Cultivation and Standardization 61 3.3.2 Fraser Valley Biomonitor Sites 64 3.3.3 Calibration of Biomonitor Response With Crop Response 66 3.3.4 Symptomatology and Injury Assessment 68 3.3.5 0 3 Exposure Indices 69 3.3.6 Data Analysis 70 3.4 Results and Discussion 71 3.4.1 Fraser Valley Biomonitor Survey 71 3.4.2 Crop Calibration Experiment 83 4. DESIGN OF EXPERT SURVEY 91 4.1 Attributes of Expertise 91 4.2 Theoretical Model of an Expert 98 4.3 Design and Testing of Questionnaire 100 4.4 Sampling Frame and Distribution of Questionnaire 103 5. EXPERT SURVEY: RESULTS AND DISCUSSION 105 5.1 Description of Sample and Reponse Rate 105 5.2 Descriptive Statistics on Attributes of Expertise 109 5.2.1 Descriptive Statistics: Substantive Expertise 110 5.2.2 Descriptive Statistics: Normative Expertise 119 5.2.3 Descriptive Statistics: External Credibility 130 5.2.4 Summary of Importance of Indicators 130 5.3 Underlying Attributes of Expertise 134 5.3.1 Purpose and Methods 134 5.3.2 Results of Stage I Factor Analysis 135 v TABLE OF CONTENTS (Continued) Page 5.3.3 Results of Stage II Factor Analysis 157 5.3.4 Construct Validity of the Questionnaire 162 5.4 Comparison of Expert Profile with Experts Nominated by Peers 163 5.4.1 Purpose and Method 163 5.4.2 Results 165 5.5 Predicting the Probability of Being Nominated as an Expert 170 6. EXPERT JUDGEMENTS OF OZONE EFFECTS ON ERASER VALLEY CROPS 174 6.1 Rationale for Use of Experts 174 6.2 The Concept of Subjective Probability 176 6.3 Risk Assessment Methods 177 6.4 Selection of Experts 178 6.4.1 Selection of Experts in Previous Studies 178 6.4.2 Selection of Experts in Present Research 180 6.5 Design of Expert Risk Assessment Questionnaire 181 6.6 Results of Expert Risk Assessment Survey 183 6.6.1 Expert Biographical Information 183 6.6.2 Relative Sensitivity of Crops 185 6.6.3 Mean Crop Loss Estimates 185 6.7 Individual and Aggregate Crop Loss Models 190 7. GENERAL DISCUSSION AND APPLICATION: PREDICTED CROP LOSSES IN THE ERASER VALLEY 199 7.1 Introduction 199 7.2 Fraser Valley Crop Loss Assessment 202 7.2.1 Biomonitor Project 203 7.2.2 Crop Loss Risk Assessment by Experts 207 7.2.3 Comparison of Crop Loss Assessment Models 213 7.2.4 Predicted Crop Losses in the Fraser Valley 218 vi TABLE OF CONTENTS (Continued) Page 8. RESEARCH SUMMARY AND CONTRIBUTIONS 222 BIBLIOGRAPHY 229 Appendix A: Questionnaire to Determine Attributes of an Expert in Ozone Effects on Crops Appendix B: Crop Loss Questionnaire Appendix C: Descriptive Statistics vii LIST OF FIGURES Page Figure 1-1 Location of Fraser Valley, British Columbia 2 Figure 1-2 Modes of Inquiry Involved in Environmental Decision Making 6 Figure 2-1 Reactions of Plants as Individuals, Populations, and Communities to Ozone, Depending on Various Response-Influencing Factors 13 Figure 2-2 Sources of Uncertainty in Assessment of Crop Losses from 0 3 Air Pollution . . 17 Figure 2-3 Original Open-Top Field Chamber Design 38 Figure 2-4 Layout of Zonal Air Pollution System at University of British Columbia 41 Figure 3-1 Diagram of a Standardized Cultivation Set for Biomonitor Plants 62 Figure 3-2 Location of Biomonitor Stations and Ambient Air Quality Monitors 65 Figure 3-3 Biomonitor Leaf Injury Index - 1985 72 Figure 3-4 Biomonitor Leaf Injury Index - 1986 74 Figure 3-5 Biomonitor Leaf Injury Index - 1987 75 Figure 3-6 Relative Biomonitor Injury in 1985 76 Figure 3-7 Relative Biomonitor Injury in 1986 78 Figure 3-8 Relative Biomonitor Injury in 1987 79 Figure 3-9 Linear, Exponential, Reciprocal Yield-Injury Models 88 Figure 6-1 Crop Loss Risk Estimates by Experts: Green Bean 192 Figure 6-2 Crop Loss Risk Estimates by Experts: Pea 193 Figure 6-3 Crop Loss Risk Estimates by Experts: Potato 194 Figure 6-4 Crop Loss Risk Estimates by Experts: Broccoli 195 Figure 6-5 Crop Loss Risk Estimates by Experts: Raspberry 196 viii LIST OF FIGURES (continued) Page Figure 6-6 Crop Loss Risk Estimates by Experts: Corn 197 Figure 6-7 Crop Loss Risk Estimates by Experts: Forage 198 Figure 7-1 0 3 Air Quality Trends in Lower Mainland of British Columbia 204 ix LIST OF TABLES Page Table 2-1 Commonly Used 0 3 Exposure Indices in Crop Response Experiments 24 Table 3-1 Location of Biomonitor Sites by Year 67 Table 3-2 Correlations Between Fraser Valley Weekly Leaf Injury Indices (LH) and Various 0 3 Indices 81 Table 3-3 Leaf Injury Index by Biomonitoring Period: UBC 1986 84 Table 3-4 Relationship Between UBC Weekly Leaf Injury Indices (LII) and Various 0 3 Indices 85 Table 3-5 Comparison of LII and Crop Yields in 1986 UBC Experiment 87 Table 3-6 Residual Sums of Squares (RSS) Associated with Biomonitor Models 89 Table 5-1 Questionnaire Response by Group 106 Table 5-2 Description of Sample 108 Table 5-3 Descriptive Statistics: Indicators of Substantive Expertise Il l Table 5-4 Descriptive Statistics: Indicators of Normative Expertise 120 Table 5-5 Descriptive Statistics: Indicators of External Credibility 131 Table 5-6 Summary of Importance of Expert Indicators 133 Table 5-7 Results of Factor Analysis on "Education (Degree Obtained)" 137 Table 5-8 Results of Factor Analysis on "Education (Discipline)" 139 Table 5-9 Results of Factor Analysis on "Academic Performance (Grade)" 141 Table 5-10 Results of Factor Analysis on "Length of Career Experience" 142 Table 5-11 Results of Factor Analysis on "Scientific Recognition (Publications)" 144 Table 5-12 Results of Factor Analysis on "Type of Career Experience" 146 Table 5-13 Results of Factor Analysis on "Cognitive Abilities" 149 x LIST OF TABLES (continued) Page Table 5-14 Results of Factor Analysis on "Personal Qualities" 152 Table 5-15 Relative Importance of Stage I Factors 155 Table 5-16 Groups with Significantly Different Factor Scores 156 Table 5-17 Results of Stage II Factor Analysis 158 Table 5-18 Independent Variables Used in Regression Analysis 164 Table 5-19 Nominated Individuals in Expert Attributes Model 166 Table 5-20 Results of Regression Analysis Predicting Logarithms of Number of Nominations 168 Table 5-21 Analysis of Collinearity in Regression Model 169 Table 5-22 Logit Model Parameters and Results for Predicting the Probability of Receiving a Specified Number of Expert Nominations 172 Table 5-23 Comparison of Nomination Probabilities from Logit Model to Actual Nominations 173 Table 6-1 Experts Selected for Crop Loss Risk Assessment 182 Table 6-2 Expert Respondent Information 184 Table 6-3 Expert Judgments of Relative Sensitivity of Crops to 0 3 186 Table 6-4 Expert Aggregate Crop Loss Estimates: Low 0 3 Scenario 187 Table 6-5 Expert Aggregate Crop Loss Estimates: Medium 0 3 Scenario . . 188 Table 6-6 Expert Aggregate Crop Loss Estimates: High 0 3 Scenario 189 Table 7-1 Crop Losses Predicted by Various Models 215 Table 7-2 Crop Production Volumes and Value: Fraser Valley 219 Table 7-3 Potential Gross Crop Losses in Fraser Valley 220 xi ACKNOWLEDGEMENTS I am grateful to my two supervisors, Drs. Vic Runeckles and Ilan Vertinsky, for their considerable support and guidance. I appreciate the direction and assistance provided by the other members of my committee, Drs. Douw Steyn, Peter Nemetz and Les Lavkulich. Several individuals provided technical assistance, and I wish to acknowledge their contribution. Dr. Vinay Kanetkar and Mrs. Ying Kwan assisted with the data analysis. Drs. John Collins, Don Wehrung, Shelby Brummelle, Ken McCrimmon, Neil Guppy and George Grey assisted with design of the two research questionnaires. Ms. Elaine Wright provided crop loss data from her experiments at U.B.C., and Mr. Al Percival of the Greater Vancouver Regional District provided historical ozone air quality data for the Lower Mainland. I am grateful to Pierrette Ste-Germaine, Gail Viner and Mary-Ann Gaitan, who competently shared in the typing of the manuscript, and to Rick Connery, who assisted with the graphics. Financial assistance in the form of scholarships was provided by the Canadian Environmental Assessment Research Council, the Air and Waste Management Association, and the University of British Columbia. Additional financial assistance was provided by Western Research of Calgary, Alberta. Finally, I wish to acknowledge the support of my wife, Pam, who consistently provides encouragement throughout all our experiences, and to Mr. Elmer Berlie, for providing much of the initial inspiration to undertake the Ph.D program. xii CHAPTER ONE INTRODUCTION AND RESEARCH OBJECTIVES 1.1 Background to the Problem Air quality monitoring in the Fraser Valley east of Vancouver, British Columbia, Canada (Figure 1-1) has demonstrated that Canadian air quality objectives for photochemical oxidants as ozone (03) are regularly exceeded during the spring and summer months.1 Agricultural crops represent one class of potentially sensitive receptors in this region, and local government authorities have expressed concern regarding the potential impact of 0 3 on crops (Wilson et al., 1984). Although there is a considerable amount of scientific information available regarding the effects of 0 3 on different crops, very limited exposure-response information has been obtained for Fraser Valley Crops and growing conditions (Runeckles and Wright, 1988; Wright, 1988). The use of indirect exposure-response information (obtained for different crops and/or growing conditions) is subject to considerable uncertainty because of differences in response between and among crop species, and because local growing conditions can dramatically affect the response of vegetation to 0 3 (National Research Council, 1977; U.S. Environmental Protection Agency (U.S. EPA), 1987; Heck et al., 1988). Although 0 3 is a natural constituent of the troposphere, it has been identified as an important air pollutant since the late 1950's. Tropospheric 0 3 is a secondary air pollutant which results from photochemical atmospheric reactions involving nitric oxides and reactive 1 Canadian air quality objectives for ozone are described in Chapter Two. 1 FIGURE 1-1 LOCATION OF FRASER VALLEY, BR IT I SH COLUMBIA hydrocarbons. Solar radiation plays a crucial role in the formation of 0 3 ) which builds up to maximum concentrations on warm, sunny, stagnant days. In the Fraser Valley of British Columbia, peak 0 3 concentrations are often observed during the primary agricultural growing season, May to September (Wilson et al., 1984). As a result of man's increasing use of fossil fuels, particularly petroleum and natural gas, and the widespread use of the internal combustion engine, the atmospheric burden of the oxides of nitrogen and hydrocarbon precursors of tropospheric 0 3 has increased. Photochemical reactions leading to the formation of 0 3 in the lower layers of the troposphere are now commonplace throughout the world, particularly in large urban areas with high vehicle densities. In spite of the fact that 0 3 is a highly reactive gas, it may be transported over considerable distances when diluted in air, depending upon meteorological conditions. It often moves away from its urban origins into rural areas, and elevated levels of 0 3 may be encountered hundreds of kilometres from major urban centres (U.S. EPA, 1986). While the phytotoxicity of 0 3 was first documented in southern California approximately 30 years ago, it is now known that 0 3 air pollution affects vegetation more than any other single air pollutant (National Research Council, 1977; U.S. EPA, 1978; U.S. EPA, 1986). The effects of 0 3 at the individual plant level are well documented for several species (Heck et al., 1988). Numerous studies over the past 20 years or so have clearly identified 0 3 as the source of important chronic exposure effects that result in impairment of plant growth and hence productivity. Such chronic exposure effects frequently do not involve visible symptoms and thus are difficult to detect. Concern and uncertainty regarding the possible magnitude of this problem resulted in the establishment of the National Crop Loss Assessment Network 3 (NCLAN) research program of the U.S. Environmental Protection Agency (Heck et al., 1988). In the United States, the monetary value (economic surplus) of loss estimates for agricultural crops due to 0 3 ranges from $2.2 to $3.0 billion annually (U.S. EPA, 1986). Annual agricultural benefits in the range of $500 million to $1 billion (1986 $) would be expected to result from increased productivity of major crops in the United States, if 0 3 concentrations were reduced by 25% of the difference between current and background levels (U.S. Congress, Office of Technology Assessment, 1989). In Canada, elevated levels of 0 3 have been observed in the Lower Mainland of British Columbia, southern Ontario and Quebec, Nova Scotia, and New Brunswick (Wilson et al., 1984). In Ontario, it was estimated that agricultural crop losses due to 0 3 result in losses in the range of $15 to $23 million annually (1980 $) (Ontario Ministry of the Environment, 1984). An unpublished B.C. government report recently predicted that, in the Lower Mainland of B.C., the annual value of crop losses from 0 3 will be approximately $8.9 million (Rafiq, 1986). It was acknowledged that substantial uncertainty was associated with this estimate, due to the lack of needed exposure-response information for local crops. Accurate estimates of losses and associated economic value rely upon a good knowledge of exposure-response relationships. The use of inappropriate crop loss models can potentially result in errors in predicted economic losses equivalent to millions of dollars on a regional or national scale. Unfortunately, considerable uncertainty is associated with the predictions of crop losses (Krupa and Teng, 1982; Krupa and Nosal, 1989). Nevertheless, due to the potential magnitude of the problem, assessment of the impact of elevated levels of 0 3 cannot be delayed while the scientific community reduces the uncertainty associated with crop loss models. Delay in the 4 assessment of air pollution effects on agricultural resources is risky, as stated by D.S. Lang of the Minnesota Environmental Quality Board (MEQB, 1984): The (MEQB) is charged with the broad responsibility of environmental impact assessment. It is the primary objective of this Board to mitigate adverse impacts for the purpose of improving and maintaining the quality of the environment. This is a difficult but manageable task when the character of the impact is acute causing directly measurable effects and occurs over a short period of time. When the character of the impact is indirect and occurs over a long period of time, it is often neglected because the means of assessment are not available. While impacts remain unknown, speculation and controversy result with the effect of delaying the assessment process. The potential costs of neglected or delayed impact assessment of chronic environmental problems is extremely high and may well prove irrecoverable at the time the true magnitude of the problem becomes known. It can be expected that during the next several decades many environmental problems will result from lack of information and understanding at a previous time. Because our current understanding of the significance of these problems is exceeded by our technological abilities to impose problems, it is behooving for agencies with environ-mental responsibilities to act in an anticipatory manner to best represent the public interest. A decision making model proposed by Beanlands and Duinker (1983) provides some insight as to how chronic environmental problems involving substantial scientific uncertainty should be addressed. The various modes of inquiry associated with environmental decision making are portrayed in Figure 1-2. The classical experimental approach to scientific inquiry, Mode 1, involves strict scientific methodologies and control of variables. Experimental results are generally consistent, but are frequently of limited use in environmental policy because they rarely embody the complex mix of real world conditions. Unaided intuitive judgment, Mode 6, involves an uncertain data base, no control of variables, no statistical control and inconsistent logic rules which are seldom made explicit. Preferred modes of inquiry in environmental decision making are Modes 3 or 4, MODE OF COGNITION ANALYTICAL I N T U I T I V E 11) TRUE Y. EXPERIMENT __J (2) HYBRID] EXPERIMENTAL (STATISTICAL METHODJf Y/A I3J QUASI-EXPERIMENTt AIOEOl JUDGMEN V7 INTUITIV" JUOGMENT 7] (0ATAr /frNOWNj JZZZZ. COVERTNESS OF PROCESS [NTUITIV IJUDGMENTT/ 3 u l A T A [ 7 / „ JNKNOWNJlX CONFLICT-REDUCING CONFLICT-PRODUCING CONFLICT POTENTIAL F IGURE 1-2 MODES OF INQUIRY INVOLVED IN ENVIRONMENTAL DEC IS ION-MAK ING (FROM BEANLANDS AND DUINKER. 1983) 6 involving some experimentation and statistical analysis, computer simulation modelling and analysis of expert opinion and judgment (Beanlands and Duinker, 1983). 1.2 Problem Statement Given the limited exposure-response data presently available for locally important crops (Runeckles and Wright, 1988; Wright, 1988), two methods were proposed in the present research for providing additional information relevant to the assessment of crop losses in the Fraser Valley. These methods, which relate to the third and fourth mode of inquiry in Beanlands and Duinker's (1983) environmental decision making model (Figure 1-2), include: (i) use of biomonitor plants in important agricultural areas throughout the Lower Mainland, and (ii) environmental risk assessment involving crop loss experts. While each of these methods has the potential to provide valuable information, both methods are subject to certain limitations, as described below. 1.2.1 Assessment Using Biomonitor Plants The nature and magnitude of vegetation response to air pollution can be dramatically affected by numerous environmental factors, such as time of day, weather conditions, water and nutrient availability, and soil type and fertility. Ambient air quality data provide information about pollution levels only; they do not provide information about the influence of environmental conditions that can modify plant response to 03. Because plants integrate the effects of pollution with climatic, edaphic and biotic factors that can modify response, they can 7 provide direct information regarding the impact that air pollution can have on the biological components of the environment (Posthumus, 1982; MEQB, 1984; Tingey, 1989). Despite its intuitive appeal, the usefulness of visible biomonitor injury data is generally limited to provision of qualitative information regarding possible vegetation effects (Jacobson and Feder, 1974). Because of the influence of local environmental conditions, the injury response of biomonitor plants cannot be used to predict 0 3 concentrations accurately. However, biomonitor injury observations can be used as an "early warning system" (Posthumus, 1982) of potentially phytotoxic 0 3 levels. Biomonitor injury has not been quantitatively related, or calibrated, to yield reduction of important agricultural crops. Evidence of such a relationship would greatly improve the value of injury response data obtained from biomonitor plants. 1.2.2 Risk Assessment by Experts Environmental risk assessment has been proposed as a vehicle for improving the communication of scientific information between scientists and decision makers (Morgan et al., 1979; Feagans and Biller, 1981; Ruckelhaus, 1983). The method is not intended to be a substitute for scientific research; rather, it is intended to improve the usefulness of the information available from existing scientific knowledge. It does this by explicitly representing the opinions of scientific experts in the form of subjective probability distributions. The availability of such probability distributions, representing independent scientific viewpoints, allows analysts and decision makers to determine not only what is known, but also what is not known about the subject being examined. 8 Scientists are often reluctant to provide subjective information regarding environmental risk as it may appear to involve sacrificing the scientific integrity that they value very highly (Orians, 1986). This is in spite of the fact that all complex technical problems involve extensive judgment and opinion in the design of solutions (Keeney and von Winterfeldt, 1988). Since the application of scientific opinion and judgment is frequently implicit, policy analysts are usually unable to judge how confident they should be with the results or conclusions generated. There is considerable potential, through the use of environmental risk assessment methods, to expand the role and input of science and expert scientific judgment in environmental decision making. For this reason, the number of studies involving expert judgments of environmental effects under uncertainty has greatly increased in recent years (see for example, Keeney et al., 1984; Whitfield and Wallsten, 1984; Morgan et al., 1985; North et al., 1985; Peterson and Violette, 1985; Amaral, 1988). However, because the risk assessment method is relatively new and undeveloped, various methodological problems are presently associated with its use. One apparent deficiency is that the desirable attributes of an "expert" are not presently documented. Although the quality of data collected clearly depends on the competence of experts that are selected, there is a paucity of information to provide guidance in the selection of experts for risk assessment. An increased understanding of the necessary and desirable attributes of expertise and provision of a systematic means for selecting experts would increase the credibility of the environmental risk assessment method. 9 1.3 Research Objectives The first objective of the research was to utilize two methods of providing information relevant to assessment of agricultural crop losses as a result of 0 3 pollution in the Fraser Valley of British Columbia. These methods were: (i) the use of biomonitor plants to provide evidence of plant injury from ambient 0 3, and (ii) the provision of expert judgments regarding the risk of crop losses under ambient 0 3 conditions. As an understanding of the attributes of expertise was essential to the selection of experts for the Fraser Valley crop loss risk assessment, the second objective of the research was to identify and document the attributes of an expert in 0 3 effects on agricultural crop yields. Related sub-objectives were to: (i) determine whether there was consensus within the scientific community regarding the profile of an expert, (ii) determine whether the norms held by the scientific community regarding the desirable attributes of expertise were reflected in the ranking of experts by peer nomination, and (iii) develop a statistical model to assist in the selection of experts for risk assessment of agricultural crop losses from 0 3 air pollution. 10 1.4 Outline of the Thesis Chapter Two discusses sources of uncertainty associated with exposure-response information regarding crop losses due to 0 3 air pollution. Chapter Three provides a description of the results of the biomonitor survey conducted during the summers of 1985 through 1987 in important agricultural areas throughout the Fraser Valley. It also describes an experiment designed to calibrate the response of the biomonitor plant with yield losses of two agricultural crops. Chapters Four and Five describe the design and findings, respectively, of a survey conducted to obtain information on the desirable attributes of crop loss experts, and to nominate top experts in crop losses from 0 3 air pollution. Chapter Six describes the results of the Fraser Valley crop loss risk assessment by nominated experts. Chapter Seven is a general discussion of the research and results, and includes an assessment of probable agricultural crop losses due to 0 3 in the Fraser Valley, and the approximate monetary value of the crop losses. Chapter Eight summarizes the findings and contributions of the research. 11 CHAPTER TWO SOURCES OF UNCERTAINTY ASSOCIATED WITH CROP LOSS ASSESSMENT 2.1 Introduction The prediction of plant responses to an air pollutant is a complex and uncertain task (Runeckles and Brown, 1986). The reaction of a plant to air pollution, as described by Guderian et al. (1985), depends on the ambient exposure, the amount of pollutant diffusing into the leaves, and the plant's autonomous and environmentally modified resistance. The degree of resistance is controlled primarily by the plant's genetic complement and its developmental stage at the time of exposure, and by the modifying influence of various external factors such as climate, soil and biotic factors. The response can range from acute or chronic effects in individual plants to changes in plant communities, their composition and structure. The relationships among ambient exposures, environmental factors and plant responses are illustrated in Figure 2-1. Past research has shown that ozone (03) pollution results in visible symptoms of injury to sensitive vegetation when the ambient 0 3 level exceeds a threshold of about 40 parts per billion (ppb)2, maintained for approximately four hours (Heck et al., 1966; Ashmore et al., 1978). Research has also clearly shown that important chronic exposure effects may occur, which may or may not involve visible symptoms, but which lead to effects on plant growth and productivity (U.S. EPA, 1986). It is this latter awareness that led to the recent extensive research effort on 0 3 effects on crops, in particular the large National Crop Loss 1 1 ppb is equivalent to 1 nL/L. 12 FREQUENCY OF OCCURRENCE SOIL CLIMATE BIOTIC INTERFERENCES 1 I ND IV IDUAL PLANTS OR POPULATIONS SUBTLE CHRONIC ACUTE AMBIENT OZONE CONCENTRATION I EXPOSURE LOW INTERMEDIATE HIGH T PLANT MODE OF ACTION \ EFFECTS DURATION OF EXPOSURE GENETIC CHARACTERISTIC DEVELOPMENTAL STAGE 1 PLANT COMMUNITIES NO ALTERATIONS EXTENSIVE SIGNIFICANT IN STRUCTURE SIMPLIFICATION ALTERATIONS AND UP TO TOTAL COMPOSITION - DESTRUCTION IMPAIRMENT OF ECOLOGICAL FUNCTION IMPAIRMENT OF ECONOMIC PERFORMANCE IMPAIRMENT OF GENETIC RESOURCES SOURCE: MODIFIED FROM GUOERIAN et a l . 1985 F IGURE 2-1 REACTIONS OF PLANTS AS IND IV IDUALS . POPULATIONS, AND COMMUNITIES TO OZONE, DEPENDING ON VARIOUS RESPONSE - INFLUENC ING FACTORS 13 Assessment Network (NCLAN) program of the U.S. Environmental Protection Agency (Heck et al., 1988). The exposure-response concept is the primary experimental approach which can be used for purposes of predicting the effects of air pollutants on plants. Exposure-response information is needed for setting air quality standards and for the development of models which predict plant damage or yield reductions as a function of pollutant loading (Heck, 1982). Several empirical exposure-response models for 0 3 effects on various crop species have been described, involving both linear and nonlinear relationships (U.S. EPA, 1986; Heck et al., 1988; Lee et al., 1987; Runeckles and Wright, 1989). There have been many attempts to construct a "usable bridge" between various expressions of "average" air pollution concentrations and the effects of specific exposures to plants, with varying degrees of success (Runeckles and Brown, 1986; Krupa and Kickert, 1987). A major problem relates to the choice of the pollutant exposure term used, since the exposure term is a surrogate for the true dose to which the plants are exposed. As pointed out by Guderian et al. (1985), the decisive factor in protecting vegetation is to maintain the ambient pollutant exposure below the threshold effects level; this requires an accurate knowledge of both the ambient exposure characteristics and the resulting responses. Although an extensive research effort has been directed at the effects of ambient 0 3 on major agricultural crops (U.S. EPA, 1986; Heck et al., 1988), significant sources of uncertainty associated with the use of empirical models to predict crop losses from 0 3 still exist (Krupa and Nosal, 1989). 14 2.2 Characteristics of Anthropogenic Ozone The general terms "oxidants" and "photochemical air pollutants" include a large number of components which are formed as reaction products of certain primary pollutants such as nitric oxide (NO), nitrogen dioxide (N02) and reactive hydrocarbons (RHC), in the presence of sunlight. Major reaction products (secondary pollutants) include 0 3, peroxyacetyl nitrate (PAN), higher oxides of nitrogen, aldehydes and ketones, and several gaseous and particle-bound organic and inorganic acids. Of all these components, 0 3 is generally the most important phytotoxic constituent. While PAN is more phytotoxic than 0 3, elevated levels of 0 3 are more widely distributed and generally occur in higher ambient concentrations (Guderian et al., 1985). Although much of the public concern about 0 3 is presently focused upon its depletion in the stratosphere, its elevated levels in the troposphere near the earth's surface represent one of our most important air pollution problems today (National Research Council, 1977; U.S. EPA, 1986). Evidence exists that 0 3 is capable of being carried long distances without diminishing its phytotoxic concentration (Manning and Feder, 1980; Lefohn, 1984). Ozone concentrations in rural areas frequently exceed those in upwind urban airsheds, due to photo-chemical induction time, physical transport processes, additivity of polluted air masses and reduced efficacy of 0 3 scavenging processes (Taylor, 1984). Since 0 3 levels vary significantly in both space and time, values that should be assumed for the purpose of regional impact assessment are uncertain. In urban situations, the buildup of 0 3 levels during the day typically starts with a buildup of nitrogen oxides and hydrocarbons in the morning, leading to an 0 3 peak in the early afternoon. At rural sites, an early morning (sunrise) increase in ground level 0 3 is frequently 15 observed; this has been suggested as being due to downmixing from aloft (Kelly et al., 1984). The subsequent increase in 0 3 is due to photochemical reactions of precursor pollutants and further downmixing. Ozone depletion at night in both urban and rural areas generally occurs as a result of chemical scavenging reactions primarily involving nitric oxide (NO), while nocturnal inversions prevent mixing with air from aloft (Kelly et al., 1984). However, persistence of 0 3 at night in rural areas is possible where the air typically contains smaller amounts of NO and other compounds that scavenge 0 3 (Jacobson, 1982). A non-linear relationship exists between 0 3 and its precursor pollutants. As a result, changes in hydrocarbon and/or nitrogen oxide emissions rarely produce equal changes in 0 3 levels (Horowitz, 1982). Primary pollutants such as sulphur dioxide (S02) and nitrogen oxides (NOx), on the other hand, tend to be linearly related to emissions near a point source. 2.3 Sources of Predictive Uncertainty in Crop Loss Assessment According to Rowe (1977), Turner (1985) and Suter II et al. (1987), there are three major categories of uncertainty related to environmental risk assessment, including stochastic, scientific and analytical uncertainty. In the following discussion, uncertainties associated with crop loss impact/risk assessment are described under each of these three categories. Figure 2-2 summarizes the various sources of uncertainty described in the text. 16 PREDICTIVE UNCERTAINTY m STOCHASTIC UNCERTAINTY NATURAL VARIATION IN PHYSICAL AND BIOLOGICAL SYSTEMS S C I E N T I F I C UNCERTAINTY DESCRIPTIVE UNCERTAINTY I MECHANISM OF PLANT INJURY 0 3 FXPOSURE INDEX ANALYTICAL UNCERTAINTY MEASUREMENT UNCERTAINTY I EXPERIMENTAL DESIGN INDIRECT EMPIRICAL DATA • LIMITED MODEL VARIABLES • MODEL FUNCTIONAL FORM BACKGROUND 0 3 LEVEL F IGURE 2-2 SOURCES OF UNCERTA INTY IN ASSESSMENT OF CROP LOSSES FROM OZONE A IR POLLUT ION 2.3.1 Stochastic Uncertainty Many of the difficulties associated with predicting the effects of pollutants are due to the inherent variability in natural physical and biological systems (Krupa and Teng, 1982; Beanlands and Duinker, 1983). Inherent stochasticity in these systems leads to difficulties in defining the value of certain state variables that should be assumed for impact assessment, including meteorological conditions, pollution levels, growing conditions and seasonal crop yields. The natural variability inherent in many physical and biological systems involves seasonal fluctuations and multi-year cycles, with the result that some variables may change over time. As a result of the overall natural variation or "noise" associated with biological systems, minor or subtle pollutant effects can be very difficult to detect. Additionally, although it is believed that most natural systems operate within certain stability ranges, the time typically available for environmental research precludes anything but an approximation of the true range of natural variability (Beanlands and Duinker, 1983). Natural variability restricts the establishment of true controls under field conditions, creating serious difficulties for the application of appropriate experimental procedures. The determination of realistic loss models requires the collection of exposure-response data, at individual sites, over several years. Out of the universe of similar environmental systems, a given percentage would be expected to show an effect; this is an estimate of risk (Suter II et al., 1987). 18 2.3.2 Scientific Uncertainty Scientific uncertainty arises from lack of information, knowledge or scientific agreement regarding cause-effect relationships (Turner, 1985). The activity of science, which comprises the gathering of data, testing of hypotheses and comparison of theories, usually results in consensus among scientists about the values of measurable quantities, at least in the long run (CMU, 1984). However, where system relationships are not fully understood or where data are difficult to obtain, uncertainty and thus disagreement among scientists may exist. Scientific uncertainty itself has two main components: descriptive uncertainty and measurement uncertainty (Turner, 1985). Descriptive uncertainty refers to knowledge about what parameters and interrelationships define the system. With regard to crop loss assessment, sources of descriptive scientific uncertainty include: (i) the mechanism of plant injury from O,, and (ii) specification of a biologically realistic 0 3 exposure index for use in exposure-response models. Measurement uncertainty relates to the quantitative values to be assigned to system parameters, and thus includes measurement error. In crop loss assessment, sources of measurement uncertainty include: (i) choice of experimental design, and (ii) use of indirect empirical data (e.g., data representing a different plant species or cultivar than that being evaluated). 19 2.3.2.1 Mechanism of Plant Injury There is a lack of clear understanding of the physiological mechanisms responsible for plant response, including detoxification and repair, although a conceptual model has been proposed to explain differences in plant responses to 0 3 (Tingey and Taylor, 1982). According to this model, there are three general processes that control the magnitude and expression of injury or response: (i) processes that control the flux of gases into the leaves, (ii) processes that influence the distribution of 0 3 or its reaction products within the leaf, including 03-scavenging reactions, and (iii) processes that repair or compensate for the perturbations caused by 0 3 or its reaction products. The interaction of these processes controls the amount of injury. The relative importance of these three processes varies among plants and is influenced by internal and external conditions. Symptoms of 0 3 phytotoxicity range from characteristic visible leaf injury to changes in the distribution of growth and harvestable yield in the absence of visible injury. The mechanism of 0 3 injury probably involves either 0 3 itself or the superoxide or hydroxyl-free radicals resulting from its uptake, interacting within the cell to produce changes in membrane permea-bility (U.S. EPA, 1986). Whether or not there is direct oxidation of membrane components is still not known. Superoxide anion has a possible role in mediating 0 3 injury (Fong, 1984), and has recently been shown to occur in tissues developing 0 3 injury symptoms (Vaartnou, 1988). The linkages relating altered biochemical processes, foliar injury and plant yield are not well understood. For example, the well established effect of 0 3 in reducing photosynthesis may reflect the direct impairment of chloroplast function or reduced CO z uptake resulting from 03-induced stomatal closure, or both. Regardless of the mechanism, a sustained reduction in 20 photosynthesis will ultimately adversely affect growth, yield and vigor of the plant (Reich and Amundson, 1986; U.S. EPA, 1986). 2.3.2.2 Ozone Exposure Index Possibly the most significant source of uncertainty associated with existing exposure-response models is the difficulty in characterizing a biologically realistic exposure index that reflects the "effective dose" (Runeckles, 1974) received by the plant, as well as the pollutant concentration in ambient air (i.e., the "exposure"). Plants appear to be affected primarily by the 0 3 that diffuses into the leaves, and an effect will only occur if a sufficient amount reaches the sensitive cellular sites within the leaf. Injury will not occur if the rate of uptake is sufficiently small so that the plant is able to detoxify or metabolize 0 3 or its reaction products, or if the plant is able to repair or compensate for the perturbation (Tingey and Taylor, 1982). Factors which influence uptake cause an uncertainty in determining the actual dosage to which the plant can respond (Lefohn and Runeckles, 1987). It is difficult to measure and quantify the relationship between exposure and dose because of environmental and biological variables which can modify uptake and plant response. This phenomenon was first documented by MacDowall et al. (1964) and Mukammal (1965). Their attempts to relate various measures of exposure to sensitive tobacco leaf injury failed to yield a satisfactory response curve. However, when the method of expressing exposure was modified to include a measure of the crop's capability for absorbing 0 3 from the ambient air, based upon coefficient of evaporation and other measurements, a significantly improved exposure-response relationship was obtained. It was suggested that the coefficient of 21 evaporation was proportional to evapotranspiration and hence provided an approximate measure of gas exchange. Several researchers have shown that pollutant uptake or flux, rather than ambient concentration, can be used to define an effective dose ( Bennett, 1979; Black and Unsworth, 1979; Amiro et al., 1984). However, because it has not yet been possible to quantify the relationship between exposure and dose, it is necessary to examine the relationships which exist between various exposure characteristics and plant response (Lefohn and Runeckles, 1987). The ultimate impact of an air pollutant on vegetation depends on the integrated effect of air pollution exposure over the growing season. Effects on vegetation are dependent on the variation of pollutant concentration with time, both in the long- and short-term. Effects are dependent on the frequency of exposures, duration of exposures, length of time between exposures and the magnitude of fluctuation concentrations (Jacobson, 1982). Research has not yet clearly defined which components of exposure are most important in causing vegetation responses (Lee et al., Lefohn and Runeckles, 1987; 1987; Tingey et al., 1988). However, the following is known regarding plant response to 0 3 (modified from Lee et al., 1987): (i) Plant response increases with concentration and duration of exposure, and concentration is more important than exposure time (Heck et al., 1966). (ii) Peak concentrations are more significant than the mean exposure in determining vegetation response (Bicak, 1978; Musselman et al., 1983; Larsen and Heck, 1984; Hogsett et al., 1985; Runeckles, 1987). (iii) The concentration of peaks is probably more important than their frequency (Heck et al., 1966; Bicak, 1978; Musselman et al., 1983). 22 (iv) Low concentrations of 0 3 may predispose plants to injury from subsequent high concentrations (Heck et al., 1966), or may render them less susceptible (Runeckles and Rosen, 1977). (v) Plant response to repeated peaks may increase (Mukammal, 1965) or decrease (Nouchi and Aoki, 1979) as the time between peaks is increased. (vi) A threshold dose of 0 3 must be exceeded before plant response occurs (Amiro et al., 1984), There is a general agreement that, from the toxicological perspective, it is the peaks or concentrations above some level that are most likely to have the greatest impact (U.S. EPA, 1986). Defining the index that best relates plant response to exposure necessitates consideration not only of the method of characterizing the temporal variations, but also of the underlying biological basis for the response. The index should not be uniquely related to the specific experimentation with which it was determined, but be universal in its applicability, permitting its use with aggregates of response data obtained in different locations and years, and with different crops. Table 2-1 summarizes several 0 3 exposure indices utilized in numerical models of pollution exposure and vegetation response. Common exposure indices used to date include mean values for 1, 7, 12 or 24 hours, both daily and averaged over the growing season; seasonal maximum hourly mean concentrations; seasonal sums of absolute hourly average concentrations above a threshold (e.g., 0.08, 0.10 or 0.12 ppm); seasonal sums of fractions of hourly concentrations above a threshold, and others. In recent years, the exposure index most widely used to assess 0 3 effects on crops is the "M7"; the seasonal mean concentration of daily 7-h (0900 to 1559 h) exposures (Heck et al., 23 TABLE 2-1 Commonly Used 0 3 Exposure Indices in Crop Response Experiments Index Description Original Source M7 M12 Ml (7) Ml (12) P7 PI SUM08, SUM10 SUM 12 HRS08, HRS10 HRS12 IE08, IE10, IE12 (Integrated exposure) EFFMEAN (Effective mean) Season-long mean of daily 7-h mean (0900-1559h) Season-long mean of daily 12-h mean (0900-2059 h) Season-long mean of daily 1-h max (0900-1559 h) Season-long mean of daily 1-h max (0900-2059 h) Peak 7-h mean observed in season Peak 1-h mean observed in season Sums of the. fractions of all 0 3 concentrations during the growing season above thresholds of 80, 100 or 120 ppb Sums of the number of hours during the growing seasons when thresholds of 80, 100 or 120 ppb are exceeded Sums of the absolute 0 3 concentrations during growing season when thresholds of 80, 100 or 120 ppb are exceeded An extension of M7, which gives greater weight to peak concentration (described in text) Heck et al. (1984b) Heagle et al. (1987) Heck et al. (1984b) Wright (1988) Heck et al. (1984b) Heck et al. (1984b) Oshima et al. (1976) Lee et al. (1987) Lefohn and Benedict (1982) Larsen and Heck (1984) ...continued 24 TABLE 2-1 (cont'd) Commonly Used 0 3 Exposure Indices in Crop Response Experiments Index Description Original Source Median TOTDOSE Sigmoid function (W95, W126) GPWCI Median of growing season or growth stage 0 3 record Total 0 3 exposure (ppb-h) observed during growing season or growth stage Designed to give lower weight to low 0 3 concentrations and higher weight to high 0 3 concentrations (described in text) Generalized phenologically weighted cumulative impact (described in text) Nosal (1984) Lee et al. (1988) Lefohn and Runeckles (1987) Lee et al. (1988) Hogsett et al. (1988) 25 1984a; Heck et al., 1984b). The daily 7-h exposure period corresponds to the time of day when plants were thought to be most susceptible to 0 3 and when maximum 0 3 concentrations were thought to occur. Other researchers have recommended the integration of the full 24-h exposure when computing summary indices, since there is strong evidence that plants are susceptible to 0 3 for more than seven hours per day, and because maximum 0 3 concentrations often occur outside this 7-h window (Lee et al., 1987). In either case, the long-term average treats all concentrations with the same weight. Thus, an infinite number of hourly distributions can be used to generate the same seasonal mean, ranging from those containing many peaks to that containing none (Lefohn and Runeckles, 1987). Several researchers have shown that it is possible for two adjacent and identical sites with the same seasonal means but with different types of distributions to experience different yield losses (Musselman et al., 1983; Hogsett et al., 1985; Larsen and Heck, 1984; Lefohn et al., 1988). These studies have shown that fluctuating episodic 0 3 exposures result in greater losses than non-fluctuating daily exposures. As a result, the long-term seasonal mean is an inadequate 0 3 exposure index. An additional problem associated with the use of the mean as an exposure index is that 0 3 concentrations are not normally distributed, but are usually non-symmetrical, strongly skewed and have a high variance relative to the mean (Nosal, 1983). Ozone concentrations tend to have log-normal or Weibull distributions, and the arithmetic mean is an inappropriate summary index for these types of distributions (Nosal, 1984; Krupa and Nosal, 1989). Furthermore, 0 3 distributions often differ markedly among and between rural and urban sites, particularly with respect to the relative abundance and magnitude of the short-term peak concentrations, which 26 are of considerable importance in the exposure-response relationship (Oshima et al., 1976; Lefohn and Benedict, 1982; Lefohn and Benedict, 1985). A further limitation associated with the seasonal mean indices is that, by definition, they exclude the long-term duration of exposure, unless they are computed for a specific and standardized time interval. For example, plant response resulting from a 10-day exposure period is likely to be very different from that resulting from a 50-day exposure period, although the values of the seasonal mean may be approximately the same (Runeckles, 1988). In response to concerns about the use of means, alternative exposure indices that weight the peak concentrations more than the lower concentrations have been suggested (Table 2-1). Several studies have compared the performance of the various indices that weight the peaks with the seasonal mean indices, and there is now strong evidence that the peak indicators are better summary indices for exposure-response applications (Larsen and Heck, 1984; Lefohn and Runeckles, 1987; Lefohn et al, 1988; Lee et al., 1987; Lee et al., 1988; Hogsett et al., 1988; Tingey et al., 1988; Tingey et al., 1989). A qualification of this statement is necessary, however. Several of the studies using NCLAN data and comparing the fit of exposure-response functions have shown that the seasonal mean often provides as good a fit (in terms of coefficient of determination or minimum residual sum of squares) as any of the indices that weight the peaks (Lee et al., 1987; Lefohn et al, 1988). The reason for this is that the NCLAN exposure regime consisted of repetitive daily peaks in excess of 0.10 ppm and, with this type of exposure, which occurs only infrequently in the real world (Lefohn and Benedict, 1985), there is a high correlation between the peak indicators and the seasonal mean. Under natural conditions there is generally a low frequency of peaks > 0.10 ppm and the application of the seasonal mean, rather than one 27 which captures the influence of the peaks, would appear to be erroneous. For sites where there is a high frequency of peaks in excess of 0.10 ppm, such as within the Los Angeles region, the use of the M7 exposure index might be appropriate, although it still suffers from the problem of ignoring the actual duration of the exposures. Ozone exposure indices formulated in an attempt to capture the influence of the peaks include PI, P7, SUMXX, HRSXX, EFFMEAN, W95, W126 and the GPWCI (Table 2-1). Peak indices, including PI and P7, were initially investigated by NCLAN (Heck et al., 1984b). It was soon concluded that they were inappropriate because they represented the extreme high end of the distribution and, over different sites and different years, varied considerably more than observed plant response. The indices proposed by Oshima et al. (1976) and by Lefohn and Benedict (1982) are integrated exposures above an arbitrary threshold. Both give equal weight to the higher hourly concentrations and ignore those below an assumed threshold of 0.10 ppm. Other arbitrary threshold indices include those which integrate exposure above 0.08 and 0.12 ppm (Lee et al., 1987; Lefohn et al., 1988). The HRXX index is conceptually similar, and represents the number of hourly occurrences above an arbitrary threshold. A disadvantage of the use of integrated values above an assumed threshold is that the contribution of the lower concentrations to plant response is removed. In order to avoid this problem, Lefohn and Runeckles (1987) proposed a sigmoid function, which gives more weight to the higher concentrations but includes the lower concentrations in the integrated exposure. 28 The proposed sigmoid function is: W; = 1 / [1 + M • exp (-A • Q)] where: M, A = arbitrary constants W i = weighting factor for concentration i Q = concentration i The utility and fit of this function were evaluated by Lefohn et al. (1988) using NCLAN data. In this work, greater emphasis was given to concentrations above 0.08 ppm (W126) and 0.10 ppm (W95). Other exposure indices were fitted to the NCLAN response data, including the M7, P7, HR08, HR10, HR12, SUM08, SUM10 and SUM12. None of these exposure indices consistently provided a best fit using the linear or Weibull exposure-response models. However, the W126, W95, SUM08 and HR08 indices were found to fit the data as well as or better than the seasonal mean M7 index. Nosal (1983) proposed a more complex exposure parameter for use in a "mixed multivariate polynomial-fourier regression" response model. Response was related to three sulphur dioxide (S02) exposure parameters, including: (i) total cumulative exposure over the growing season, (ii) peak concentration of individual episodes, and (iii) total number of episodes above a certain assumed threshold over the growing season. Response data were fitted to exposure data and a good fit was achieved. Larsen and Heck (1984) proposed the "effective mean" index (EFFMEAN) as a means of assigning unequal weighting to 0 3 concentrations. The EFFMEAN model is based on the 29 following mathematical relationship which stresses the importance of the peak concentrations by the use of an exponent: me = [ (ZCV*) n]-v where: me = effective mean Q = individual 1-hour average 0 3 concentrations n = number of 1-hour averages in the exposure period v = an exposure time-concentration parameter Plant injury studies in which both concentration and time are set at different values are needed to compute a value for the v parameter in the above equation. Using an estimated v value of -0.376, the EFFMEAN model predicted that two separate locations with the same daytime arithmetic mean 0 3 concentration may have quite different effective mean 0 3 concentrations and crop reductions (Larsen and Heck, 1984). Although this exposure index would appear to be a significant improvement over the seasonal arithmetic mean, its utility is, o f course, limited by the length of exposure period problem. Larsen et al. (1988) recently argued that the M7 was a good exposure index because models using the M7 and the EFFMEAN predicted similar yield losses. The major flaw with this argument is that there was no assessment of whether either index was predicting accurate yield losses, only that the two indices are closely correlated with respect to specific sets of air quality data. Since most sites do not experience many peaks above 0.10 ppm (Lefohn and 30 Benedict, 1985), crop losses predicted by models using both mean exposure indices would be expected to be similar. None of the exposure indices discussed above addresses the sensitivity of the plant at the time of exposure. Plants vary widely in their susceptibility to 0 3 throughout their growth stages. At very early developmental stages, few functional stomata are present, so that most plants are relatively resistant at this stage (U.S. EPA, 1986). With increasing leaf age, stomata become rapidly functional and intercellular spaces form; these are important for internal pollutant distribution within the leaves. Gas exchange per unit area reaches a maximum just prior to full leaf expansion (Guderian et al., 1985). It is at this time that plants are generally most susceptible to injury. Plants may be affected by different 0 3 episodes at several growth stages throughout the growing season, and it is known that sequential exposures of plants to pollutants alters growth response and sensitivity (Runeckles and Rosen, 1977; Runeckles and Palmer, 1987; McCool et al., 1988; Kohut et al., 1988). If it is necessary to characterize the temporal distribution of exposures within a growing season to define plant response adequately, the current exposure indices used by researchers are inadequate (U.S. EPA, 1986). One index which attempts to overcome this shortcoming is the "generalized phenolog-ically weighted cumulative impact" (GPWCI), an integrated exposure index which utilizes a sigmoid concentration weighting function and gamma function weighting to simulate phenology (Lee et al., 1988; Hogsett et al., 1988). Using NCLAN and CERL (Corvallis Environmental Research Laboratory) data, the best fit was achieved when exposures occurring 10 to 40 days before harvest were given maximum weight. The researchers observed that the fit of the same exposure-response function was "clearly better" when the GPWCI was used, compared to the fit 31 when the M7 was used. Additionally, it was shown that over several experiments, there was a clear trend showing increasing yield loss with increasing GPWCI exposure; this was not evident with the M7. Several of the experiments exhibited similar M7 values but considerably different exposure durations; this caused the lack of a clear trend when the M7 was used. Additional issues that are not addressed by any of the indices discussed, including the GPWCI, include: (i) the effective dose that enters the plant at the time of exposure, (ii) the length of exposure within each episodic event and (iii) the time between exposures (the respite or recovery time). If these factors are as important or more important than ambient air pollutant exposure, then a given pollutant exposure will elicit varying biological responses at different times for the same crop (Krupa and Teng, 1982; Lefohn and Runeckles, 1987). Recent work by Kohut et al. (1988) demonstrated that respite periods between exposures are important in reducing the sensitivity of plants to subsequent 0 3 exposures, while McCool et al. (1988) found that plants became tolerant or desensitized to second fumigations only if the exposures were tightly spaced. Such inconsistent experimental results will make it difficult to establish a universally superior exposure index. Indeed, given the complexity of ambient 03-crop response relationships, which are inherently stochastic in nature, Krupa and Nosal (1989) question the validity of the concept of regional scale Cyinduced loss assessment. A recently reported study (Krupa and Nosal, 1989) may provide an explanation as to why previous efforts to identify a single, universally applicable exposure term have been unsuccessful. In an experiment with alfalfa that involved comparisons of the relationship between various exposure indices (mean, median, percentiles, peaks, frequency of exposure and cumulative integral of exposure) and yield losses during specific growth periods, it was found that the exposure index explaining the majority of the variation in crop growth changed with 32 degree of 0 3 flux density. When 0 3 uptake was slow, the median hourly 0 3 concentration was the best exposure index. When 0 3 uptake was rapid, the cumulative integral of 0 3 exposure was best. The characteristics of flux density appeared to determine the type of exposure parameter that can best describe crop response at any given time. These researchers concluded that, since crop phenology is also important, appropriate exposure indices must be identified for each growth stage and utilized in the final equation(s) explaining the crop yield responses. A major consideration in the selection of an 0 3 exposure index relates to the intended use of the index. If it is intended to be used in exposure-response models to assess the economic impacts of pollution, it should incorporate all contributing factors and account for the variation in exposure-response (Tingey et al., 1988). On the other hand, an exposure index that is applicable to the development of ambient air quality standards should be easy to measure and apply. The type of evidence that is needed for air quality standards relates some expression that defines air quality either to: (i) the absence of detectable effects on vegetation, or (ii) some quantifiable adverse response, selected on the basis of tradeoffs related to environmental protection (Lefohn and Runeckles, 1987). The aim of air quality management strategies is generally to control exposure over a long period of time to some average level of pollution, as well as to control exposure over a short period of time to high concentrations of pollutants (Simpson, 1988). Existing Canadian 0 3 objectives (Environment Canada, 1981) are as follows:3 Maximum Desirable: 1-h average 100 jug/m3 (0.05 ppm) 24-h average 30 ju,g/m3 (0.015 ppm) The relationship between absolute (^ tg/m3) and mole fraction units (ppm) is temperature and pressure dependent. The values quoted are for 25°C and 101.3 kPa. 33 Maximum Acceptable: 1-h average 160 )ug/m3 (0.08 ppm) 24-h average 50 jug/m3 (0.025 ppm) annual average 30 Atg/m3 (0.015 ppm) Maximum Tolerable: 1-h average 300 /ig/m3 (0.015 ppm) It has recently been recommended by a federal advisory committee that the 24-hour values for the desirable and acceptable objectives be rescinded, and that the annual value for acceptable objectives be rescinded and replaced with a "seasonal mean" of 0.032 ppm (three-month mean of the daily seven-hour means for the period 0900 - 1559 h) (personal communication, Pauline Erickson, Environment Canada, Environmental Protection Services, December 1989). The consideration of seasonal mean standards for Canada is surprising, considering the problems associated with a seasonal mean index, as described previously. In the United States, the primary (health effects) and secondary (welfare effects) 0 3 standard is 0.12 ppm (one-hour average), not to be exceeded on more than one day per year (Federal Register 44 FR 8202, 1979). The U.S. air quality standard is also under review (U.S. EPA, 1986), and a seasonal mean value was being considered, based on adoption of such an exposure index in the NCLAN work (Heck et al., 1988). However, it has recently been concluded by the EPA that "the use of cumulative indices to describe exposures of 0 3 for predicting crop effects appears to be a more rational approach than using long term averages" (Lefohn et al., 1989). There is clearly a conflict in objectives between exposure indices formulated for standard setting and those formulated for crop loss assessment. A suitable air quality standard for 34 protection of vegetation may have the form "no more than X days per year in which the one-hour average 0 3 concentration of Y ppm shall be exceeded" (Lefohn and Runeckles, 1987). An appropriate exposure term for a crop loss assessment model, on the other hand, might incorporate a more complicated exposure index that more accurately represents the influence of the exposure dynamics that are involved. The attractiveness of a simple summary exposure index is undeniable (Runeckles, 1987), although it clearly represents a compromise in the features of the "best" exposure index (Tingey et al., 1988). According to Lefohn et al. (1989), "no simple standard can be expected to accommodate the sources of variability [that affect] the magnitude of plant response. They will always be present and contribute to the uncertainty of any simple standard that is designed to protect vegetation." It is becoming obvious that the more sophisticated models and/or indices being investigated at this time may be too complicated for the purpose of standard setting. This was pointed out by Runeckles (1988), who suggests that "unless proved to the contrary, simple explanations or relationships are preferable to complex ones." Without a clearer understanding of the biological factors influencing pollutant uptake and time of increased plant sensitivity, no simple response models based on biological processes can be developed (Lee et al., 1987). Pure research-oriented process or mechanistic models that attempt to portray physiological processes within the plant may ultimately provide the direction that is needed. 35 2.3.2.3 Choice of Experimental Design A variety of exposure-response experimental designs have been employed, ranging from highly controlled laboratory experiments to outdoor field experiments involving minimal control of variables. Methods used include: (i) laboratory chambers and greenhouses, (ii) field chamber exposure systems, (iii) open-air field exposure systems, (iv) natural 0 3 gradients, (v) protective chemicals, and (vi) differential cultivar susceptibility. Three basic requirements needed to reach a conclusion regarding the effects of air pollution on plants are: (i) representativeness of the range of typical conditions, (ii) assurance of cause-effect associations, (iii) quantification of exposure-response relationships, and (iv) knowledge of interspecific and intercultivar differences in plant response. All three requirements cannot be met with any single exposure method, and controversy presently exists regarding the preferred method. The results from the more controlled experimental designs are difficult to extrapolate to the real world. On the other hand, the results from experiments with less control may be insufficient to establish cause-effect relationships and to quantify exposure-response. (a) Laboratory Chambers and Greenhouses The stability of highly controlled environmental conditions exhibited by laboratory exposure and greenhouse systems allows precise measurement of plant responses to various pollution levels. These methods are important for increasing our understanding of the effects of pollutants on the biological processes basic to plant growth (U.S. EPA, 1986). However, their use provides results that are difficult to extrapolate to the real world (Heck, 1982). 36 (b) Field Chamber Exposure Systems Open-top chambers provide a less controlled and more realistic growing environment for plants than laboratory chambers or greenhouses, and are presently the most extensively used experimental system (U.S. EPA, 1987). The original open-top field chamber design is shown in Figure 2-3. The use of field chambers allows standard agricultural practices to be carried out during field preparation, seeding and early crop growth, before the chambers are set in place. Chambers can be selectively placed in the field to maximize soil uniformity. Filtered air or ambient air is blown into the chamber through a perforated duct (plenum) forming the lower half of the wall of the 3 m diameter chamber. Air moves up through and laterally into the plant canopy and out the chamber top where its flow minimizes the incursion of ambient air. A frustum, or baffle around the top of modern chambers (not shown in Figure 2-3) further restricts ambient air incursion. The open top allows some ambient rainfall to enter, minimizes light restriction, helps prevent temperature increases, and permits some entrance of natural pests (U.S. EPA 1987). The major disadvantage of field chambers is that they result in unavoidable modifications of growing conditions, with the result that plants grown within such chambers generally exhibit different susceptibilities to pollutant injury and reductions in growth and yield than those grown naturally in the field. These growth and yield differences (termed "chamber effects") are due to various factors (Jacobson, 1982). Airflow is different between field- and chamber-grown plants. Windspeeds fluctuate significantly in the field and cause considerable variation in 0 3 flux. In the chamber, air flows from the lower portion out through the open top; this air flow pattern is different from that in the open field. This might alter the influence of 0 3 on plants and was the primary motivation for the development of "downdraft chambers" (Runeckles et al., 1978), which allow the deposition of 0 3 as one would expect under ambient conditions. In addition, 37 SOURCE: U.S. EPA, 1987 FIGURE 2-3 ORIGINAL OPEN-TOP FIELD CHAMBER DESIGN 38 physiological chamber effects can result from artificial air flows (wind speed is constant), humidity, light, C0 2 , soil drying rate and variations in incidence of biopathogens. A further limitation associated with open-top chamber studies is the relatively small plot size. According to Heagle (1989), "the large variation in results for some experiments was probably due to experimental variability associated with small sample size, rather than to differences in 0 3 concentrations or interactions between 0 3 and climatic factors." An additional potential limitation mentioned by Heagle (1989), and largely unaddressed to date, is the possibility that charcoal filtration removes "something" from the air that plants require for normal growth. If this is the case, then impacts based on comparison with filtered controls may be overestimated. Although there is evidence that the open-top chamber approach adopted by NCLAN may yield valid relative data, the magnitude of any interactive effects between pollutant treatments and chamber effects cannot be determined directly (Heck et al., 1983; Heck et al., 1984a). Recognition of the potential importance of chamber effects resulted in the development of the field exposure methods described below. (c) Open-Air Field Exposure Systems There are two types of open-air field exposure systems: (i) plume exposure systems, and (ii) air exclusion systems (U.S. EPA, 1987). An example of the plume exposure system is the Zonal Air Pollution System (ZAPS) (Lee et al., 1975; Runeckles et al., 1981; Wright, 1988; Runeckles and Wright, 1988; Runeckles et al., 1989), which consists of manifolds of perforated pipes that release the pollutant above the plant canopy. The design consists of a series of plots, each of which receives a different 39 long-term 0 3 exposure (Figure 2-4). Plants are exposed to pollutants under real world growing conditions. The primary difficulty with this method is that small changes in wind turbulence can cause major changes in 0 3 concentration, requiring a high intensity of monitoring. Modern ZAPS systems include intensive monitoring within plots and feedback control of 0 3 when wind speed drops (Runeckles et al., 1990). Air exclusion systems generally use inflatable perforated PVC ducts between plant rows to blow filtered air up through the plants during ambient pollution episodes, thereby preventing the incursion of ambient air into the plant canopy (Jones et al., 1977; U.S. EPA, 1987). Pollutants can be added as well as excluded; exclusion is not possible with plume exposure systems. Disadvantages of the method are similar to the plume exposure systems, and include loss of control over any dispensed pollutant and the need for intensive and extensive monitoring. Additionally, the air flow pattern is not as representative of ambient conditions as with plume exposure systems. (d) Natural 0 3 Gradients Natural 0 3 gradients, such as those occurring in the Los Angeles Basin of California, have been used to correlate exposure patterns with the yield of tomato (Oshima et al., 1975) and alfalfa (Oshima et al., 1976). More recently, Ashmore et al. (1988) used natural pollution gradients around London, England to quantify the impacts of air pollution on various crops. The advantage of such studies is that air pollution effects on crops can be observed under true field conditions, unmodified by enclosure within a chamber. Disadvantages of the method include the need for strong 0 3 gradients and the fact that differences in environmental, edaphic and cultural variables between sites can modify response to 03. However, where climatological conditions are reasonably similar throughout an area, 0 3 40 FIGURE 2-4 LAYOUT OF ZONAL AIR POLLUTION SYSTEM AT UNIVERSITY OF BRITISH COLUMBIA (a) GENERAL LAYOUT OF PLOTS (b) INDIVIDUAL PLOT 41 exposure and climatological conditions can be monitored at each site, allowing the development of exposure-response models for the plant species tested. (e) Protective Chemicals (Antioxidants) Antioxidants are compounds that prevent the reactions of organic compounds with molecular oxygen or 03. They can be used as chemical protectants of vegetation and, when it has been shown that they elicit no growth and yield effects themselves, their application may permit the estimation of yield losses. Ethylene diurea (EDU) is such an antioxidant. It is an experimental chemical that was developed specifically as a chemical protectant for 0 3 (Carnahan et al., 1978), although its commercial development has not been pursued. The yield in plots treated with EDU is thought to represent the yield that would occur in the absence of 0 3 , and several studies have demonstrated higher yields in EDU-treated plots in regions with high ambient 0 3 levels (U.S. EPA, 1986). A disadvantage with the use of protective chemicals is that only a single pollutant treatment is possible at each location; this prevents the development of exposure-response functions. Additionally, the chemical protectant itself may alter plant growth and the amount of protection provided for different plant species, by different methods and rates of application, is not known (U.S. EPA, 1986; Heagle, 1989). Hence, its use is currently limited to attempts to validate the estimates of yield loss predicted by exposure-response functions obtained by techniques such as the use of open-top chambers. Indeed, such EPA studies have led to the suggestion that the predictions for soybean yield reductions based upon NCLAN functions are substantially overestimated (Smith et al., 1987). 42 (f) Differential Cultivar Susceptibility Differences in yield of two cultivars, that vary in sensitivity to 0 3 and are grown at the same site under identical conditions, may be attributed to the effects of 03. Where cultivar differences in pollutant sensitivity are known, the method seems to have good potential for monitoring responses directly attributable to ambient air pollution (MEQB, 1984). However, as with antioxidant studies, exposure-response functions cannot be developed because of the single treatment at each site. 2.3.2.4 Use of Indirect Empirical Data It has been shown that considerable genetic variability in sensitivity to 0 3 exists between and among plant species. Relative plant resistance is based upon the expression of genetic traits which may vary during development, as well as by previous and existing environmental conditions (National Research Council, 1977; U.S. EPA, 1986; Heck et al., 1988). As a result, empirical results obtained under specific biological and physical conditions are not directly applicable to other situations. When direct empirical exposure-response information is lacking, indirect information (for example, other cultivars, species or growing conditions) may be substituted in its place. An unknown amount of uncertainty is associated with the use of such indirect information. Expertise is needed to judge the degree of similarity between the two situations, and hence, to determine how to adjust the observed results to apply to the situation of interest. 43 2.3.3 Analytical Uncertainty This source of uncertainty is related to the inability of predictive models, which are simplified versions of the real world, to faithfully represent reality (Turner, 1985). Sources of analytical uncertainty related to exposure-response models include: (i) the use of a small number of variables to represent a large number of complex phenomena, and referred to as approximation uncertainty (CMU, 1984) or aggregation error (Suter II et al., 1988), (ii) the choice of a biologically realistic functional form for the response model, and (iii) the choice of an appropriate background 0 3 level for predicting yield losses. 2.3.3.1 Approximation Uncertainty (Aweeation Error) The effects of 0 3 on crops cannot be reliably predicted unless all other factors that affect plant response, such as climatic and edaphic factors and biotic interactions, are held constant or can be explained. Light, temperature, relative humidity, soil moisture and soil fertility interact to provide the conditions for plant growth and to govern plant growth. Short-term changes in one or more of these environmental conditions may modify the susceptibility of the plant to 0 3 (U.S. EPA, 1986). An exposure-response model which accurately represents reality would be of the form: Plant Response = f (03 exposure + all other factors affecting plant response to 0 3 exposure) 44 Such a predictive model, which captures the entire system relationship, is in practice impossible to achieve. A major consideration is the extent to which the effect of the several important non-pollutant factors that modify plant response to 0 3 should be incorporated into the model. Incorporation of most or all of these factors into the response model would be an admirable undertaking, but one which in reality is probably not worth the cost and effort. Thus, for practical purposes, the number of explanatory variables in any model is restricted to those which appear to be most important in determining plant response. For example, NCLAN explicitly incorporated only 0 3 and effects of water stress, which potentially reduces plant response to 0 3 (Heck et al., 1984a). In experiments since then, water stress interactions have been found in some studies but not in others (Miller et al., 1989; Heagle, 1989). At this time, it would appear that only severe water stress will reduce the impact of 03. It is not always possible to define the most important variables. Scientific experiments may demonstrate that particular environmental variables can have a large effect on growth and performance, and these are therefore retained in the exposure-response model. Other variables may individually influence response less dramatically, so that their individual effects may not be statistically significant in experiments in which their influence is tested, and thus they are excluded. Unfortunately, the excluded variables may collectively have a large effect on response, and hence reduce the validity of the model by their omission. With the exception of the influence of water stress, little attention has been paid to the influence of environmental variables in the development of recent empirical models that predict crop yield as a function of some measure of 0 3 (U.S. EPA, 1986; Krupa and Kickert, 1987; Lefohn and Runeckles, 1987; Lee et al., 1987). 45 A model that incorporates the influence of developmental stage on 0 3 sensitivity (the three-dimensional response surface envisaged by Krupa and Teng, 1982) has not yet been validated, although two models have recently been proposed that capture phenological influences (Hogsett et al, 1988; Krupa and Nosal, 1989). 2.3.3.2 Choice of Model Functional Form The types of information generally required in the development of exposure-response models are threefold: (i) an appropriate index to describe pollutant exposure, (ii) a measure of crop response, and (iii) a mathematical function that relates exposure to plant response. Development of response functions is essentially an exercise in optimization, since independent choices must be made for each of the above (Runeckles, 1988). Unfortunately, it is not always obvious when poor choices have been made in one or more of the parameters. For example, it is possible to compensate for the selection of a biologically unrealistic 0 3 exposure index with the selection of a flexible exposure-response function (Lee et al., 1987). Prior to about 1987, in the empirical development of the most appropriate exposure-response models, greater emphasis was placed on selecting the functional form than in selecting an appropriate exposure index (Lee et al., 1987). Since that time, the emphasis has shifted to the exposure index (Lee et al., 1987; Lefohn and Runeckles, 1987; Tingey et al., 1988), possibly because of the growing need to provide alternative and better indices for use in setting air quality standards. The U.S. EPA (1986) warns that care must be taken to ensure that there is no systematic deviation of the exposure-response model from the observed data, and that results 46 cannot be extrapolated beyond the range of 0 3 concentration used to construct the model. Additionally, it is important to recognize that a response model should not be used to predict crop losses under conditions that differ from those used to estimate the model (Runeckles and Brown, 1986). To possess any general applicability, a model must account for the variation in environmental (and genetic) conditions in space and time (Male, 1981). Additionally, since separate models should be developed for individual cultivars, a large number of models is needed to predict impacts over large agricultural regions. Numerical models of air pollutant exposure and vegetation response have recently been reviewed by U.S. EPA (1986), Krupa and Kickert (1987), and Lee et al. (1987). Various exposure-response models, including the linear, quadratic, plateau-linear, Weibull and Box-Tidwell, have recently been used to represent crop response as a function of 0 3 exposure. Linear models are desirably simple, but they cannot represent threshold levels below which no yield reduction occurs, and many exposure-response relationships are non-linear. The plateau-linear model can represent thresholds, although it cannot accommodate curvilinear data patterns. Higher degree polynomials, in particular the quadratic model, can incorporate the effect of low level pollutant stimulations and deleterious effects at higher doses (Runeckles and Brown, 1986). However, although they allow curvature and gradual changes in slope, quadratic models tend to "waste" degrees of freedom; they change direction within the range of the data, producing false peaks; and they are not capable of generating the observed range of curvilinear patterns (Lee et al., 1987). Nonlinear models that have received considerable attention include the Weibull (Rawlings and Cure, 1985) and Box-Tidwell models (Lee et al., 1987). Statistical theory is not 47 as well developed for nonlinear models and confidence intervals are usually not fitted (U.S. EPA, 1986). Additionally, the nonlinear models have three parameters which, like the quadratic, use an extra degree of freedom. Most NCLAN experiments used four to seven distinct 0 3 treatments, severely restricting the class of possible exposure-response models to those possessing two or three parameters (Lee et al., 1987). A standard response function that can represent 0 3 exposure and plant response across species, sites and years should possess the following characteristics: (i) it should be flexible and span the observed range of biological responses, (ii) it should be biologically reasonable, and (iii) its parameters should be readily interpretable (Lee et al., 1987). The Weibull model has the form (Rawlings and Cure, 1985): Y = a • exp [ -(x/s)c ] + e where: Y = estimated crop yield a = maximum yield with no 0 3 x = 0 3 exposure index s = 0 3 exposure at which maximum yield (a) is reduced by 63% c = dimensionless shape parameter The growing recognition of the inadequacy of seasonal average exposure indices, and the growing popularity of the cumulative exposure indices, led to the suggestion for the use of simple exponential decay models, reciprocal models and gamma function models (Runeckles and Wright, 1989). The gamma function model, described in Section 2.3.3.3, can accommodate 48 initial yield increases followed by subsequent decreases at higher exposures. The exponential model is a two-parameter model, equivalent to the Weibull function with parameter c = 1. The reciprocal model is: Y = a/(b«x< + 1) where: Y = estimated crop yield a = hypothetical maximum yield with no 0 3 b,c = shape parameters x = exposure index A simple two-parameter reciprocal model eliminates the shape parameter c. The three-parameter model can accommodate an initial plateau, while the two-parameter model cannot. There does not appear to be a general consensus at this time as to the most appropriate exposure-response model functional form. Although the nonlinear models can be used to estimate the proportional yield loss necessary to combine data across sites, cultivars and years, an unknown degree of uncertainty is associated with the use of such aggregate models (as confidence intervals are not reported). Additional parameter uncertainty is introduced when such aggregate response models are used to make predictions under conditions that differ from those associated with collection of the original data. The most appropriate response model relating 0 3 exposure to vegetation loss would be based on actual biological processes. However, without a complete understanding of the biological factors involved, it is difficult to develop a theory-based response model (Lee et al., 49 1987; Krupa and Kickert, 1987). It is not possible at this time to define a single, standard response model that is appropriate for "real-world" conditions. 2.3.3.3 Natural Background Level of Ozone For any region, in order to determine the magnitude of effects due to air pollution, it is necessary to know what the natural (unpolluted) background level is. This is not a constant value, as background levels of air pollution, including 0 3, differ over time and space, and between urban and rural areas (Pratt et al., 1983). NCLAN (Heck et al., 1984a) used 0.025 ppm as a reference for background 0 3 in agricultural areas, although Evans et al. (1983) and Lefohn (1984) reported a high frequency of hourly observations above this value for a network of remote national forest sites in the U.S. Similar findings have been reported for rural locations in Alberta (Peake and Maclean, 1983; Angle and Sandhu, 1986). On the other hand, Altshuller (1987) reported that the natural background of 0 3 during the warmer summer months, in the U.S. and in Western Europe, is in the range of 0.010 to 0.020 ppm. Assumptions related to the natural (background) level of 0 3 are critical since apparent stimulatory effects have been observed at low 0 3 concentrations, in relation to plants grown in filtered air (Bennett et al., 1974; Runeckles and Wright, 1989). Because low levels of 0 3 are normal in unpolluted air, it seems likely that widespread selection for this background concentration has occurred. Crop yields in many NCLAN studies appeared to be highest under low 0 3 levels, rather than in charcoal-filtered air containing sub-ambient 0 3 levels. However, Rawlings et al. (1988), 50 after investigating the NCLAN data sets in regard to this phenomenon, concluded that there was no statistical support for the view that low 0 3 levels may result in stimulatory effects at low 0 3 concentrations. Runeckles and Wright (1989) argued that the NCLAN experiments, designed to determine negative yield responses, virtually precluded the likelihood of any stimulatory effects being demonstrated under low 0 3 conditions. Because of insufficient data in this range, Lee et al. (1987) reported that potential stimulatory effects are ignored in the model-building process; ie., exposure-response models are monotonic with maximum yield at zero concentration. A gamma function model, recently introduced by Runeckles and Wright (1989), can respond both positively and negatively to changes in the independent variable. The gamma function model is: Y = a»(x + l)c • exp-(b"'° where: Y = estimated crop yield a = hypothetical yield with no 0 3 x = 0 3 exposure index b,c = dimensionless shape parameters The gamma function model can respond positively at low values of the independent variable, although it does not require such positive data. 51 2.4 Epilogue Environmental policy making has as its goal the formulation of rational decisions that incorporate scientific knowledge as well as social and economic values. Decision makers should recognize that scientific uncertainty is associated with predictions about the effects of 0 3 on agricultural crops. Although this uncertainty cannot be eliminated, it can be reduced through appropriate scientific research (Runeckles and Brown, 1986). However, because of the complexity of the issues involved, scientific judgment does and will continue to have a major role in the formulation of rational policies and regulations related to the protection of agro-ecosystems from 0 3. 52 CHAPTER THREE CROP LOSS ASSESSMENT USING A BIOMONITOR PLANT 3.1 Role of Biological Systems in Crop Loss Assessment The previous chapter provided a description of sources of uncertainty associated with the current popular methodology for crop loss assessment. A major problem associated with the use of exposure-response models is that, in order to truly account for biological response, the interactive effects of all relevant conditions present during the growth of the crop must be integrated. Such large multivariate data sets are, in practice "impossibly complex and financially unrealistic" (MEQB, 1984). As an alternative, biological effects may be directly monitored using 03-sensitive plants. Through the use of biological rather than physical measurement systems, the impact of 0 3 can be directly observed as it affects vegetation in the environment. Because plants are products of their environment, their biological response reflects the integrated effect of air pollution conditions as well as other climatic, edaphic and physical factors that interact during the growth of the plant. Plants react in a number of ways to phytotoxic substances such as air pollution. While the most obvious symptom is the presence of visible foliar injury, plants may also exhibit developmental changes in: (i) growth and maturation rates; (ii) flower, fruit and seed formation; (iii) reproductive processes; (iv) biomass accumulation; and, (v) economic yield (Teng, 1982). 53 The concept of using plants as biological indicators of environmental quality is not new, as they have been used for more than 75 years to detect the presence and amount of airborne pollutants (Weinstein and Laurence, 1989). Plants were used as indices of smoke pollution in England prior to 1920 (Jacobson and Feder, 1974). In the 1960's, sensitive tobacco plants (Nicotiana tabacum L. cv. Bel W-3) were introduced as monitor plants for oxidant pollution (Heggestad and Menser, 1962). Since then, Bel W-3 tobacco has been used as a biological monitor or indicator of 0 3 in several studies in Canada (MacDowall et al., 1964; Mukammal, 1965), the United States (Heck and Heagle, 1970; Craker et al., 1974; Jacobson and Feder, 1974; Feder et al., 1975), Holland (Posthumus, 1982), Israel (Naveh et al., 1978), the British Isles (Ashmore et al, 1978; Ashmore et al., 1980a; Ashmore et al., 1980b), Australia (Horsman, 1981) and India (Bambawale, 1986). In air pollution studies, a "bioindicator" is a plant which exhibits a specific symptomatology when exposed to phytotoxic concentrations of air pollutants (Jacobson and Feder, 1974). These symptoms may not always be quantifiable. In contrast, a "biological monitor" plant exhibits injury symptoms that can be quantitatively related to ambient pollution levels (Teng, 1982), and/or the intensity of effects related to the exposure (Tonneijck and Posthumus, 1987; Tingey, 1989). In the context of crop loss assessment, both indicator and monitor plants have potential roles to play. Regional crop losses are generally predicted using air quality data collected at only a few sites. Bioindicator plants established in agricultural areas that do not contain physical air quality monitors provide information about phytotoxic 0 3 levels (Ashmore et al., 1978; Posthumus, 1982; Teng, 1982). The type of information collected, although qualitative in the sense that it cannot be used to predict air pollution levels or crop yield losses, provides direct evidence that vegetation has or has not been affected. In this 54 respect, for the purpose of regional crop loss assessment, bioindicator response data obtained at several sites throughout a region may be quite valuable. The role of plants would be even more valuable in regional crop loss assessment if their response could be "calibrated" with a response measure that is directly relevant to regional crop loss assessment (Laurence and Greitner, 1980; Posthumus, 1982; Teng, 1982; Tonneijck and Posthumus, 1987). There would appear to be two methods whereby the response of sensitive plants might be calibrated with crop response. The first method involves attempting to relate visible foliar injury exhibited by an 03-sensitive plant over a growing season to yield losses incurred by specific crops. If such a quantitative relationship could be defined, the visible response of the biomonitor plant could be used to predict the level of yield losses that occurred. The second method by which plants may be used as direct monitors of yield loss was described in the previous chapter. It involves the use of two crop cultivars with differing sensitivities to 03. Given identical growing conditions, the difference in response between the two cultivars represents the effect of the 0 3 exposure. This concept was explored in a study of the effects of SOz on vegetation and crops in Minnesota (MEQB, 1984), and has been utilized in Holland for several years in conjunction with a physical air quality monitoring network (Posthumus, 1982). Surprisingly, in spite of its intuitive appeal, there would appear to be relatively little interest other than this to date, perhaps because of a general lack of information on relative cultivar sensitivity to 03. If this is the case, there may be increased interest in biological monitoring in the near future, since recent research within the NCLAN program (Heck et al., 1988) has increased knowledge of the relative sensitivity of several cultivars. 55 3.2 Objectives and Rationale for the Biomonitor Project In addressing the issue of assessing the impact of 0 3 on crops in the Fraser Valley, the potentially useful information that could be obtained through definition of biological injury gradients suggested the use of a network of biological monitoring plants. Such a network could provide spatial and temporal information about potentially injurious 0 3 levels in locations removed from instrumented monitoring sites. With calibration, the use of biomonitor plants also offered the possibility of predicting crop losses. 3.2.1 Phytotoxic O, Trends The first objective of the present biomonitor research was to attempt to obtain a correlation between ambient 0 3 levels and plant foliar injury. The necessary data could be obtained by locating biomonitor sites adjacent to one or more instrumented ambient air quality monitoring stations. This information was desired in order to confirm that biomonitors established within the Fraser Valley are indeed valid indices of 0 3 concentrations, in which case they can serve as "early warning systems" (Posthumus, 1982) of potentially harmful 0 3 levels. Such relationships have been established by others. In an early study, Heggestad and Menser (1962) reported that, on average, 40% of Bel W-3 tobacco leaves were "flecked" by exposure to 53 to 100 ppb 03. Shortly thereafter, MacDowall et al. (1964) and Mukammal (1965) reported a significant correlation between injury to commercial cultivars of tobacco and 0 3 levels, after the data were corrected for evapotranspiration. Heck et al. (1966) reported the existence of a non-linear relationship between tobacco injury and 0 3 concentration. Manning 56 and Feder (1980) reported that a linear relationship exists between leaf injury and cumulative hours of 0 3 in excess of 40 ppb, when a large number of observations were aggregated. Naveh et al. (1978) found a significant linear relationship between leaf injury and ambient 0 3 exposure in Israel. In the British Isles, a weak correlation (r = 0.40, p < 0.01) was found between leaf injury and number of hours when 0 3 exceeded 40 ppb (Ashmore et al., 1980a). A stronger relationship (r = 0.65, p < 0.05) was found near Melbourne, Australia (Horsman, 1981). One of the most extensive studies using tobacco as a biomonitor of 0 3 pollution involved several sites throughout the northeastern U.S. over the years 1968 to 1971 (Jacobson and Feder, 1974). Over this period, widespread injury to Bel W-3 tobacco from 0 3 was observed throughout the region (in both urban and rural areas). It was concluded that the then existing air quality standard for photochemical oxidants (0.08 ppm one-hour average) may not have been adequate to protect susceptible plant species from the injurious effects of oxidants. A correlation coefficient between foliar injury and 0 3 concentrations was not reported. One study that observed a particularly strong relationship (R2 = 0.79, p < 0.01) between plant injury response and 0 3 levels was conducted by Oshima (1974) in southern California. For this work, pinto bean was used rather than tobacco. The biomonitor system was designed "to satisfy the need for an inexpensive method of monitoring ambient air pollutants in areas devoid of instrument stations." Some important experimental methods were incorporated into Oshima's project which contributed to its overall success. Plants were grown in standardized soil and nutrient regimes, and regularly replaced so that there was a constant supply of 10 to 12-day old plants. It was believed that this would capture the most susceptible plant growth stage. A photo-reference system involved standardized assessment of injury based on a direct comparison of injured leaves with reference photos of several levels of injury. Each 57 biomonitor location included a water reservoir and tubing that distributed water to the plants by gravity. Additionally the strong 0 3 gradients in southern California probably contributed to the success of the project. Biomonitor studies using Bel W-3 tobacco and several other species have been conducted for several years by Posthumus (1982) in the Netherlands. The techniques used were similar to those reported by Oshima (1974), except that filter candles in the soil were used to provide a continuous water source, and no photo reference system was used. Although the predictive ability observed by Oshima (1974) was not achieved, it was concluded that biomonitor plants should be used in conjunction with instrumented ambient air quality monitors, so that information about 0 3 concentrations as well as direct biological effects is obtained. 3.2.2 Calibration of Biomonitor Response with Crop Response The second objective of the biomonitor research was to calibrate the foliar response of Bel W-3 tobacco with yield reductions of two locally important agricultural crops, processing peas and potatoes. The existence of such relationships was thought to be possible on the basis that chronic symptoms such as yield reduction are known to result from a series of irritations, or "insults", to the plant from short-term 0 3 peaks throughout the growing season. Many of these short-term irritations could result in visible foliar injury to the biomonitor plant, while "invisible" damage might be inflicted on certain crop species. It was hypothesized that, over the growing season, the cumulative effect of these irritations might result in a level of crop loss that could be related to the cumulative foliar injury to Bel W-3 tobacco. 58 The relationship between foliar injury to a sensitive biomonitor plant and yield losses of agricultural crops has received little attention by researchers. This might be due to the general belief that plant injury may be either classified as being due to acute (short-term, high concentration) or chronic (long-term, low concentration) exposure to 0 3, each of which results in different responses. It is known that decreases in yield are not a direct result of leaf function impairment, because there is uneven competition among the several sinks that receive photosynthate and because compensatory responses to 0 3 can produce rapid recovery from injury (Jacobson, 1982). As a result, foliar symptom production has been shown to be not closely correlated with yield losses of several crops. Good correlations between foliar injury and yield losses may be expected when single exposures to 0 3 are applied at a stage when the harvested product is growing rapidly, but recurrent 0 3 exposures over the growing season were believed unlikely to be closely related, except when the harvested product is in the foliage. Although the relationship between foliar injury to specific crops and final yields of the same crops has been shown to be not closely correlated, Tonneijck and Posthumus (1987) reported a quantitative relationship between 0 3 injury to tobacco Bel W-3 and visible injury to sensitive and tolerant bean cultivars. They suggested that these relationships allow prediction of bean "effect intensities" by assessing 0 3 injury to tobacco. A similar relationship was reported for tobacco Bel W-3 and three spinach cultivars. The work reported by Tonneijck and Posthumus (1987) involved relating visible injury of the biomonitor plant with visible injury of specific crops. It did not relate visible biomonitor injury (acute responses to short term 0 3 episodes) with crop yield losses (chronic response to 59 the entire 0 3 distribution over the growing season). An attempt was made in the present research to determine if such a relationship could be observed. One argument against the feasibility of relating foliar injury of a biomonitor plant with yield losses of other species relates to the "respite period" between exposures. Recent evidence (Kohut et al., 1988) indicates that plant may fully recover during these respite periods, to the extent that no crop losses are observed at final harvest. The explanation of this phenomenon is that transitory reductions in photosynthesis will not necessarily result in yield impacts, particularly if tissue is not injured. This finding suggests that 0 3 episodes which may produce injury to the biomonitor plant, and may temporarily affect the growth of other crop species, may not result in permanent damage to the latter, if it is allowed to recover from the 0 3 episode. In this case, it may not be possible to accurately relate foliar injury of the biomonitor to final yield loss of other crop species. Another potential problem in regard to the establishment of such a relationship is that chronic yield losses of agricultural crops may result under 0 3 concentrations that are too low to result in acute injury to the biomonitor plant. Data collected by Wright (1988) showed that 0 3 episodes above 25 ppb can result in crop losses. The 0 3 exposure threshold for injury to tobacco Bel W-3 is approximately 40 ppb for four hours, under favourable environmental conditions (Ashmore et al., 1978). Under, these circumstances, the existence of biomonitor injury would provide a strong indication that crop losses occurred, although a lack of biomonitor injury would not necessarily mean that there were no losses. 60 3.3 Methods and Materials Bel W-3 tobacco was selected for the project because it exhibits several important biomonitor attributes. It is sensitive to 0 3, it is genetically uniform and it will grow well in the region of interest (in this case, the Fraser Valley). Additionally, it grows throughout the season, it produces characteristic Cyinjury symptoms and it responds proportionally to 0 3 exposure (Manning and Feder, 1980). 3.3.1 Cultivation and Standardization Bel-W3 and Bel-B seed was supplied by Dr. H.E. Heggestad, Agricultural Research Service, U.S. Department of Agriculture, Beltsville, MD. It was sown on the surface of a standard sterilized potting soil in 10 cm pots and then placed in greenhouses. After about two weeks, the seedlings were transplanted into flats, with 50 seedlings per flat. Two weeks after this, seedlings were individually planted into small (10 cm) pots. The seedlings were transplanted one more time, approximately two weeks later, into larger (30 cm) field pots. A diagram of the biomonitor station setup is shown in Figure 3-1. The field pots contained a water-soaked ceramic filter candle used for automatic watering. A tygon tube filled with water was connected to the ceramic filter candle in each pot, with the lower end of the tube below the common water reservoir. As the soil dried out due to plant use and evaporation, fresh water was drawn into the soil from the reservoir via the filter candle. In an ideal biomonitor network, the only variable that should significantly vary between sites is the 0 3 pollution level. In the present study, growing conditions for the biomonitor plants were standardized with respect to cultivation, soil, nutrients, watering and age of the 61 PLASTIC LINER GROUND LEVEL CERAMIC FILTER CANDLE SOIL 1 h / CERAMIC FILTER CANDLE HOLE FOR OUTLET OF EXCESS WATER TYGON TUBING -1 r-WATER LEVEL 1.22 m TOP VIEW ^ — 0.6 m SOURCE: MODIFIED FROM POSTHUMUS. 1982 F IGURE 3-1 DIAGRAM OF A STANDARDIZED CULT IVAT ION SET FOR BIOMONITOR PLANTS plants. However, it was not possible to control local climatic variables, including temperature, precipitation and sunshine. Within the Fraser Valley, these local climatic conditions are known to vary somewhat. In particular, during the summer, average daily temperatures increase as one moves east from Vancouver into the Fraser Valley. Climatic variables that could not be controlled should ideally be measured so that their influence can be explained. While it was not possible to do this in the present study, as the cost of instrumentation was prohibitive, the overall influence of the variation in climatic factors was probably not a major source of error, based on the work of Oshima et al. (1976). Plants were hardened off in an outdoor sheltered position for one to two days before being transported to the various exposure sites. Each biomonitor station consisted of four replicate plants placed in the top of a square wooden box lined with plastic, which served as the water reservoir. Each box was buried in the ground so that the plants were approximately at ground level. Biomonitor surveys were conducted in the summer months of three years, 1985 through 1987. During each survey, an attempt was made to replace each batch of plants every four weeks. This was not possible in 1985 due to unfamiliarity with cultivation procedures, but fresh plants were set out approximately every four weeks in 1986 and 1987. In 1985 and 1986, each biomonitor station consisted of four pots, each containing one plant. These plants were removed and replaced by four young plants approximately every four weeks. In 1987, a modification was devised which was intended to allow the comparison of injury exhibited by older and younger plants, as suggested by Ashmore et al. (1978), since susceptibility to 0 3, and hence injury, is known to vary with growth stage. For this purpose, two plants were put in each pot, and two pots only were initially established at each biomonitor station (a total of four plants per station). After approximately four weeks, two additional pots 63 were added, each containing two young plants (a total of eight plants per station). Subsequent to this, injury exhibited by both the older and younger plants was recorded. The exposure periods were July 11 to August 19 in 1985, May 25 to September 3 in 1986, and June 9 to September 1 in 1987. 3.3.2 Fraser Valley Biomonitor Sites Biomonitor sites identified through discussions with representatives from the B.C. Ministry of Agriculture and Food were selected in different agricultural areas within the Fraser Valley. In 1985, seven bioindicator stations were established, at Annacis Island, Anmore, Pitt Meadows, Abbotsford, Chilliwack, Fort Langley and Cloverdale (Figure 3-2). Although Annacis Island and Anmore are not important agricultural areas, they were selected because they were known to experience relatively high 0 3 levels. It was deemed necessary to locate some of the biomonitor stations adjacent to ambient air quality monitors so that an attempt could be made to correlate the injury response of the plants with known 0 3 levels. Ambient air quality monitors, owned and operated by local government agencies, were located at Abbotsford, Chilliwack, Annacis Island and Anmore in 1985. Most biomonitor stations were located within 100 m of air quality monitoring stations, except at Anmore and Chilliwack where the separation distances were 500 m to 750 m. These separation distances were not considered to be a significant source of error, since 0 3 is a regional pollutant resulting from an urban area source (Kosusko and Nolen, 1989). Ozone levels and trends were assumed to be relatively constant over the separation distances associated with this study. 64 * OZONE BIOMONITORING STATION (SURVEY YEARS) * OZONE AIR QUALITY MONITOR NOTE: BIOMONITOR STATIONS WERE NAMED FOR CONVENIENCE BASED ON THE CLOSEST COMMUNITY BUT ARE NOT NECESSARILY LOCATED WITHIN THAT COMMUNITY F I G U R E 3 -2 LOCAT ION OF BIOMONITOR STAT IONS AND AMBIENT A IR QUAL ITY MONITORS In 1986 and 1987, twelve biomonitor sites were established (See Figure 3-2). In addition to the seven 1985 locations, plants were established at Mission, Matsqui, Surrey, Boundary Bay and Richmond. In 1986, ambient air quality monitors existed at seven of these twelve locations. In 1987, the Annacis Island biomonitor site was relocated to south Burnaby, adjacent to where the ambient air quality monitor had been relocated. The locations of biomonitor sites and air quality monitoring sites are summarized in Table 3-1. 3.3.3 Calibration of Biomonitor Response With Crop Response In 1986 and 1987, field experiments on crop response to 0 3 were conducted using the ZAPS facility (see Figure 2-4) on the U.B.C. campus. The operation of the ZAPS facility and the crop response have been described elsewhere (Wright, 1988; Runeckles et al, 1990). In 1986, biomonitor stations were established immediately adjacent to experimental plots (four treatment plots and one control plot) that contained two important Fraser Valley crops, potato (Solanum tuberosum L. cv. Russet Burbank) and pea (Pisum sativum L. cv. Puget). The biomonitor stations were situated on the west side of the plots, so that changes in wind direction would affect the plants equally with respect to 0 3 released within the plots. The experiment was repeated in 1987, with 14 biomonitor plots located within each of 14 ZAPS plots. In the 1987 experiment, an additional cultivar of tobacco (Bel-B), which is relatively resistant to 0 3, was planted in each of the biomonitor plots. The objective was to aid in the confirmation of injury by 0 3 to Bel W-3, since if the symptoms were the result of other causes (e.g., sunscald or Altemaria infection), they would probably show up on both cultivars. 66 TABLE 3-1 Location of Biomonitor Sites by Year 19851 19862 19873 Site No. Location** GVRD* Monitor Site No. GVRD* Location** Monitor Site No. GVRD* Location * * Monitor 1 Annacis Island T13 1 Annacis Island T13 1 Burnaby — 2 Anmore T7 2 Anmore T7 2 Anmore T7 3 Pitt Meadows (North) — 3 Pitt Meadows T16 3 Pitt Meadows T16 4 Abbotsford T i l 4 Mission — 4 Mission — 5 Chilliwack T12 5 Chilliwack T12 5 Chilliwack T12 6 Fort Langley — 6 Abbotsford T i l 6 Abbotsford T i l 7 Cloverdale — 7 8 Matsqui Fort Langley : 7 8 Matsqui Fort Langley : 9 Surrey T15 9 Surrey T15 10 Cloverdale — 10 Cloverdale — 11 Boundary Bay — 11 Boundary Bay — 12 Richmond T17 12 Richmond T17 Survey Period: 1 July 11 - August 19 2 May 25 - September 3 3 June 9 - September 1 * GVRD = Greater Vancouver Regional District. ** Biomonitor stations were named for convenience based on the closest community, but are not necessarily within that community. 67 The Cygenerating system consisted of a Grace Model GS 4060 0 3 generator, which added 0 3 to the treatment plots in proportion to the ambient level. In 1986 the 0 3 generator was supplied with compressed oxygen, while in 1987 it was supplied with compressed air. 3.3.4 Symptomatology and Injury Assessment As with most broadleaf plants, tobacco leaves that are just approaching full size are most susceptible to 0 3 injury (MacDowall et al., 1964; Guderian et al., 1985). Within hours of exposure to phytotoxic levels of 0 3, the injured area appears water-soaked due to damage to palisade cell membranes, which release their cell sap. Subsequently, when dry, the injured areas of the leaf exhibit characteristic "flecks" or "stipples" (Manning and Feder, 1980; Guderian et al., 1985; U.S. EPA, 1986). Flecks are small lesions formed when groups of palisade cells die and the associated epidermal cells collapse. Flecks are yellow or tan in color, and if the injury is extensive, the entire leaf surface may appear bronzed. Individual flecks may coalesce to form bifacial lesions that appear on both leaf surfaces (U.S. EPA, 1986). Stipples are small groups of red, purple or black pigmented palisade cells. Stippling is viewed through the uninjured epidermal layer of the upper leaf surface (U.S. EPA, 1986). The colors result from the Cyinduced biosynthesis of specific anthocyanins and polyphenols in the palisade cells (Guderian, 1985). Acute injury to Bel W-3 tobacco from 0 3 exposure results in "classic" symptoms easily distinguishable from those caused by other stresses, including drought, insects and pathogens (Manning and Feder, 1980). 68 At approximately one week intervals, each biomonitor site was visited and a visual assessment made of injury. The value of the data obtained depended on the ability of the assessor to distinguish 0 3 injury from other types of leaf injury, and to assess accurately the percentage of each leafs area covered by 0 3 injury (Ashmore et al., 1980). Identification of 0 3 injury was confirmed by Drs. V.C. Runeckles and RJ. Copeman, Department of Plant Science, U.B.C. Two other types of injury identified were sunscald and infection by the fungus Altemaria. Leaf injury was visually estimated as recommended by Ashmore et al. (1980a), Manning and Feder (1980), and Posthumus (1982). Originally, it was intended to classify injury according to a seven-point scale based on the following injury levels: 0%, 0-5%, 5-10%, 10-25%, 25-50%, 50-75%, and 75-100%. However, after the first few visits, it was obvious that the level of injury was going to be relatively low, in the range of 0 to 5%. Thus, the above scale was abandoned and actual percentages were visually estimated. A transparent grid was used to assist in the determination of level of new injury. The percentage of new injury on each leaf of each plant was recorded, as well as the number of live leaves. A mean leaf injury value, referred to as the leaf injury index (LII) (Ashmore et al., 1978), was computed for each station for each visit. In addition, a cumulative leaf injury index (Cumulative LII) was obtained for each site by summing weekly leaf injury indices. 3.3.5 <93 Exposure Indices The most common exposure index used in previous biomonitor studies is the number of hours in which 0 3 exceeded 40 ppb during the period of exposure. For the present biomonitor 69 project, several measures of 0 3 were computed, including the M12, M7, SUM40 and SUM50, HR40 and HR50, and P7. Definitions for these exposure indices are provided in Table 2-1, although weekly values were computed rather than seasonal values. Additional exposure indices computed included the overall mean (M24); total 0 3 dose based on 24-hour, 12-hour (0700 to 1859 h) and 7-hour (0900 to 1559 h) daily periods (TOTDOSE 24, 12, 7); and the peak 1-hour 0 3 level observed in 24-hour and 12-hour daily periods (P24, P12). Fraser Valley 0 3 data were obtained from the Greater Vancouver Regional District, and FORTRAN programs permitted the extraction of the relevant 0 3 data by location and time period, and computation of the weekly 0 3 indices defined above. Ozone levels in the experimental field plots at U.B.C. were measured by a Dasibi Model 1003-AH 0 3 monitor, operated on a time-share basis among the various field plots. All data were recorded by plot and time period using a Campbell Scientific Model 21-X data logger. The time-share system resulted in the collection of two-minute average data for each plot, two or three times per hour (Runeckles et al., 1989). The various exposure indices were then computed based on the one-hour means for each experimental plot. 3.3.6 Data Analysis Pearson correlations were computed using the SAS statistical package to examine the relationship between weekly leaf injury indices and the various 0 3 exposure indices. Data from individual sites, as well as aggregate data for all sites, were analyzed for each year and for all three years together. 70 For the purpose of attempting to calibrate biomonitor response to crop yield, weekly leaf injury indices over the experimental exposure period were summed. These cumulative leaf injury indices were regressed against final yields of the potato and pea crops. 3.4 Results and Discussion The leaf injury index is ratio scale data which is amenable to correlation analysis. However, it is recommended by Zar (1984) that, since percentages form a binomial rather than a normal distribution, percentile data should be transformed to their arcsines. This data transformation was not reported in the previous studies cited earlier. For the present study, both the original data and arcsine-transformed data were used in the statistical analysis. 3.4.1 Fraser Valley Biomonitor Survey 3.4.1.1 Biomonitor Injury Levels Various measures of leaf injury index by site and time period for 1985 are shown in Figure 3-3. Highest leaf injury index levels on average were observed for the week of July 18 to 24 (LII = 0.53), followed by August 2 to 12 and August 20 to 28 (0.44 for both periods). The highest injury levels, on average, were observed at Anmore (LII = 0.68), followed by Annacis Island (0.37) and Fort Langley (0.37). Overall average LII for all sites was 0.28 in 1985. 71 LEAF INJURY INDEX SITE WEEKL Y ANNACIS ISLAND ANMORE PITT MEADOWS ABBOTSFORD CHILLIWACK 1.05 1-35 0.00 0.00 0.00 0.00 0.00 0.73 0.00 0.07 Q.02 0.07 0.43 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.19 0.00 2.00 4.74 1.93 1.32 0.00 FORT LANGLEY CLOVERDALE AVERAGE ALL SITES NEW PLANTS ESTABLISHED 1.14 0.00 0-13 °- 1 B 0.09 0.13 0 0 0 „ 0 0 0.53 0.34 0.44 n P 1 0.44 0.00 — , „ , „ 0.00 0.07 0.2B 2.59 0.51 1.96 JULY JULY JULY 25- AUG AUG AUG AUG 20-11-17 1B-24 AUG 1 2-12 13-19 20-28 SEPT 6 FIGURE 3-3 BIOMONITOR LEAF INJURY INDEX 1985 Overall leaf injury was much lower in 1986, with an overall average LII of 0.03 for all sites (Figure 3-4). Highest injury levels on average occurred during the week of May 25 (LII = 0.07) and the week of August 8 (0.08). The sites experiencing the highest injury levels, on average, included Anmore (LII = 0.15), Abbotsford (0.06), Pitt Meadows (0.05), Chilliwack (0.05) and Surrey (0.04). In 1987, leaf injury levels were again quite low, with an overall average LII of 0.04 for all sites (Figure 3-5). The greatest amount of injury was observed during the week of June 16 (LII = 0.23) and the week of June 23 (0.11). In 1987, the greatest amount of injury was observed at Chilliwack (0.08), Abbotsford (0.06), Cloverdale (0.06) and Mission (0.05). One method used by researchers to portray spatial phytotoxic 0 3 trends over large geographic regions has been to plot cumulative or average LII values observed at various sites and to construct isolines of injury (Ashmore et al., 1978; Posthumus, 1982). For this purpose with the present data, cumulative LII values were obtained for each site by summing weekly LII values. For plotting, three arbitrary levels of injury were selected: (i) negligible injury, represented by cumulative LII values of 0.0 to 0.10%; (ii) low injury, with cumulative LII values of 0.11 to 0.70%; and (iii) medium injury, represented by cumulative LII values greater than 0.70%. The results for 1985 are presented in Figure 3-6. Only seven biomonitor sites were included in the 1985 survey, and the approximate boundaries of coverage are outlined. Within these boundaries, medium injury was observed everywhere except in the vicinity of Cloverdale, where injury was low, and Chilliwack, where injury was negligible. The lack of injury observed at the Chilliwack site was surprising, since ambient 0 3 levels measured at Chilliwack were frequently sufficient to induce injury symptoms. It is felt that the biomonitor site was too sheltered in 1985, and it was moved to a less sheltered location in 1986. 73 LEAF INJURY INDEX SITE WEEKL Y /SEASONAL, 'AVERAGE, CUMUL. ANNACIS I SLAND 0.22 0 1 3 0.00 °- 0 4 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.39 ANMORE 0.25 0.23 0.21 °- 3 3 0.50 • • • 0.20 o.oo IHI^HItaHH °' o u °-00 HHI °'00 P ^ ^ ^ H H ^ H H H 0.15 1.B8 P I T T MEADOWS 0.31 0.15 0.13 0.00 o.oo °-00 °'00 °-00 ^ ^ ^ ^ ^ ^ m _ ° ^ 0 5 0.05 0.64 MISS ION 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 CH ILL IWACK 0.31 0.15 0.13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 o.oo ^ ^^^^^m 0.00 0.05 0.59 ABBOTSFORD 0.22 • • ^ ^ M 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 . 0 0 ^ ^ ^ ^ ^ ^ 0.06 0.72 MATSQUI 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 FORT L A N G L E Y °- 0 6 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.005 0.06 SURREY 0.20 0.18 o.oo o.OO ° ' 0 0 ° - 0 0 ° - 0 0 • • • ° - 0 0 0.04 0.48 C L O V E R D A L E 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 BOUNDARY BAY 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 RICHMOND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 AVERAGE A L L S I T E S 0.07 0.04 0.05 0.03 0 00 0.00 0 00 °- 0 3 0 00 °- 0 1 °' 0 B °- 0 5 °- 0 4 0.03 0.40 NEW P L A N T S E S T A B L I S H E D y y y MAY JUNE JUNE JUNE JUNE JUNE 27-JULY JULY JULY 25-31 1-6 7-12 13-20 21-26 JULY 4 5-10 11-25 26-31 FIGURE 3-4 BIOMONITOR LEAF INJURY INDEX 1986 AUG AUG AUG AUG 22-1-7 B-14 15-21 SEPT 3 LEAF INJURY INDEX SITE HEEKL Y SEASONAL, 'AVERAGE CUMUL. BURNABY ANMORE PITT MEADOWS MISSION CHILLIWACK ABBOTSFORD MATSQUI FORT LANGLEY SURREY CLOVERDALE BOUNDARY BAY o.oo o.on 0^08 n nn n nr n no n on o.oo o.oo o.oo o.oo o.oo 0.32 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.12 0.12 0.00 0.00 0.00 0.00 0.10 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.02 o.OO 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.007 0.03 0.06 0.00 O.OB 0.36 0.73 0.00 RICHMOND 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 o.oo 0.00 0.00 o.oo 0.00 0.00 AVERAGE ALL SITES 0.00 0.01 0.01 0.01 0.01 0.01 0.00 0.00 0.00 NEW PLANTS ESTABLISHED 0.04 JUNE JUNE JUNE JUNE 30- JULY JULY JULY JULY 28- AUG AUG AUG AUG 25-9-15 16-22 23-29 JULY 6 7-13 14-20 21-27 AUG 4 5-11 12-18 19-24 SEPT 1 F IGURE 3 -5 BIOMONITOR LEAF INJURY INDEX 19B7 0.44 FIGURE 3-6 RELAT IVE BIOMONITOR INJURY (CUMULATIVE L I I ) IN 1985 In 1986, medium injury occurred in the vicinity of Anmore and Abbotsford, low injury was observed near Pitt Meadows, Surrey and Chilliwack, and injury was negligible everywhere else within the boundaries of the study region (Figure 3-7). A small area between Abbotsford and Matsqui was assumed to exhibit low injury. In 1987, medium injury occurred near Cloverdale, Abbotsford and Chilliwack (Figure 3-8). The majority of the study area exhibited low injury in 1987, extending from Anmore and Surrey in the west to Mission in the east. Again, injury was negligible in the extreme western portion of the study area. On July 20, new injury was detected at Mission on the new batch of plants but not on the old batch of plants. This was the only time that injury was observed when both new and old plants were exposed at the same time. The corresponding LII was based on the average overall injury to the old and new plants. Because differential injury to old and new plants was only observed on this occasion, no attempt was made to incorporate the influence of leaf age into the LII values, as was done by Ashmore et al. (1978). The information presented in Figures 3-6 through 3-8 cannot be used to quantitatively predict crop loss levels, but it does indicate where crop damage was most likely to have occurred, and where it was probably most significant. Similarly, the cumulative injury maps outline areas where risk to crops was quite low. This information is particularly useful in parts of the study area where ambient air quality monitors presently do not exist (for example, at Boundary Bay, Cloverdale, Fort Langley, Matsqui and Mission). 77 * OZONE 8I0M0NIT0RING STATION (SURVEY YEARS) * OZONE AIR QUALITY MONITOR NEGLIGIBLE INJURY (LII = 0.0-0.10 %) LOW INJURY (LII = 0.11-0.70 %) MEDIUM INJURY (LII > 0.70 %) FIGURE 3-7 RELATIVE BIOMONITOR INJURY (CUMULATIVE L I I ) IN 1986 FIGURE 3-8 RELATIVE BIOMONITOR INJURY (CUMULATIVE LII) IN 1987 3.4.1.2 Ozone Exposure I Biomonitor Correlations Several attempts were made to relate weekly LII values with various 0 3 indices. When all paired data were included, the majority of which contained some positive value for 0 3 but no injury, no significant correlations were observed. Similarly, when the data containing zero injury were dropped, no significant correlations were observed for individual years. However, when data containing zero injury were dropped and all three years were aggregated, significant correlations (p < 0.05) were observed for three of the exposure indices, including SUM40, SUM50 and M24 (Table 3-2). The correlation coefficients were similar for the transformed and untransformed data; r-values ranged from .300 to .345. These values correspond to the typically low correlations between 0 3 and injury levels observed by others and cited previously. The low r-values reflect the influence of the various environmental and internal physiological factors that affect pollutant uptake and response. However, the results suggest that Bel W-3 tobacco can serve as an "early warning system" of the presence of potentially phytotoxic 0 3 levels in the Fraser Valley. Several additional aggregate data exposure indices were statistically significant at the p < .10 level. These included P24, P12, P7, HR50 and M7 for the untransformed data, and HR50 for the transformed data. The exposure indices that emphasize or give weight to the peak 0 3 concentrations appear to have outperformed the mean 0 3 indices. 80 TABLE 3-2 Correlations Between Fraser Valley Weekly Leaf Injury Indices (LII) and Various 0 3 Indices Data Set Ozone Index Pearson Correlation Coefficient (p-value)12 Untransformed Arcsine Data Transformed Data Fraser Valley 1985 (n=10) Fraser Valley 1986 (n=19) TOTDOSE24 .432 (.21) .272 (.45) TOTDOSE12 .292 (.41) .147 (.69) TOTDOSE7 .281 (.43) .144 (.69) P24 .375 (.29) .291 (.41) P12 .375 (.29) .291 (.41) P7 .392 (.26) .310 (.38) SUM40 .378 (.28) .356 (.31) SUM50 .331 (.35) .345 (.33) HR40 .500 (.14) .386 (.27) HR50 .341 (.34) .310 (.38) M24 .630 (.07) .592 (.09) M12 .394 (.29) .396 (.29) M7 .405 (.28) .415 (.27) TOTDOSE24 .163 (.50) .191 (.43) TOTDOSE12 .115 (.64) .126 (.61) TOTDOSE7 .089 (.72) .093 (.70) P24 .262 (.28) .242 (.32) P12 .255 (.29) .235 (.33) P7 .242 (.32) .219 (.37) SUM40 .159 (.51) .154 (.53) SUM50 .253 (.30) .245 (.31) HR40 -.074 (.76) -.071 (.77) HR50 .100 (.69) .110 (.66) M24 -.130 (.59) -.126 (.61) M12 -.196 (.42) -.216 (.37) M7 -.178 (.47) -.206 (.40) ... continued 81 TABLE 3-2 (Cont'd.) Correlations Between Fraser Valley Weekly Leaf Injury Indices (LII) and Various 0 3 Indices Data Set Ozone Index Pearson Correlation Coefficient (p-value)1,2 Untransformed Arcsine Data Transformed Fraser Valley 1987 TOTDOSE24 .153 (.53) .302 (.21) TOTDOSE12 .162 (.51) .239 (.32) (n=19) TOTDOSE7 .118 (.63) .188 (.44) P24 .036 (.88) .089 (.72) P12 .036 (.88) .089 (.72) P7 .059 (.81) .116 (.64) SUM40 .153 (.53) .207 (.39) SUM50 .119 (.63) .176 (.47) HR40 .224 (.36) .282 (.24) HR50 .164 (.50) .226 (.35) M24 .139 (.57) .298 (.22) M12 .181 (.46) .281 (.24) M7 .129 (.60) .220 (.37) Fraser Valley TOTDOSE24 .228 (.12) .151 (.31) Aggregate Data TOTDOSE12 .134 (.37) .068 (.65) (1985-1987) TOTDOSE7 .136 (.36) .075 (.61) P24 .254 (.08) .214 (.14) (n=48) P12 .254 (.08) .214 (.14) P7 .263 (.07) .224 (.13) SUM40 .313 (.03) * .304 (.04) * SUM50 .338 (.02) * .345 (.02) * HR40 .227 (.12) .177 (.23) HR50 .266 (.07) .258 (.08) M24 .322 (.03) * .300 (.04) * M12 .226 (.13) .215 (.15) M7 .241 (.10) .229 (.12) Correlation coefficients significant at < 0.05 are flagged with an asterisk (*). Observations where leaf injury was not observed were excluded. 82 3.4.2 Crop Calibration Experiment 3.4.2.1 Relationship Between and Biomonitor Injury During the crop calibration experiment conducted at U.B.C. in 1986, considerably more biomonitor injury was observed (Table 3-3) than during the Fraser Valley biomonitor survey. Significant correlations (p < .05) were observed for all 0 3 indices except for the P12 (Table 3-4). R-values ranged from 0.382 to 0.638 for the untransformed data; the transformed data performed similarly. The observed relationship reflects the expected situation whereby higher injury was observed with increasing 0 3 treatment levels. The M12 and M7 0 3 indices slightly outperformed the indices that emphasize the peaks in the 0 3 distribution, indicating that the mean indices correlated well with increasing treatment levels. 3.4.2.2 Relationship Between Crop Yield and Biomonitor Injury The biomonitor exposure period at the U.B.C. field plots in 1986 was June 2 to August 18, which corresponded to the experimental crop exposure period. As shown in Table 3-3, no injury was observed in the control plot exposed to ambient 0 3 levels, while injury was observed in all four of the treatment plots. As indicated previously, the relative amount of biomonitor injury observed in the treatment plots was correlated to the 0 3 treatment level. The cumulative LII was lowest at Site 1 (1.87), with greater injury at Site 2 (3.41), followed by Site 3 (5.61) and Site 4 (14.41). Average injury levels observed at Sites 1 and 2 corresponded approximately to average injury levels observed in the Fraser Valley in 1985, while injury levels 83 TABLE 3-3 Leaf Injury Index by Biomonitoring Period: UBC 1986 Site Period 1 Period 2 Period 3 Period 4 Period 5 Period 6 Period 7 Period 8 Period 9 Period 10 Period 11 Seasonal Cumulative June 2-9 June 10-17 June 18-23 June 24-30 July 1-5 July 6-11 July 12-18 July 19-24 July 25-31 Aug. \S * Aug. 9-18 Average Injury co Control 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1 0.00 0.50 0.50 0.11 0.29 0.06 0.09 0.37 0.00 0.00 0.00 0.17 1.87 2 0.00 0.21 0.00 0.86 0.93 0.10 0.31 0.59 0.37 0.00 0.00 0.31 3.41 3 0.00 1.08 0.00 1.14 0.67 0.40 0.98 0.77 0.27 0.33 0.00 0.51 5.61 4 0.00 0.71 0.17 0.86 2.60 3.08 1.82 2.79 0.78 0.00 1.61 1.31 14.41 Average 0.00 0.50 0.13 0.59 0.90 0.73 0.64 0.90 0.28 0.07 0.32 0.46 5.06 New batch of plants TABLE 3-4 Relationship Between UBC Weekly Leaf Injury Indices (LII) and Various 0 3 Indices Data Set 0 3 Exposure Index Pearson Correlation Coefficient (p-value) 1,2 Untransformed Data Arcsine Transformed Data UBC Experiment 1986 (n=31) TOTDOSE24 .382 (.03)* .394 (.03)* TOTDOSE12 .432 (.02)* .444 (.01)* TOTDOSE7 .504 (.004)* .508 (.004)* P24 .476 (.007)* .417 (.02)* P12 .289 (.11) .252 (.17) P7 .460 (.01)* .403 (.02)* SUM40 .492 (.005)* .461 (.009)* SUM50 .440 (.01)* .415 (.02)* HR40 .570 (.001)* .530 (.002)* HR50 .490 (.005)* .460 (.009)* M24 .561 (.001)* .528 (.002)* M12 .598 (.001)* .563 (.001)* M7 .638 (.001)* .600 (.001)* Correlation coefficients significant at < 0.05 are flagged with an asterisk (*). Observations where leaf injury was not observed were excluded. 85 observed at Sites 3 and 4 were on average higher than those observed in the Fraser Valley (Table 3-3 and Figures 3-3, 3-4 and 3-5). Crop yields for 1986 in the control and four treatment plots were obtained from Wright (1988). These data and the cumulative LII data are presented in Table 3-5. Preliminary observation of these data shows that, where biomonitor injury occurred, apparent crop yield losses also occurred. The scatter plots of crop yield as a function of seasonal cumulative leaf injury indicated a negative relationship, as expected. The concave shape of the scatter plots was the result of even the lowest level of biomonitor injury being associated with substantial crop losses. Although it was possible to fit both linear (Figure 3-9(a)) and non-linear models (Figures 3-9(b) and (c)), none of the models should be considered reliable due to the limited data base. Of the three models, the reciprocal model consistently provided the best fit, as demonstrated by minimum residual sums of squares (Table 3-6). In 1987, no injury to Bel W-3 tobacco was observed throughout the entire experiment, which was run from early June through mid-August, again corresponding to the crop growth period. Similarly, analysis of potato and pea yield data in the 1987 field experiment indicated that significant yield reductions attributable to 0 3 did not occur. The lack of impact in 1987 was due to unusually low ambient levels of 0 3 that prevailed throughout the experiment. Increasing the output of the 0 3 generator to its maximum output was insufficient to result in injury. Although only limited biomonitor data were obtained, primarily as a result of atypically low 0 3 conditions observed during the project (discussed in Chapter Seven), the results are encouraging. It was demonstrated that Bel W-3 tobacco can be used as an inexpensive indicator of phytotoxic 0 3 trends in the Fraser Valley. It would also appear that it may be 86 TABLE 3:5 Comparison of Cumulative LII and Crop Yields in 1986 UBC Experiment Site Potato Weight (grams) Pea Weight (grams) Cumulative LII Control 1022 38 0.00 1 593 19 1.87 2 767 15 3.41 3 661 22 5.61 4 731 17 14.41 87 1 2 0 0 1 6 2 4 6 8 1 0 1 2 1 4 SEASONAL CUMULATIVE LII ( X ) ' 2 4 6 8 1 0 1 2 1 4 SEASONAL CUMULATIVE LII (X) FIGURE 3-9A LINEAR YIELD-INJURY MODELS SEASONAL CUMULATIVE LII (X) SEASONAL CUMULATIVE LII (X) FIGURE 3-9B EXPONENTIAL YIELD-INJURY MODELS 1 2 0 0 , y = 817.1 / (0.0178X + 1) 2 4 S 8 1 0 1 2 1 4 SEASONAL CUMULATIVE LII (X) 2 4 6 8 10 12 14 SEASONAL CUMULATIVE LII (X) 1 6 FIGURE 3-9C RECIPROCAL YIELD-INJURY MODELS 88 TABLE 3-6 Residual Sums of Squares (RSS) Associated with Biomonitor Models Model RSS (a) Pea Yield = f (Cumulative LII) Linear y = 26.5 - 0.844x 249 Exponential y = 28.3 e"00567* 227 Reciprocal y = 33.5/(0.166x+l) 177 (b) Potato Yield = f (Cumulative LII) Linear y = 800.1 - 8.953x 96970 Exponential y = 806.9 e"00137* 95862 Reciprocal y = 817.1/(0.0178x+1) 94334 89 possible to "fine-tune" the calibration between biomonitor injury response and crop losses in future experimental work. On the basis of the results obtained in this research, it may be concluded that: (i) The existence of injury to tobacco Bel W-3 is an indication that yields of some important agricultural crops may be reduced due to phytotoxic 0 3 exposure. (ii) The lack of injury to tobacco Bel W-3 does not necessarily mean that crop losses due to 0 3 exposure have not occurred. 90 CHAPTER FOUR DESIGN OF EXPERT SURVEY Environmental policy makers are frequently required to make regulatory decisions based on inadequate technical information and scientific knowledge of the risks and benefits that are involved. Environmental risk assessment strategies have recently been used to predict human and environmental risks from air pollution in several projects (e.g., Morgan et al., 1979; Feagans and Biller, 1981; Ruckelhaus, 1983; Morgan et al., 1984; Fraser et al., 1985; Morgan et al., 1985; North et al., 1985; Peterson and Violette, 1985; Turner, 1985; Amaral, 1988). However, use of expert scientific judgment as a basis for implementing environmental policy remains controversial (Russell and Gruber, 1987). One of the reasons for the controversy may be that there is a serious deficiency of information regarding the necessary and desirable attributes of expertise. The quality of expert judgments clearly depends on the competence of the experts that are selected. An increased understanding of expertise and provision of a systematic means for selecting experts should greatly increase the credibility of environmental risk assessment. 4.1 Attributes of Expertise According to Webster's Ninth New Collegiate Dictionary (Merriam-Webster Inc., 1988), an "expert" is a person having or displaying special skill or knowledge derived from training or experience, or a person with the special skill or knowledge representing mastery of a particular subject. "Expertise", according to Webster's, is "the skill of an expert", or "expert opinion or commentary". 91 The concept of the expert seems to be quite straightforward for most people, since "experts" are frequently called upon to provide "expertise" within their particular area of special knowledge. As indicated by Webster's, a person is frequently acknowledged as an expert once he or she has achieved a certain level of skill or knowledge in a particular area or discipline, on the basis of training or experience, or a combination of both. This definition of the expert obviously covers a very wide range of knowledge and/or skill. For certain work tasks, mastering some skill, and thus achieving expertise, involves less training or experience than other work tasks. For example, a person completing a secretarial course may be considered an expert on the typewriter, while to be considered an expert physician, much more preparation and training is necessary. Although the use of experts for decision making in society is increasing due to the increasing complexity of our environment, there is very little published information related to the necessary and desirable attributes of expertise. There are no guides or manuals available to help us select an expert on a particular topic. Experts used in legal proceedings ("expert witnesses") are usually selected on the basis of meeting certain minimum criteria, such as completion of a certain educational program, and on the basis of their reputation for being a "good expert", or a "good witness" (Mildred, 1977). The credentials of experts are seldom questioned, with the exception of provision of expert testimony in courts of law or at public inquiries. In this case, an expert's qualifications are seldom reviewed by his or her peers; rather, legal professionals frequently judge whether an individual has the credentials to qualify as an expert. Among other requirements, the basic 92 qualifications required of an expert witness in courts of law include the following (Mildred, 1977): (i) Relevant professional qualifications and experience in the field of the dispute. (ii) The ability to weigh facts and to draw logical conclusions from them. (iii) The ability to view a problem impartially. Empirical studies of people's abilities as probability assessors (for example, the ability of weather forecasters to accurately predict the weather) has primarily emphasized their "calibration" (Morgan et al., 1979). Calibration is a measure of the degree to which assessed probabilities match empirical frequency. For an underconfident judge (or expert), the assessed probabilities are nearer to 0.5 than they should be. More typically, judges are overconfident and their probabilities are assessed too near certainty (0 or 1). For example, weather forecasters have been shown to be well calibrated (Fischoff et al., 1977); this is attributed to the fact that their forecasts receive rapid and accurate feedback. Where feedback is not as rapid (for example, in medical prognostication), experts are not well calibrated (Brehmer, 1986). The concept of calibration appears, at present, to be restricted to psychological research studies of human judgment capability. Mildred (1977) states that an expert must constantly strive to keep himself informed of developments in all fields affecting his professional expertise, through the reading and hearing of learned papers, and through contact with professional colleagues. It is paradoxical but nevertheless true that a person with the longest experience may not be as well informed as a person with less experience who has taken greater pains to keep his knowledge up-to-date. 93 Any person who has the "special knowledge, skill, experience, training, or education" necessary to become an expert in a particular field may be qualified to testify as an expert (Bradley, 1983). Generally, the weight given to expert testimony is greater when a witness of accredited skill and experience forms an opinion from personal observation or examination of the subject at issue, and less when general conclusions are drawn on the basis of secondary or indirect study. The convincing or persuasive force of expert testimony may be weakened by a showing of strong personal interest or bias (Bradley, 1983). An expert's authority traditionally rests upon assumptions that scientific data are objective, that scientific conclusions are based upon rational and logical procedures, and that scientific advice or testimony is a product of the scientific method. However, to some degree, the tradition of an objective, neutral scientist is changing. While part of the change is internal, scientists are reexamining social responsibilities in a changing world; most of the change seems to come from a need by society for scientific assessments of the consequences of large scale technological innovations (Ravetz, 1978). According to Bradley (1983), the most important role of an expert witness in court is that of educator. In his words: Experts deal in a scarce commodity—knowledge. Sharing this knowledge so that it might be used to reach rational and justifiable solutions is an expert's unique role. To be an expert is to be fundamentally an educator. Education is more than collecting or presenting facts; experts educate . . . by interpreting facts. Experts explain the significance of information and help to mold and direct the understanding of the problems at issue . . . Expertise and professionalism are nearly equal in the eyes of the public. Professional demeanour is essential. The fundamental need seems clear—better scientific information for court use, presented in ways that encourage instead of discourage its adoption. 94 Although information related to the attributes and credentials of experts is sparse, there has been a lot of attention and research on human judgment processes and decision making. A survey conducted by Arkes and Hammond (1986) showed that articles related to judgment and decision making appear in more than 500 professional journals. These researchers summarized the relationship between judgment and expertise (Arkes and Hammond, 1986): Nowhere is the role of judgment more important than in the work of the expert, for the essential value of experts lies in their judgment . . . Experts exercise judgment; that is they seek out, select, organize information, and offer, on the basis of their expertise—a judgment, a diagnosis, a plan—that would be produced in no other way. For nearly two decades, psychologists have been studying the cognitive abilities of expert decision makers (e.g., Tversky and Kahneman, 1974; Kahneman et al., 1982; Arkes and Hammond, 1986). These studies have concluded that experts, because of cognitive limitations, are generally inaccurate, unreliable, biased and lack self-insight. Moreover, except for being more confident, experts appear to gain little with experience. Overall, previous psychological studies have painted a rather dismal picture of the decision making abilities of experts. One frequent explanation for this low level of performance is that experts reportedly rely on heuristics (mental rules of thumb) in making judgments. Heuristics are necessary because of the limited cognitive processing capacity of the human brain (Simon, 1986); these heuristics often lead to biases or judgmental errors. Similar biases have been reported for both novice and expert decision makers (Kahneman et al., 1982). Some studies have found that expert judgments are based on surprisingly little information. Court judges have been found to use only a portion of the available information when sentencing defendants (Shanteau, 1987); medical pathologists have been reported to be equally limited (Einhorn, 1986). 95 Research conducted by James Shanteau at the Department of Psychology, Kansas State University, has provided an alternate (and optimistic) view of experts (Shanteau, 1987). Experts were found to use various strategies to help them overcome the effects of cognitive limitations. These strategies are as follows (summarized from Shanteau, 1987): (i) Experts are willing to make adjustments in initial decisions. They take advantage of subsequent feedback in dynamic environments and avoid rigid adherence to prior decision policies. Rigidity and blind commitment to prior choices are characteristic of inexpert decision makers; experts have learned that being right is more important than being consistent. (ii) Top decision makers rely on others to assist in making decisions. Experts seldom work in isolation, but operate either with a group or have the opportunity to gain feedback, additional insights and perspectives from others. Isolation from associates usually leads to inferior decision making. (iii) Experts know how to learn from past decisions and to make appropriate changes in future decision strategies. Although they may not learn in the most efficient fashion, they are responsive to past successes and failures. In contrast, novices frequently appear more interested in rationalizing or defending past decisions than in learning from them. (iv) Experts in various domains often have developed various informal decision aids which allow them to avoid the biasing effects of heuristics. These aids may take the form of written records of prior decisions to eliminate hindsight biases, or they may be normative guides to calibration, as in soil category decisions. 96 (v) Although experts may make small errors, they tend to avoid large mistakes. This appears to reflect a dual strategy of first coming up with a rough "ball park" estimate and then conducting a more careful analysis. (vi) Experts frequently follow some kind of "divide-and-conquer" strategy. They break large problems into smaller parts, find solutions to the parts, and then use these partial solutions to estimate the larger solution. How this is done varies with the content area. Although such an approach is often prescribed as part of normative decision analysis, experts have often evolved this strategy without reference to formal analysis. Shanteau has also found that experts tend to share several psychological characteristics. These include, but are not limited to, the following (summarized from Shanteau, 1987): (i) Experts generally have highly developed perceptual/attentional abilities. They are able to extract information that non-experts either overlook or are unable to see. The difference is that experts are able to see patterns that novices cannot. (ii) Experts seem to have a sense of what is relevant and what is irrelevant when making decisions. (iii) Experts have an ability to simplify complex problems—an expert is "someone who can make sense out of chaos". (iv) Experts can effectively communicate their expertise to others. (v) Experts are able to handle adversity and stress better than non-experts, possibly due to the development of well-learned strategies to deal with them. 97 (vi) Experts know that it is important to be selective in picking decision problems to solve. In comparison, novices frequently adopt one of two extremes: they either want to be perceived as "decisive" and thus aggressively take on all decisions, or they are so reluctant to make a mistake that they avoid making any decisions. (vii) Almost all experts show strong outward confidence in their decision making ability. Although this might be viewed as arrogance, it generally comes across as a highly developed faith in one's own abilities. Experts believe in themselves and their capacity to make decisions. (One respected agricultural judge, when confronted with an inconsistent decision about which of two animals was best-of-show, commented "there must have been two grand champions".) (viii) Almost without exception, experts have an extensive and up-to-date content knowledge. They know a lot and pride themselves in staying up to date with the latest developments in their field. (One recently retired agronomist commented that he felt unqualified because he had not kept up with developments in the past few months.) 4.2 Theoretical Model of an Expert It may be hypothesized that experts selected to provide risk estimates for the purpose of improved policy decisions would exhibit certain desirable characteristics or attributes. For the purpose of the present research, expertise was assumed to be a function of three major clusters of attributes, or dimensions, as follows: 98 Substantive expertise refers to a high level of knowledge of a particular topic area; this being the effects of 0 3 on crops in the present research. Substantive expertise is obtained through learning, which in turn is a function of education and career experience. Normative expertise refers to the ability to make good intuitive judgments on the basis of information which has been processed and transformed by the human brain. According to Simon (1986), "in any field of expertise, possession of an elaborate discrimination net that permits recognition of any one of tens of thousands of different objects or situations is one of the basic tools of the expert and the principal source of his intuitions". Degree of normative expertise is largely a function of differences in various cognitive abilities, including general intelligence, memory, imagination, creativity, ability to ignore irrelevance, observational ability, ability to integrate diverse opinion and other attributes. External credibility relates to a person's ability to convince others of his or her expertise. This dimension encompasses various personal or professional qualities, skills or styles related to the credibility of experts as viewed by others. Honesty and communication ability are two examples of attributes that are important to an expert's credibility. How an expert is viewed in terms of motives and/or biases is also important in terms of external credibility. 99 In order to test the validity of this assumed expert model, a questionnaire was designed and prepared to address the following primary questions: (i) What is an expert? What are the necessary and desirable attributes of expertise related to assessment of crop losses from 0 3 pollution? Is there consensus between plant scien-tists and other "stakeholders" (persons with an interest in 0 3 effects on crops) regarding the desirable profile of expert attributes? (ii) Who is an expert? Who are the most highly qualified individuals to provide expert judgments of risks to crops from 0 3 pollution in the Fraser Valley of B.C.? Are the norms held by plant scientists regarding the desirable attributes of experts reflected in the ranking of experts by peer nomination? The underlying purpose of the survey was to identify and select experts for the purpose of assessing crop losses due to 0 3 in the Fraser Valley/ 4.3 Design and Testing of Questionnaire Design of the questionnaire involved an eclectic method of defining potentially important indicators of expertise within the three major theoretical dimensions. Indicator variables thought to contribute to substantive expertise were defined through personal research experience in Q 3 effects of crops, and through discussion with U.B.C. Plant Science professors Hereinafter, this project is referred to as the Fraser Valley Risk Assessment Project. 100 (Drs. V.C. Runeckles and W.G. Wellington) and graduate students. Defining the indicators of normative expertise involved an extensive examination of the literature related to judgment and decision making (e.g., Kahneman et al., 1982; Arkes and Hammond, 1986; von Winterfeldt and Edwards, 1986). A study of the relative importance of indicators of physician performance (Price et al., 1971) provided several indicators of external credibility. The questionnaire was pretested to identify inconsistencies and to determine approximate completion time. Graduate students in the U.B.C. departments of Plant Science, Botany and Geography, and in the School of Community and Regional Planning, were included in the pretest. Ninety questionnaires were distributed, of which 37 were returned completed. It appeared that respondents understood and were able to answer the questions without difficulty. Suggestions for additional indicators of expertise were minimal. To test the reliability (consistency) of the questionnaire, a test-retest approach was employed. The questionnaire was re-administered to the sample of individuals who completed and returned the questionnaire. Seventeen respondents completed the questionnaire a second time. Normally, it is sufficient to retest ten individuals (Politt and Hungler, 1983). Two tests were used to assess reliability. A paired t-test was used to compare the test and retest scores (a basic t-test is not appropriate because the two groups of respondents are not independent). The criterion is the difference between mean test and mean retest scores, and involves testing the null hypothesis that the mean of differences is zero. The computed t-values for 10 of the 97 variables tested exceeded the critical values (p < .05). In the second test, correlation coefficients were computed between test and retest scores. A low correlation coefficient is indicative of relatively large differences in scores by at least some individuals that were retested. The majority of the computed coefficients were 101 relatively high, thereby indicating consistency between test and retest scores. However, coefficients less than 0.4 were observed for 20 of the 97 variables, and less than 0.5 for 33 of the 97 variables. All variables that produced inconsistent results were examined. In most cases, wording of these variables was revised; in some cases, variables were deleted. Following the pretest, the questionnaire underwent a final edit and revision prior to distribution to the target sample. The final questionnaire is reproduced in full in Appendix A. A summary of the information requested in the questionnaire is presented below: (i) Substantive Expertise. Respondents were asked to comment on the relative importance of experts having a certain level of education, a particular academic background, their academic performance, length of career experience, types of career experience and frequency of contributions to literature. (ii) Normative Expertise. Respondents were asked to rate the relative importance of an expert possessing certain cognitive or mental skills. For this purpose, 22 items were included in the survey. Examples include general intelligence and/or intellect, mental quickness, memory, imagination and creativity, and ability to ignore irrelevance. (iii) External Credibility. Nineteen items representing potentially important personal qualities, as they relate to credibility of an expert, were included. Examples are forthrightness, willingness to admit ignorance, honesty, articulateness and communications ability. (iv) Nomination of Experts. Respondents were asked to nominate up to five top experts in 0 3 effects on agricultural crops. 102 (v) Respondent Information. Respondents were asked to provide information regarding their education, discipline, years of career experience, location, work description and employer. 4.4 Sampling Frame and Distribution of Questionnaire The population for which the survey was designed consisted of plant scientists or others knowledgeable about the effects of 0 3 on agricultural crops. The primary geographic unit of interest was North America, although scientists in Europe were also considered. The names and addresses of individuals comprising the sample were obtained from several sources. The primary sources included the past attendance list (names and addresses) of the "Air Pollution Workshop", an annual international meeting of researchers, scholars and professionals with an interest in the effects of air pollution on plants. The 1987 mailing list consisted of approximately 500 North Americans and 100 Europeans. This list was reviewed with Dr. V.C. Runeckles and names were deleted if they did not appear to conform to the target population (for example graduate students with minimal experience, or engineers who were unlikely to have specific training in pollution effects on plants). Additionally, names were added of scientists who did not appear on the list but who were known to be studying 0 3 effects on crops. Additional names and addresses for the sample were obtained from: (i) the "Alberta Acid Deposition Research Program", whose mandate included air pollution effects on crops, (ii) the Air Pollution Control Association, a local chapter of an international society, and 103 (iii) published lists of conservation groups within North America with an interest in air pollution effects (U.S. EPA, 1985; B.C. Environmental Network, 1988). The edited sampling frame consisted of a total of 449 names. The questionnaire was distributed to the entire sampling frame to minimize the risk of obtaining an insufficient number of completed questionnaires. Each questionnaire was numbered and coded to an individual on the master mailing list, so that individuals not returning the questionnaire could be sent a follow-up letter. Questionnaires were distributed by mail to each of the 449 individuals on the mailing list. A stamped return envelope was included for Canadian and American respondents. As completed questionnaires were received, they were coded and the numerical data were entered in a computer file. Approximately four weeks after the initial distribution of the questionnaire, a follow-up letter was sent to those individuals who had not returned it. This letter encouraged respondents to return the completed questionnaire, or if this was impossible or undesirable, to provide a reason for not returning it. Only one follow-up letter was distributed as the overall response rate was very good. The results of the survey are presented in the following chapter. 104 CHAPTER FIVE EXPERT SURVEY: RESULTS AND DISCUSSION This chapter describes the results of a questionnaire survey conducted to determine the desirable attributes of experts selected to provide risk estimates of crop losses due to 03. The relative importance of 76 indicators of expertise are described and the underlying dimensions of expertise are identified. Differences in perceptions of expertise between various groups with different backgrounds are identified and described. Models estimated to assist in the selection of top experts from a group of potential candidates are also described. 5.1 Description of Sample and Response Rate Table 5-1 summarizes the response rate by the five main groups comprising the sampling frame. Of 449 questionnaires mailed out, 16 were returned by the post office as undeliverable, 29 were returned by individuals who indicated they felt unqualified to complete the questionnaire, 18 were returned by persons who were too busy or not interested, and one completed questionnaire was mailed but never received. A total of 238 completed questionnaires were returned for an overall response rate of 53%. The groups within the sample providing the highest responses included North American (60%) and European (50%) Air Pollution Workshop attendees, and local (B.C.) Air Pollution Control Association Members (50%). The response from the Alberta Acid Deposition Research program scientists was lower (43%), probably because several of the scientists contacted were only indirectly involved with assessment of air pollution effects on crops (this was determined after the questionnaire had been distributed). The response from the conservation groups was lowest, with only 16% 105 TABLE 5-1 Questionnaire Response by Group Post Office Overall Original could not Too busy/ Lost by Returned Response Group Questionnaires deliver Unqualified Not interested Post Office Completed (%) Air Pollution Workshop attendees (North America) 326 11 19 12 1 195 60 Air Pollution Workshop attendees (European) 14 1 ~ 3 7 50 Alberta Acid Deposition Research Scientists 14 - ~ ~ -- 6 43 Local Air Pollution Control Association Members 44 - 8 3 - 22 50 Conservation Groups _5_L _4 _2 ~ ~ _8 16 Total Sample 449 16 29 18 1 238 53 responding. The reason for the low response rate is not known, but it is possible that persons within these organizations generally felt unqualified to respond to the questionnaire. It is difficult to estimate the exact number of completed questionnaires received without the influence of the follow-up letter. There was a period shortly after the letter was sent during which several questionnaires were received, but it seemed unlikely that these persons were responding to the reminder letter. The majority of the questionnaires were returned prior to the reminder letter being distributed; approximately 170 completed questionnaires had been received by this time. A summary description of the sample is presented in Table 5-2. Of the 238 respondents, the majority held Ph.D. degrees (159; 68%) while the remainder held Masters degrees (45; 19%) or Bachelor's degrees (20; 9%). Ten individuals (4%) did not specify their educational status. The majority of the sample received their education in Plant Science or Botany (112; 49%), while 27 individuals (12%) had degrees in "other biology"; these were primarily foresters, zoologists or general biologists. Forty-three individuals (19%) had an education in some "other science" discipline, while eight persons (4%) were trained in social science. Forty individuals (17%) were educated in some "other" discipline; it is believed these persons were primarily engineers. Eight persons did not provide information on their educational discipline. The sample, overall, was very experienced, as 169 individuals (73%) had 10 or more years of relevant career experience since their most recent graduation. Sixty-one individuals (26%) had more than 20 years experience. The number of inexperienced persons in the sample was quite low, with only 15 persons (6%) having less than six years experience, and 48 individuals (21%) with six to ten years work experience. Six individuals did not provide details of their career experience. 107 TABLE 5-2 Description of Sample Variable Frequency Percent Education (highest degree obtained) B.S. degree 20 8.5 M.S. degree 45 19.2 Ph.D. degree 159 67.9 Other 10 4.4 Missing 4 — 238 100.0 Educational discipline Plant Science or Botany 112 48.7 Other Biology 27 11.7 Other Science 43 18.7 Social Science 8 3.5 Other 40 17.4 Missing 8 — 238 100.0 Years of experience since most recent graduation Five years or less 15 6.5 Six to ten years 48 20.7 Eleven to twenty years 108 46.5 More than twenty years 61 26.3 Missing 6 238 100.0 Present location (Country) Canada 56 24.3 U.S.A. 164 71.4 Other (Europe) 10 4.3 Missing 8 — 238 100.0 Present work description1 Scientific research 180 75.6 Consulting 54 22.7 Policy/regulations 47 19.7 Enforcement 20 8.4 Administration 55 23.1 Teaching 45 18.9 Other 26 10.9 Present employer1 University 89 37.4 Government 98 41.2 Consulting 25 10.5 Industry 28 11.8 Other 30 12.6 Several persons within each of these categories fitted into more than one of the subcategories, with the result that the total sums to a figure exceeding the total sample Persons residing in the U.S.A. comprised the majority of the sample (164; 71%), although Canadians were fairly well represented (56; 24%). Ten Europeans (4%) were included in the sample and six persons did not provide information on their location. The majority of the respondents conducted scientific research (180; 76%), while other job descriptions included: consulting (54; 23%), policy or regulations development (47; 20%), enforcement (20; 8%), administration (55; 23%), teaching (26; 11%) and unspecified (26; 11%). Employers included: universities (89; 37%), government (98; 41%), consulting firms (25; 11%), industry (28; 12%) and unspecified other (30; 13%). Again, several respondents were employed by more than one of these categories of employers. 5.2 Descriptive Statistics on Attributes of Expertise The majority of the items within the questionnaire were responded to by circling one of the following: E VI MI SI NI where: E = essential (coded value was 5) VI = very important (4) MI = moderately important (3) SI = slightly important (2) NI = not important (1) 109 This scale should be referred to when the mean scores for each item in the questionnaire are being considered. For example, a mean score close to 5.0 indicates that the item is essential, and a mean score close to 4.0 indicates that the item is very important. 5.2.1 Descriptive Statistics: Substantive Expertise Descriptive statistics for indicators of substantive expertise are presented in Table 5-3. The mean scores, standard deviations and frequencies of individuals in each importance category were computed for each item in the questionnaire. The items were ordered in terms of decreasing importance. This is different from the order of items in the questionnaire. 5.2.1.1 Level of Education There is general consensus that an expert must have at least a B.Sc. degree (mean score = 4.71). However, 49% of the sample felt that the M.Sc. degree is essential. Only 20% felt that the Ph.D. degree is essential to become an expert. The post-doctorate was not considered important by many (mean score = 2.87), with only 12% of the sample considering it essential and an additional 20% rating it very important. 5.2.1.2 Academic Background There is general consensus that a background in Plant Science is very important to essential (mean score = 4.67), with 97% of the sample rating it within these categories. Botany would appear to be an acceptable alternative (mean score = 4.07), with 72% of the sample 110 TABLE 5-3 Descriptive Statistics: Indicators of Substantive Expertise Variable Mean Standard Frequency (%) by Score1  Score Deviation N 1 2 3 4 5 What level of education should an expert have who provides risk estimates of 0 3 impacts on crop yields? B.S. degree or equivalent 4.71 0.77 220 4(2) 4(2) 7(3) 21(10) 184(84) M.S. degree or equivalent 4.16 1.04 221 8(4) 7(3) 34(15) 64(29) 108(49) Ph.D. or equivalent 3.63 1.00 230 8(3) 17(7) 74(32) 84(37) 47(20) Post-doctorate or equivalent 2.87 1.28 230 44(19) 44(19) 68(30) 46(20) 28(12) What major disciplines or fields of study are an important background for an expert who provides risk estimates of 0 3 impacts on crop yields? Plant Science 4.67 0.54 235 0(0) 0(0) 8(3) 61(26) 166(71) Botany 4.07 0.88 232 0(0) 9(4) 54(23) 80(34) 89(38) Statistics 3.74 0.83 234 1(4) 11(5) 80(34) 98(42) 44(19) Atmospheric Science 3.46 0.84 236 0(0) 27(11) 100(42) 83(35) 26(11) Chemistry 3.01 0.94 232 12(5) 51(22) 105(45) 51(22) 13(6) Mathematics 2.71 0.91 230 18(8) 76(33) 98(43) 30(13) 8(3) Physics 2.23 0.86 228 39(17) 118(52) 55(24) 12(5) 4(2) ... Continued 1 Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all TABLE 5-3 (Cont'd) Descriptive Statistics: Indicators of Substantive Expertise Variable Mean Standard Frequency (%) by Score1  Score Deviation N 1 2 3 4 5 How important is academic performance (as measured by grades) while in university or college? Pass (50% - 64%, C) 4.67 0.87 224 7(3) 2(1) 11(5) 18(8) 186(83) Second Class (65% - 79%, B) 4.03 0.99 227 8(4) 6(3) 41(18) 88(39) 84(37) First Class (> 80%, A) 3.12 1.10 228 28(12) 20(9) 96(42) 64(28) 20(9) How important is length of career experience (including graduate work)? 5 years or less 4.78 0.56 223 0(0) 2(1) 9(4) 26(12) 186(83) 6 to 10 years 3.83 0.91 234 5(2) 11(5) 57(24) 108(46) 53(23) 11 to 20 years 2.79 1.07 230 36(16) 46(20) 84(37) 58(25) 6(3) 21 years or more 2.10 0.99 228 81(36) 61(27) 69(30) 16(7) 1(<1) What types of career experience should an expert have who provides risk estimates of 0 3 impacts on crop production? Studied the majority of published articles dealing with 0 3 effects on crops 4.35 0.79 237 1(<1) 4(2) 28(12) 83(35) 121(51) Participated in experiments dealing with 0 3 effects on crops 4.28 0.80 237 1(<1) 7(3) 25(11) 95(40) 109(46) ... Continued 1 Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all TABLE 5-3 (Cont'd) Descriptive Statistics: Indicators of Substantive Expertise Variable Mean Score Standard Deviation N Frequency (%) by Score1 Authored scientific papers dealing with Oj effects on crops 3.91 Conducted statistical analysis related to defining plant dose-response models 3.56 Supervised experiments dealing with 0 3 effects on crops 3.49 Regularly attended conferences and workshops dealing with 0 3 effects 3.49 Served as a peer reviewer in evaluating research regarding 0 3 effects on crops 3.16 Peer-reviewed scientific papers submitted to journals dealing with 0 3 effects 3.14 Regularly gave speeches/ lectures dealing with 0 3 effects on crops 2.51 1 Meaning of score: 5 = essential 4 = very important 3 0.94 0.91 0.92 0.99 0.91 0.89 0.98 237 237 237 237 237 237 237 1(<1) 23(10) 41(17) 104(44) 68(29) 4(2) 18(8) 96(41) 80(34) 39(16) 8(3) 22(9) 77(32) 105(44) 25(11) 7(3) 31(13) 70(30) 96(41) 33(14) 10(4) 42(18) 97(41) 77(32) 11(5) 10(4) 38(16) 107(45) 72(30) 10(4) 36(15) 86(36) 78(33) 32(14) 5(2) Continued = moderately important 2 = slightly important 1 = not important at all TABLE 5-3 (Cont'd) Descriptive Statistics: Indicators of Substantive Expertise Variable Mean Standard Frequency (%) by Score1 Score Deviation N 1 2 3 4 5 Conducted research primarily in a university setting 2.38 1.16 235 73(31) 50(21) 69(29) 36(15) 7(3) Appeared as an expert witness to provide testimony dealing with 0 3 effects 2.32 1.04 237 59(25) 81(34) 66(28) 25(11) 6(3) Conducted research primarily for government 2.26 1.06 235 77(33) 49(21) 83(35) 23(10) 3(1) Conducted research primarily for private industry 2.07 1.02 235 90(38) 61(26) 63(27) 20(9) 1(4) Held a teaching appointment at a post-secondary educational institution 1.63 0.77 237 127(54) 73(31) 34(14) 3(1) 0(0) How important is frequency of contributions to the literature? 1-5 publications 4.44 0.91 232 5(2) 3(1) 27(12) 46(20) 151(65) 6-10 publications 3.62 1.06 234 10(4) 21(9) 68(29) 83(35) 52(22) 11-20 publications 2.72 1.08 230 35(15) 59(26) 82(36) 44(19) 10(4) More than 20 publications 2.15 1.04 228 76(33) 71(31) 55(24) 23(10) 3(1) 1 Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all regarding it very important to essential. Some training in statistics (mean score = 3.74), atmospheric sciences (mean score = 3.46) and chemistry (mean score = 3.01) was deemed desirable. 5.2.1.3 Academic Grades Superior academic grades as a student are in general not essential to development of expertise. While a passing grade was considered essential by the majority of the sample (83%) and second class was considered very important (mean score = 4.03), achievement of first class grades was considered only moderately important (mean score = 3.12). Only 9% of the sample felt that a first class grade was essential in order to become an expert. 5.2.1.4 Length of Career Experience It was generally considered essential for an expert to have accumulated one to five years of career experience (mean score = 4.78). There was a lack of consensus of the relative importance of six to ten years experience, with 23% rating it essential, 46% rating it very important and 24% rating it only moderately important. Eleven or more years of career experience was considered only slightly to moderately important. 115 5.2.1.4 Type of Career Experience Respondents were asked to score 14 separate items describing different types of career experience. The descriptive statistics are presented in Table 5-3 and summarized below in order of decreasing importance. The mean importance score appears in parentheses after the description of the item. Item 1: Studied the majority of published articles dealing with 03 effects on crops (4.35) Expertise involves not only knowledge of one's own research efforts, but also requires knowledge of the work and results of others. Item 2: Participated in experiments dealing with 03 effects on crops (4.28) Participation in experiments is important to the development of expertise, and apparently cannot be obtained by simply studying the published results of experiments. Item 3: Authored scientific papers dealing with 02 effects on crops (3.91) Disseminating one's scientific results or views and obtaining feedback from peers is known to be highly important to the development of expertise. As discussed by Hogarth (1981), learning involves the use of feedback to generate, modify, maintain or abandon hypotheses, and feedback is central to the learning of expertise. 116 Item 4: Conducted statistical analysis related to defining plant exposure-response models (3.56) Statistical models are frequently used to define the relationships between air pollution levels and plant response. Item 5: Supervised experiments dealing with 03 effects on crops (3.49) Supervision of experiments may not be essential experience for experts. In fact, only 25 individuals (11%) concluded that supervision of such experiments is essential. Item 6: Regularly attends conferences and workshops dealing with 03 effects (3.49) Attendance at relevant conferences/workshops was considered only moderately to very important. Only 14% of the sample felt that regular attendance is essential. Item 7: Served as a peer reviewer in evaluating research regarding 03 effects on crops (3.16) Persons selected to peer-review the work of others are generally perceived as experts in the particular topic area. However, this does not seem to be an essential component of an expert's career experience. Only five persons (5%) judged such experience as being essential to the development of expertise. Item 8: Peer-reviewed scientific papers submitted to journals dealing with 03 effects (3.14) Peer-review of journal articles was considered only moderately important. Only 4% of the respondents felt that peer review of journal articles was essential experience. 117 Items 9 through 14: Other career experience The remaining career experience items were all rated as being of relatively low importance. These items, and their associated importance scores, were as follows: • Regularly gave speeches/lectures dealing with 0 3 effects on crops (2.51) • Employed by a university (2.38) • Appeared as an expert witness to provide testimony dealing with 0 3 effects (2.32) • Employed by government (2.26) • Employed by private industry (2.07) • Held a teaching appointment at a post-secondary educational institution (1.63) Respondents were asked to list other types of important career experience. Frequently suggested responses included: (i) diagnosis of injury in the field, (ii) policy-making experience, (iii) consulting on 0 3 effects, (iv) risk and/or impact assessment, (v) air pollution effects other than 0 3, (vi) interdisciplinary team experience, (vii) knowledge of local farming practices, (viii) experience with variation in crop susceptibility between regions, and (ix) agricultural eco-nomics training. 118 5.2.1.5 Frequency of Publications One to five publications in learned journals was considered essential (mean score = 4.44) by the majority (65%) of the respondents. The mean importance score for contributing six to ten publications was 3.62, for 11 to 20 publications was 2.72, and for more than 20 publications was 2.15. 5.2.2 Descriptive Statistics: Normative Expertise Descriptive statistics related to the relative importance of 22 cognitive or mental skills are presented in Table 5-4. These items are presented below in order of their relative importance as determined from the survey. Mean importance scores are given in parentheses following the description of the item. 5.2.2.1 General Intelligence and/or Intellect (4.50) It was concluded some time ago by Johnson (1963) that superior general intelligence is one of the characteristics of productive scientists. In the present research, general intelligence was considered to be the most important cognitive skill possessed by experts. There was strong consensus regarding this opinion; 130 persons (56%) rated it essential and 39% rated it very important. 119 TABLE 5-4 Descriptive Statistics: Indicators of Normative Expertise Variable Mean Score Standard Deviation N Frequency (%) by Score1 How important are certain cognitive or mental skills to the ability to provide risk estimates (judgments) of 0 3 impacts on crops? General intelligence and/or intellect Alertness in terms of observing relevant information or cues Ability to do induction (infer a general conclusion from particular instances) Ability to integrate diverse opinion Consistency in judgments and decisions Independent, autonomous thinking ability Ability to ignore irrelevance Ability 10 focus attention Judgmental ability (as rated by peers) 4.50 4.29 4.22 4.16 4.16 4.08 4.05 4.03 3.98 0.63 0.66 0.74 0.71 0.77 0.83 0.81 0.85 0.86 233 231 232 232 232 232 233 233 231 1(<1) 0(0) 1(<1) 0(0) 1(<1) 1(<1) 2(1) 1(4) 2(1) 0(0) 2(1) 2(1) 2(1) 6(3) 9(4) 5(2) 10(4) 11(5) 1 Meaning of score: 5 - essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all 11(5) 91(39) 130(56) 20(9) 118(51) 91(39) 31(13) 110(47) 88(38) 36(16) 116(50) 78(34) 30(13) 114(49) 81(35) 37(16) 108(47) 77(33) 43(18) 112(48) 71(30) 45(19) 103(44) 74(32) 42(18) 111(48) 65(28) ... Continued TABLE 5-4 (Cont'd) Descriptive Statistics: Indicators of Normative Expertise Variable Mean Standard Frequency (%) by Score1 Score Deviation N 1 2 3 4 5 Ability to express internal uncertainty (i.e., think probabilistically) 3.91 0.79 231 2(1) 4(2) 59(26) 114(49) 52(23) Independent, autonomous learning ability 3.85 0.88 231 2(1) 12(5) 62(27) 98(42) 57(25) Curiosity, inquisitiveness 3.79 0.88 232 2(1) 13(6) 67(29) 99(43) 51(22) Memory 3.72 0.78 232 2(1) 9(4) 74(32) 115(50) 32(14) Imagination and creativity 3.70 0.91 233 2(1) 20(9) 69(30) 96(41) 46(20) Decisiveness 3.67 0.86 233 1(<1) 17(7) 79(34) 96(41) 40(17) Foresightedness 3.58 0.95 231 6(3) 21(9)' 74(32) 94(41) 36(16) Mental quickness 3.57 0.87 232 4(2) 14(6) 91(39) 91(39) 32(14) Liking for method, precision, exactness 3.53 0.94 231 8(3) 18(8) 78(34) 97(42) 30(13) Mathematical/statistical ability 3.50 0.83 232 1(<1) 20(9) 00(43) 83(36) 28(12) Liking for abstract thinking 3.22 0.91 231 8(3) 34(15) 106(46) 65(28) 18(8) Liking for intellectual argument and debate 3.11 1.03 230 16(7) 42(18) 93(40) 59(26) 20(9) Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all 5.2.2.2 Alertness in Terms of Observing Relevant Information or Clues (4.29) To be "alert and observant" was found to be an important characteristic of a "superior" physician (Price et al., 1971), and the research reported here suggests it is a very important cognitive skill of the expert. Thirty-nine percent of the sample rated it essential and 51% rated it very important. This cognitive skill involves the ability to select relevant cues from information available and also involves the ability to ignore irrelevant information. 5.2.2.3 Ability to do Induction (Infer a General Conclusion From Particular Instances) (4.22) The ability to make accurate inferences is what experts aspire to, and, presumably for this reason, inductive skills are perceived as being very important. According to Lopes (1982), "there is probably no psychological process more important to individual survival than the ability to do induction." In many judgment situations, it has been shown that even experienced practitioners have poor ability to do accurate induction. This is because people do not have the cognitive schemata needed for efficient performance in probabilistic tasks, and because people seldom receive accurate feedback on the accuracy of their professional judgments (Lopes, 1982). 122 5.2.2A Ability to Integrate Diverse Opinion (4.16) Judgments and decisions are dependent upon selecting and weighting cues and informa-tion from one's environment; experts have superior ability to integrate diverse opinion into an appropriate judgment. 5.2.2.5 Consistency in Judgments and Decisions (4.16) According to Einhorn (1986), high intrajudge reliability (consistent judgment) is a necessary condition for expertise. This hypothesis is supported by the present research, as consistency in judgments was rated very important. Task complexity and task uncertainty are known to affect consistency (Brehmer, 1986). This has been interpreted to mean that judgment is partially a matter of skill; although a person intends to use a certain rule, the judgment may not follow that rule. 5.2.2.6 Independent, Autonomous Thinking Ability (4.08) High autonomy and independence of judgment were also found to be important charac-teristics of productive scientists by Johnson (1963). It was pointed out by Einhorn (1986) that it is sometimes the "oddball" who did not agree with others who was later shown to be correct. 123 5.2.2.7 Ability to Ignore Irrelevance (4.05) It was suggested by Einhorn (1986) that the expert's ability to identify information or cues from the multidimensional stimuli encountered can be thought of as a problem of "extracting weak signals from a background of noise". Experts build up expectations regarding relationships, and these expectations may vary between scientists. This is largely due to different experience and training, reflecting each expert's way of organizing information into clusters or dimensions. This clustering of information reduces the complexity of information processing (Einhorn, 1986). The process of information search will, for the expert, result in selection of cues that yield the greatest amount of information, and this leads to a shortening of the decision tree used to reach a judgment. Studies on the accuracy of clinical judgments have yielded discouraging results. This is partially related to the inability of persons to ignore irrelevance. As the amount of (irrelevant) information is increased, the confidence in judgments increases, although the accuracy frequently decreases (Goldberg, 1968). 5.2.2.8 Ability to Focus Attention (4.03) Hogarth (1980) stated that focusing attention is a necessary condition for good judgment. In his words: Consider what needs to be done. Information from both the environment and the individual's memory needs to be selected. Meaning is given to the information and, indeed, such meaning may even guide the search process in the first instance. The various sources of information selected then have to be weighted and combined to form a final judgment. 124 5.2.2.9 Judgmental Ability (as Rated by Peers) (3.98) Research on human judgment has shown that statistical models consistently outperform human judgmental predictions (Hogarth, 1980). Poor performance by humans is largely due to the fact that they have limited information-processing capacity and, as a result, employ cognitive simplification strategies or heuristics which, although generally efficient for most routine day-to-day tasks, can sometimes lead to systematic errors (Tversky and Kahneman, 1974). It must therefore be acknowledged that experts, being human, are prone to several sources of judgmental error. Nevertheless, it is generally believed that certain individuals demonstrate better judgment than others. 5.2.2.10 Ability to Express Internal Uncertainty (ie.. Think Probabilistically) (3.91) As indicated previously, formal (statistical) methods have been shown to be superior to methods which rely on judgment. Humans tend not to think probabilistically, or if they do, their weightings of uncertain data are often inappropriate. However, it is believed by researchers that if one thinks probabilistically, the quality of decisions and judgments will be improved (Hogarth, 1980). 125 5.2.2.11 Independent, Autonomous Learning Ability (3.85) Independent, autonomous learning was found to be a common characteristic of productive scientists (Johnson, 1963). The importance of learning in the judgment process is well established (Hogarth, 1980). 5.2.2.12 Curiosity, Inquisitiveness (3.79) This characteristic is probably related to the cognitive skill associated with imagination and creativity (Hogarth, 1980) and is a commonly observed characteristic in academics and other successful professionals. 5.2.2.13 Memory (3.72) The results suggest an above average memory is moderately to very important. Persons who can remember more presumably have the potential of making better judgments because of the larger knowledge base at their disposal. Superior recall is a characteristic of experts observed by Shanteau (1984). According to Hogarth (1980), a good memory is a necessary condition for good judgment. Memory affects judgment in several ways: (i) the manner in which judgmental tasks are structured; (ii) cues that are selected, either from the environment or from memory; (iii) the rule used to process the information; and (iv) the interpretation and "coding" of the outcome. 126 It has been shown that human memory is frequently erroneous, even when the expert is very confident in what is being recalled (Hogarth, 1980). One researcher (Arkes, 1981) suggests that in order to make better judgments, less reliability should be placed in memory. 5.2.2.14 Imagination and Creativity (3.70) The importance of this attribute may manifest itself in the expert's ability to discern or form cues that no one has seen before (Hogarth, 1980). The expert can discern and use contingent relationships between cues, whereas the novice cannot (Einhorn, 1986). The importance of imagination and creativity was discussed by Simon (1986), who stated "a large part of our problem solving consists of the search for good alternatives, or for improvements in alternatives that we already know." 5.2.2.15 Decisiveness (3.67) Decisiveness is an important characteristic of "superior" physicians in practice (Price et al., 1971), although some deep scientific questions are not conducive to reaching rapid con-clusions. 5.2.2.16 Foresishtedness (3.58) Foresight, or predicting the future, requires "considerable powers of imagination and both the ability and willingness to entertain several hypotheses simultaneously" (Hogarth, 1980). 127 Thus, it appears that the ability to construct good causal explanations is important in prediction, and accurate prediction depends on identifying key variables in the environment and their relationship to the event predicted. 5.2.2.17 Mental Quickness (3.57) Although it has been documented that some persons have quicker recall or cognitive processing ability than others (Hogarth, 1980), there has been little or no research which addresses the following questions: Is mental quickness an important characteristic of the expert? Can persons who tend to "brood" about deep problems, or persons who do not reach rapid conclusions or judgments about problems be considered experts? 5.2.2.18 Liking for Method. Precision, Exactness (3.53) Liking for method, precision and exactness is a characteristic of a productive scientist as observed by Johnson (1963). 5.2.2.19 Mathematical/Statistical Ability (3.50) People have difficulty in adjusting base-rate probabilities by specific information; they lack the ability to do the mental calculations (Tversky and Kahneman, 1974). Persons with a statistical background are likely to be better judges under uncertainty than those without this 128 background, although there is research to indicate that even statisticians may not be immune from this problem (Tversky and Kahneman, 1974). 5.2.2.20 Liking for Abstract Thinking (3.22) Liking for abstract thinking is a characteristic of productive scientists (Johnson, 1963). It is presumably indicative of the creative mind. As pointed out by Hogarth (1980), "effective creativity requires free-wheeling, imaginative (or even irrational) thought processes, as well as logical structures to be able to evaluate the potential usefulness of ideas and solutions." 5.2.2.21 Liking for Intellectual Argument and Debate (3.11) Liking for intellectual argument and debate, a further characteristic of productive scientists (Johnson, 1963), was considered to be only moderately important by the majority of respondents in the present survey. 5.2.2.22 Additional Cognitive or Mental Skills Respondents were asked to list additional important cognitive or mental skills. Additional cognitive skills frequently suggested included: (i) ability to be objective, (ii) communicate science at a policy level, (iii) collate in the mind (distill and synthesize), (iv) avoid bias and (v) understand the distinction between scientific issues and policy issues. 129 5.2.3 Descriptive Statistics: External Credibility Descriptive statistics related to the relative importance of various personal qualities or styles related to external credibility are summarized in Table 5-5. Several items were considered as being very important to essential personal qualities, and there was generally strong consensus regarding the relative importance of these items. Items that received a mean score of 4.0 or higher, in order of decreasing relative importance, included: honesty, open-mindedness, willingness to admit ignorance, ability to communicate, forthrightness, willingness to reject group thinking, conscientiousness, interest in work/research and desire to discover scientific truth. Items receiving a mean score between 3.0 and 3.99 were considered moderately important to very important. In general, there was less consensus regarding the relative importance of these items. In order of decreasing importance, these included: confidence, constructiveness when reviewing others' work, ability to speak, emotional stability, sympathy or concern for problems underlying work/research, conservativeness in data interpretation, and satisfaction and pride in career. Personal qualities that were considered only slightly to moderately important (mean scores of 2.99 or less) included friendliness, modesty and willingness to work overtime. 5.2.4 Summary of Importance of Indicators A summary of the relative importance of the 76 indicators of expertise evaluated in the survey is presented in Table 5-6. 130 TABLE 5-5 Descriptive Statistics: Indicators of External Credibility Variable Mean Standard Frequency (%) by Score1 Score Deviation N 1 2 3 How important is it for an expert to possess an above average level of each of the following personal qualities? Honesty 4.79 0.47 232 0(0) 0(0) 6(3) 37(16) 189(81) Open-mindedness 4.41 0.75 232 2(1) 2(1) 19(8) 85(37) 124(53) Willingness to admit ignorance of a specific topic or problem 4.39 0.77 232 3(1) 2(1) 17(7) 89(38) 121(52) Ability to communicate 4.31 0.66 233 0(0) 2(1) 19(8) 117(50) 95(41) Forthrightness 4.29 0.61 229 0(0) 1(0) 16(7) 128(56) 84(37) Willingness to reject group thinking 4.27 0.80 230 2(1) 3(1) 30(13) 90(39) 105(46) Conscientiousness 4.27 0.75 232 1(<1) 4(2) 24(10) 106(46) 97(42) Interest in work/research 4.16 0.79 232 2(1) 4(2) 32(14) 112(48) 82(35) Desire to discover scientific truth 4.13 0.93 232 4(2) 8(3) 39(17) 85(37) 96(41) Confidence 3.85 0.80 233 2(1) 8(3) 58(25) 119(51) 46(20) ... Continued 1 Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all TABLE 5-5 (Cont'd) Descriptive Statistics: Indicators of External Credibility Variable Mean Standard Frequency (%) by Score1  Score Deviation N 1 2 3 4 5 Constructiveness (in reviewing work of others) 3.75 0.87 233 5(2) 10(4) 64(27) 113(48) 41(18) Ability to speak (articulateness) 3.70 0.88 233 6(3) 11(5) 66(28) 113(48) 37(16) Emotional stability 3.63 1.01 231 8(3) 21(9) 65(28) 92(40) 45(19) Sympathy or concern for problems underlying work/research 3.52 0.94 230 9(4) 20(9) 69(30) 107(47) 25(11) Conservativeness (in data interpretation) 3.33 0.85 228 7(3) 21(9) 103(45) 83(36) 14(6) Satisfaction and pride in career 3.32 1.01 233 12(5) 31(13) 88(38) 75(32) 27(12) Friendliness 2.77 1.10 232 34(15) 58(25) 80(34) 47(20) 13(6) Modesty 2.75 1.09 233 38(16) 49(21) 91(39) 44(19) 11(5) Time spent working overtime 2.19 1.07 229 79(34) 57(25) 68(30) 21(9) 4(2) 1 Meaning of score: 5 = essential 4 = very important 3 = moderately important 2 = slightly important 1 = not important at all TABLE 5-6 Summary of Importance of Expert Indicators Category Very Important to Essential1 Moderately Important to Very Important2 Slightly Important to Moderately Important3 Education (level) B.S. degree M.S. degree Education (discipline) Plant Science Botany Performance as a Student Length of Career Experience Type of Career Experience Frequency of Publications Cognitive Skills Personal Qualities Pass Second Class 5 years Reads published literature Participates in experiments 1-5 publications Intelligent Alert Can do induction Can integrate opinion Consistent Independent thinker Can ignore irrelevance Can focus attention Honesty Open-minded Wil l admit ignorance Good communicator Forthright Can reject group thinking Conscientious Enjoys work Desire to discover truth Ph.D. degree Statistics Atmospheric Science Chemistry First Class 6 to 10 years Publishes papers Conducts analysis Supervises experiments Attends workshops/conferences Peer reviews research Peer reviews papers 6-10 publications Good judgment Can express uncertainty Independent learner Inquisitive, curious Good memory Imaginative, creative Decisive Good foresight Mentally quick Likes precision Mathematical ability Abstract thinker Enjoys debate Confident Constructive reviewer Good speaker Emotionally stable Concern for environment Conservative Pride in career Post-doctorate Mathematics Physics 11 to 20 years 21 years or more Gives speeches/lectures Work in university Expert witness Work for government Work for industry Teaching appointment 11-20 publications More than 20 publications Friendly Modest Works overtime 1 Mean importance score of 4.0 to 5.0 2 Mean importance score of 3.0 to 3.99 3 Mean importance score of 2.0 to 2.99 133 5.3 Underlying Attributes of Expertise 5.3.1 Purpose and Methods The data obtained from the expert survey were analyzed using factor analysis in order to: (i) Identify and group together different measures of the major underlying attributes of expertise, and (ii) Assist in the determination of the construct validity of the questionnaire, i.e., to provide evidence whether the questionnaire measured what it was supposed to measure. The factor analysis was conducted in two stages. The first stage involved analysis of variables within separate questions from the questionnaire, including education, academic achievement, length of career experience, type of career experience, number of publications, cognitive abilities and personal characteristics. The Stage I factor analysis resulted in the identification of 23 factors based on the original 76 variables. Because there still appeared to be a certain amount of redundancy in the data, a second factor analysis (Stage II) was conducted using the 23 factors from Stage I as the input variables, as suggested by Cooley and Lohnes (1971), and Joreskog (1979). This resulted in the identification of seven major underlying factor constructs of expert attributes. An analysis was also conducted to determine the relative degree of consensus regarding the underlying factors among different categories of people within the sample. Analysis of 134 variables (ANOVA) of factor scores between groups was used for this purpose. The various groups included in the ANOVA were: Plant scientists (112 individuals) compared to others (126 individuals), • Respondents with less experience (ten years or less) (69) compared to those with more than ten years of experience (169), • Canadians (56) compared to Americans (164) and others (10), Researchers (180) compared to non-researchers (58), • Consultants (54) compared to non-consultants (184), • Policy-makers (47) compared with others (191), • Enforcement people (20) compared with others (218), Administrators (55) compared with others (183), • Teachers (45) compared with others (193), • University employees (89) compared with others (149), Government employees (98) compared with others (140), • Industry employees (21) compared with others (217), Experts nominated > five times (25) compared with others (213), • Experts nominated > ten times (11) compared with others (227), and • Members of conservation groups (14) compared with others (224). 5.3.2 Results of Stage I Factor Analysis The results of the factor analysis of grouped items from the questionnaire is presented below. Factors were initially interpreted based on factor loadings, which are the correlations 135 between the factors and the original variables. These loadings, like correlation coefficients, vary from -1.0 to +1.0. All variables having factor loadings exceeding the root mean square of all values in the correlation matrix are considered "high loadings", and are flagged with an asterisk (*). The amount of variance of a variable that is explained by the factor is called the communality and is designated by h2. Communality equals the sum of the squared loading of a variable on factors. While h2 provides an indication of the reliability of a variable to a factor, (1-h2) provides an indication of the uniqueness of a variable (Kinnear and Taylor, 1983). Factor scores were computed by selecting a set of variables that loaded highly on the factor and then dividing by the number of variables in the factor. Since the factor scores were based on the original importance scale, interpretation of the factors remains as before (e.g., 5 = essential, 4 = very important, 3 = important, 2 = slightly important, 1 = not important). The mean factor scores for the entire sample, as well as the various groups within the sample that differ significantly (p < 0.05) in their mean factor scores, are reported below. As indicated previously, ANOVA was used to test for differences in means between groups. 5.3.2.1 Education (degree obtained) Table 5-7 summarizes the findings for the education (degree obtained) dimension. There are two variables associated with each of the two factors. The post-doctorate and Ph.D. variables load highly on Factor 1 (.89 and .87, respectively), while the B.S. and M.S. degree variables load highly on Factor 2 (.91 and .72, respectively). 136 TABLE 5-7 Results of Factor Analysis on "Education (Degree Obtained)" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 h2 Completed post-doctorate •89(*) -.03 .79 Ph.D. degree or equivalent .87(*) .21 .81 B.S. degree or equivalent -.09 .91 (•) .83 M.S. degree or equivalent .45 .72 C ) .72 Eigenvalue 1.76 1.39 3.15 Percent variance explained 44% 35% 79% 1 Values greater than RMS 0.63 have been flagged by an (*). (b) Mean Factor Scores and Differences Between Groups Group (N) Total sample (238) Location Canada (56) U.S.A. (164) Other (10) Researcher (180) Others (58) Administrator (55) Others (183) Teacher (45) Others (193) Less Experience (69) Others (169) Top experts (11) Others (227) Factor 1 3.25 2.85 3.35 3.55 3.34 2.90 3.47 3.17 3.68 3.14 Factor 2 4.42 4.57 4.37 4.04 4.47 137 Factor 1, which may be interpreted as "advanced academic preparation", accounts for 44% of the explained variance. Factor 2, interpreted as "basic academic preparation", accounts for 35% of the explained variance. The two factors together explain 79% of the variance. The mean factor score for Factor 1 is 3.25 and for Factor 2 it is 4.42, indicating that advanced academic preparation is moderately to very important, while basic academic preparation is very important to essential. However, there were some differences in opinion, as shown in Table 5-7. Canadians considered advanced academic preparation less important (2.85) than Americans (3.35) or Europeans (3.55). Researchers considered advanced academic preparation more important (3.34) than non-researchers (2.90), as did administrators (3.47) compared to non-administrators (3.17), and teachers (3.68) compared to non-teachers (3.14). Respondents with ten years or less experience considered basic academic preparation (Factor 2) more important (4.57) than respondents with more than ten years experience (4.37). Top experts nominated in the survey considered basic academic preparation somewhat less important (4.04) than the remainder of the sample (4.47). 5.3.2.2 Education (discipline) The results for education of discipline are contained within Table 5-8. Two factors were identified and interpreted as follows: Factor 1: "General sciences background" Factor 2: "Plant or botanical sciences background" 138 TABLE 5-8 Results of Factor Analysis on "Education (Discipline)" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 h2. Mathematics .81 (*) .06 .66 Physics .81 (*) -.04 .66 Chemistry .64 (*) .06 .42 Atmospheric Science .63 (*) .05 .41 Statistics -56 (*) 2 9 3 9 Plant Science -.02 .84 (*) .71 Botany .15 .79 (*) .65 Eigenvalue 2.46 1.43 3.89 Percent variance explained 35% 20% 55% 1 Values greater than RMS 0.53 have been flagged by an (*). (b) Mean Factor Scores and Differences Between Groups Group fN) Fac Total sample (238) 2. Consultants (54) 2. Others (184) 2. Policy jobs (47) Others (191) University jobs (89) Others (149) Enforcement jobs (20) Others (218) 139 The mean factor score for Factor 1 is 2.40, indicating a general scientific background is only slightly to moderately important. Consultants regarded the general sciences background more important (2.75) than others (2.38). The mean importance score for Factor 2 is 4.38, indicating that a background in plant or botanical sciences is very important to essential. Individuals with policy jobs rated this background more important (4.62) than others (4.31), as did those with university jobs (4.41) compared to others (4.34), and those with enforcement jobs (4.74) compared to others (4.34). 5.3.2.3 Academic Performance (Grade) Only one factor regarding academic performance was extracted, and it accounted for 74% of the explained variance. The mean factor score is 4.18, indicating it is very important to essential to have achieved a certain level of academic performance while in university or college. There were no differences in opinion between groups on this factor (Table 5-9). 5.3.2.4 Length of Career Experience As shown in Table 5-10, two factors related to length of career were identified. These were interpreted as follows: Factor 1: "Lengthy career experience" Factor 2: "Minimal career experience" 140 TABLE 5-9 Results of Factor Analysis on "Academic Performance (Grade)" (a) Factor Loadings (one factor - no rotation)1 Factor 1 Second class average .95 (*) Pass .81 First class average .81 Eigenvalue 2.22 Percent variance explained 74% 1 Values greater than RMS 0.86 have been flagged by an (*). (b) Mean Factor Scores and Differences Between Groups Group (N) Factor 1 Total sample (238) 4.18 (No differences between groups) 141 TABLE 5-10 Results of Factor Analysis on "Length of Career Experience" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 h2 11 to 20 years .93 (*) .16 .91 21 years or more .90 (*) -.07 .70 6 to 10 years .68 (*) .50 .89 5 years or less .04 .95 (*) .82 Eigenvalue 2.14 1.19 3.32 Percent variance explained 53% 30% 83% 1 Values greater than RMS 0.65 have been flagged by an (*). (b) Mean Factor Scores and Differences Between Groups Group (N) Fact Total sample (238) 2.9: Researchers (180) 3.0( Others (58) 2.0S 142 The mean factor score for Factor 1 is 2.93, suggesting that lengthy career experience is only moderately important. Researchers consider lengthy career experience to be more important (3.00) than non-researchers (2.09). Some career experience (Factor 2) was generally considered essential (4.78). There was consensus between groups regarding the importance of this factor. 5.3.2.5 Number of Publications Two factors related to number of publications were identified (Table 5-11). These were interpreted as follows: Factor 1: "Lengthy list of relevant publications" Factor 2: "Several relevant publications" The mean score for Factor 1 is 2.46, indicating that a lengthy publications list (greater than ten) is only slightly to moderately important. However, administrators (2.76), teachers (2.93), individuals employed by universities (2.76), plant scientists (2.53) and researchers (2.52) considered a lengthy publications list more important. Having several relevant publications (Factor 2) was considered very important (4.06), and there was general consensus in this regard. 143 TABLE 5-11 Results of Factor Analysis on "Scientific Recognition (Publications)" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 h2 More than 20 publications .95 (*) .01 .90 11-20 publications .90 (*) .30 .90 6-10 publications .63 .68 (*) .87 1-5 publications .05 .97 (*) .94 Eigenvalue 2.11 1.50 3.60 Percent variance explained 53% 37% 90% 1 Values greater than RMS 0.67 have been flagged by an (*). (b) Mean Factor Scores and Differences Between Groups Group fN) Fact Total sample (238) 2.M Administrators (55) 2.7( Others (183) 2.3( Teachers (45) 2.92 Others (193) 2.3< University employees (89) 2.7t Others (149) 2.2< Plant scientists (112) 2.5: Others (126) 2.3$ Researchers (180) 2.5: Others (58) 2.2: 144 5.3.2.6 Type of Career Experience Sixteen career experience variables were grouped into four factors that were interpreted as follows (Table 5-12): Factor 1: "General research experience" Factor 2: "Conducted experiments and produced results" Factor 3: "Active involvement in scientific field" Factor 4: "Reviews up-to-date scientific data and results" General research experience (Factor 1) was only considered slightly important (2.21). Researchers (2.64), consultants (2.44), teachers (2.40) and university employees (2.40) considered this type of experience somewhat more important than others. Government employees (2.02) and experts (1.83) considered general research experience to be less important. The importance score for Factor 2 is 3.80, which indicates that conducting experiments and producing experimental results is very important. Teachers (3.98) and plant scientists (3.89) considered this type of experience to be more important than others. Factor 3, which involves active involvement and communication of scientific information, was rated slightly to moderately important (2.46). Government employees (2.38) and experts (2.20) considered it to be less important. Factor 4 involves review of scientific data and results and was considered quite important (3.54). Europeans (3.87) considered this type of experience to be more important 145 TABLE 5-12 Results of Factor Analysis on "Type of Career Experience" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 Factor 3 Factor 4 h2 Government researcher .95 (*) .06 .11 .02 .91 University researcher .91 (*) .09 .13 .03 .86 Industry researcher .91 (*) -.08 .11 -.04 .85 Participated in experiments -.04 .83 (*) .06 -.09 .70 Supervised experiments .04 .75 (*) .18 .06 .61 Authored papers -.05 .66(*) .20 .33 .58 Conducted statistical analysis .16 .48 (*) .03 .23 .31 Gave speeches .09 .14 .85 (•) -.03 .74 Attends conferences .02 .12 .68(*) .06 .48 Expert witness .34 .08 -55 (•) .08 .43 Teach at university .43 .09 -48 C) .18 .45 Studied published articles .01 .02 -.12 .79 (•) .64 Research program peer-reviewer .07 .23 .42 .58 (*) .57 Peer-reviewed journal papers -.01 .42 .28 .55 (•) .55 Eigenvalue 2.89 2.22 2.09 1.47 8.67 Percent variance explained 21% 16% 15% 10% 629 ... Continued 1 Values greater than RMS 0.39 have been flagged by an (*). 146 TABLE 5-12 (Cont'd.) Results of Factor Analysis on "Type of Career Experience" (b) Mean Factor Scores and Differences Between Groups Group (N) Factor 1 Factor 2 Factor 3 Factor 4 Total sample (238) 2.21 3.80 2.46 3.54 Location: Canada (56) 3.28 USA (164) 3.61 Others (10) 3.87 Researchers (180) 2.64 Others (58) 2.13 Consultants (54) 2.44 3.36 Others (184) 2.19 3.58 Teachers (45) 2.44 3.98 Others (193) 2.20 3.76 University employees (89) 2.40 Others (149) 2.15 Government employees (98) 2.02 2.38 Others (140) 2.40 2.55 Experts (25) 1.83 2.20 Others (213) 2.30 2.51 Plant scientists (112) 3.89 Others (126) 3.71 147 than Americans (3.61) or Canadians (3.28). Consultants considered it to be less important (3.36) than others. 5.3.2.7 Cognitive Abilities The 21 cognitive attributes comprised six separate factors (Table 5-13) that were interpreted as follows: Factor 1: "Conceptual and creative abilities" Factor 2: "Independent thinking ability" Factor 3: "Decisive and consistent" Factor 4: "Powers of inference; inductive abilities" Factor 5: "Information retrieval and processing ability" Factor 6: "Ability to synthesize and integrate" Factors 4 and 6 were considered most important, with scores of 4.04 and 4.07, respectively. General consensus existed between groups regarding the importance of Factor 4. There was some disagreement between groups regarding the importance of Factor 6. Teachers considered it to be more important (4.21), as did university employees (4.20). Government employees considered it less important (3.92). Factors 2 and 3 were considered very important, with factor scores of 3.96 and 3.99 respectively. Individuals with enforcement jobs (4.08) and university employees (4.05) regarded independent thinking ability (Factor 2) to be more important. 148 TABLE 5-13 Results of Factor Analysis on "Cognitive Abilities" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 Liking for intellectual argument/debate .70 (*) .17 Liking for abstract thinking .70 (•) .17 Curiosity, inquisitiveness .64 (*) .27 Imagination and creativity .47 (*) .13 Liking for method, precision, exactness .43 (*) .33 Independent, autonomous thinker .22 .79 (*) Independent, autonomous learner .33 .76 (*) Alertness with relevant information/cues -.07 .51 (*) Foresightedness .29 .46(*) Decisiveness .26 .06 Consistency in judgments and decisions .02 .00 Ability to focus attention .01 .25 Mathematical/statistical ability .32 .01 Ability to ignore irrelevance -.26 .13 General intelligence and/or intellect .22 -.07 Ability to do induction .16 .25 Ability to express internal uncertainty .11 .24 Memory .06 .08 Mental quickness .28 .00 Judgmental ability .17 -.11 Ability to integrate diverse opinion .03 .27 Eigenvalue 2.47 2.24 Percent variance explained 12% 11% 1 Values greater than RMS 0.31 have been flagged by an (*). Factor 3 Factor 4 Factor 5 Factor 6 h2. .24 .04 -.03 .27 .65 .07 .14 .11 .23 .61 .00 .22 .24 -.27 .66 .25 .38 .18 -.11 .49 .00 .06 .36 .06 .43 .06 .19 .03 .08 .72 .11 .07 .02 .02 .71 .09 .11 .50 -.01 .54 .40 (*) -.15 .19 .28 .59 -82(*) -.06 .12 .15 .79 .73 (•) .11 .06 .05 .55 .66(*) .33 .09 -.02 .62 .09 .64(*) -.19 -.01 .56 .08 .57 (*) .12 .37 .56 -.12 .55 (*) .45 .09 .58 .24 .45 (•) .32 .18 .49 .16 .45 (•) .20 .14 .36 .12 .15 .75 (*) -.07 .62 .20 -.04 .61 (*) .29 .58 .09 .10 .12 .76 (•) .65 .07 .16 -.04 .69 (•) .58 2.14 1.92 1.90 1.66 12.33 10% 9% 9% 8% 59% ... Continued TABLE 5-13 (Cont'd.) Results of Factor Analysis on "Cognitive Abilities" (b) Mean Factor Scores and Differences Between Groups Group (N) Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Total sample (238) 3.46 3.96 3.99 4.04 3.67 4.07 Enforcement jobs(20) 4.08 Others (218) 3.94 Teachers (45) 4.21 Others (193) 4.04 University employees (89) 4.05 4.20 Others (149) 3.89 3.99 Government employees (98) 3.74 3.92 Others (140) 3.59 4.19 Conservation groups (14) 3.59 Others (224) 3.89 There was consensus regarding the importance (3.99) of Factor 3 (decisiveness and consistency), with the exception of the conservation group, who regarded it to be of lower importance (3.59). Factors 1 and 5 were the least important of the six cognitive factors, with scores of 3.46 and 3.67, respectively. There was general consensus between groups regarding the relative importance of Factor 1, which involves conceptual-creative abilities. Regarding information retrieval and processing ability (Factor 5), only government employees disagree regarding its relative importance; they rated it slightly more important (3.74) than others (3.59). 5.3.2.8 Personal Qualities The 19 personal qualities variables comprised 4 factors (Table 5-14), with the following interpretations: Factor 1: "Intellectual honesty" Factor 2: "Career orientation" Factor 3: "Interpersonal relations" Factor 4: "Communication skills" The mean score for Factor 1 is 4.37, indicating intellectual honesty is a very important to essential attribute of the expert. However, there was some disagreement as to the relative importance of this factor. Europeans (4.65) and Americans (4.44) considered it to be more important than Canadians (4.13). Researchers (4.43) considered intellectual honesty to be 151 TABLE 5-14 Results of Factor Analysis on "Personal Qualities" (a) Factor Loadings After Varimax Rotation1 Factor 1 Factor 2 Factor 3 Factor 4 hi Open-minded -80(*) .15 .07 .02 .67 Honesty .72 (*) -.04 .07 .11 .54 Willingness to admit ignorance .70 (*) .10 .11 -.01 .52 Reject group thinking .64(*) .13 .09 .13 .46 Conscientious •57 n .45 .07 .00 .54 Discover scientific truth .47 (*) .45 .09 -.14 .46 Constructive •36 (•) .31 .33 .28 .42 Enjoys career -.01 .79 (•) .17 .05 .66 Interest in work .20 .73 (*) -.05 .15 .60 Concern for environment .20 .66 O .11 .17 .52 Works overtime .04 -51 (*) .33 .06 .37 Modest .06 .11 .86 (•) .02 .75 Friendly -.01 .32 .72 C) .22 .67 Interpret data conservatively .30 .05 .60 (*) .12 .46 Emotionally stable .17 .03 .47 (•) .39 .40 Communication .08 .03 .01 .82 (•) .68 Articulateness -.08 .06 .18 .79 (•) .66 Confident .08 .33 .15 -54(*) .43 Forthrightness .36 .06 .21 -40(*) .34 Eigenvalue 3.11 2.66 2.23 2.16 10.16 Percent variance explained 16% 14% 12% 11% 539 ... Continued 1 Values greater than RMS 0.37 have been flagged by an (*). 152 TABLE 5-14 (Cont'd.) Results of Factor Analysis on "Personal Qualities" (b) Mean Factor Scores and Differences Between Groups Group (N) Factor 1 Factor 2 Factor 3 Factor 4 Total sample (238) 4.37 3.30 3.13 4.04 Less experience (69) 3.48 Others (169) 3.23 Location: Canada (56) 4.13 3.14 USA (164) 4.44 3.34 Others (10) 4.65 3.45 Researchers (180) 4.43 Others (58) 4.17 Enforcement jobs (20) 4.29 Others (218) 4.38 Administrators (55) 4.19 Others (183) 4.00 Teachers (45) 3.35 Others (193) 3.06 Government employees (98) 4.32 3.07 Others (140) 4.41 3.15 Consultants (54) 3.42 Others (184) 3.08 153 somewhat more important than did individuals with enforcement jobs (4.29) or other government employees (4.32). Factor 2, career orientation, was considered to be moderately important (3.30) and there was consensus in this regard, with the exception of individuals with less experience (3.48). Factor 3, interpersonal relations, was also rated moderately important (3.13). Europeans considered this quality to be more important (3.45) than Americans (3.34) or Canadians (3.14). Teachers (3.35) and consultants (3.42) considered it more important, while government employees (3.07) considered it less important. Factor 4, communication skills, was considered very important (4.04). There was general consensus regarding the importance of this personal quality, with the exception of administrators, who perceived it to be more important (4.19). 5.3.2.9 Summary of Stage I Factor Analysis Factor analysis of the 76 variables resulted in 23 factors representing underlying attributes. The relative importance of these 23 attributes, based on mean factor scores from the entire sample, is summarized in Table 5-15. ANOVA of mean factor scores demonstrated that there was generally consensus between groups regarding the relative importance of the 23 attributes. Although some groups were found to differ significantly (p < .05) in their mean factor scores on the various attributes (Table 5-16), the differences in scores between groups were generally minor in comparison with the mean importance score for the overall sample. 154 TABLE 5-15 Relative Importance of Stage I Factors Category Very Important to Essential1 Moderately Important to Very Important2 Slightly Important to Moderately Important3 Education (level) Basic academic preparation Advanced academic preparation Education (discipline) Plant Science/Botany background General science background Performance as a Student Pass or better Length of Career Experience 5 years experience 6 years or more experience Type of Career Experience Conducted experiments Reviews scientific data General research Active involvement in scientific field Frequency of Publications 1-10 publications 11 or more publications Cognitive Abilities Inductive ability Ability to synthesize and integrate Conceptual and creative ability Independent thinking ability Decisive and consistent Information retrieval and processing ability Personal Qualities Intellectual honesty Communication skills Career orientation Interpersonal relations 1 Mean importance score of 4.0 to 5.0 2 Mean importance score of 3.0 to 3.99 3 Mean importance score of 2.0 to 2.99 155 TABLE 5-16 Groups with Significantly Different Factor Scores Expert Attribute 1 2 3 4 5 6 7 8 9 10 i i 12 13 14 15 Advanced academic preparation J y J J Basic academic preparation J y General science background J J Plant science background J y Student academic performance Lengthy experience (> 6 years) y Minimal experience (1 to 5 years) Lengthy list of publications (> 10) J y J J y Less publications (10 or less) General research experience y J J y y y Conducted experiments J J Active involvement in scientific field y y Reviews up-to-date data and results J J Conceptual and creative abilities Independent thinking ability J y Decisive and consistent y Inductive abilities Information retrieval and processing ability Ability to synthesize and integrate Intellectual honesty J j y Career orientation J Interpersonal relations J J J y Communication Skills J Groups: 1. Plant scientists 6. 2. More experienced 7. individuals (> 10 years) 8. 3. Canadians 9. 4. Researchers 10. 5. Consultants Policy-makers Enforcement Administrators Teachers University employees 11. Government employees 12. Industry employees 13. Top experts (> 10 nominations) 14. Experts (10 or less nominations) 15. Conservation organization 156 5.3.3 Results of Stage II Factor Analysis A second (Stage II) factor analysis was conducted, using the 23 underlying factors from Stage I as input variables. The objective of the two-stage factor analysis, as recommended by Cooley and Lohnes (1971) and Joreskog (1979), was to further reduce the number of underlying factors, since it was believed that a certain amount of redundancy still existed in the data following the Stage I factor analysis. Grouping of the Stage I factors into fewer underlying constructs eliminates this redundancy. The resulting profile of attributes, while less complex, still retains much of the information contained in the original data. The results of the Stage II factor analysis are summarized in Table 5-17. The 23 Stage I factors have been grouped into seven Stage II factor constructs. Each of these is interpreted and described below. Factor 1: Normative Expertise A total of seven Stage I factors loaded highly on this construct, including: (i) independent thinking ability, (ii) inductive ability, (iii) conceptual/creative ability, (iv) decisiveness, (v) communication skills, (vi) general sciences background, and (vii) information retrieval and processing ability. This dimension may be interpreted as consisting of various cognitive or mental skills known to be important in human judgmental processes. Persons displaying these qualities may be thought of as possessing normative expertise. 157 TABLE 5-17 Results of Stage II Factor Analysis (a) Factor Construct Loadings After Varimax Rotation1 Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 h 2 Independent thinking ability .74 (*) -.06 .20 .05 .09 .07 -.05 .61 Inductive ability .70 (*) .10 .10 .08 .08 .12 .08 .54 Conceptual, creative ability .67 (•) .11 .31 .00 .33 .04 -.09 .67 Decisiveness .66 (•) .19 .04 .12 -.04 -.18 -.09 .53 Communication skills 52 n .08 .29 .32 .03 -.15 .17 .51 General sciences background 30 C) -.03 -.18 .14 .33 .30 .31 .60 Information retrieval and processing .48 n .13 .28 .24 -.05 -.11 .31 .50 Considerable publications .12 •85C) .08 .12 .17 -.08 .01 .80 Minimal publications .15 .71 O .13 .23 -.04 .26 -.09 .67 Career orientation .31 -.07 .66(*) .15 .28 .05 -.15 .66 Intellectual honesty .42 .10 -57 O -.07 .00 .22 -.05 .57 Interpersonal relations .30 .05 57 n .11 .08 -.15 .26 .52 Academic achievement (grades) .01 .25 -52(*) .11 -.01 .35 .10 .48 Plant or botanical sciences education -.01 .02 .05 .72 n -.08 .17 .18 .59 Conducted experiments and produced results .15 .36 -.03 .67 O .14 .02 -.12 .64 Reviews up-to-date scientific data and results .21 .18 .09 ss n .30 .12 -.13 .51 Synthesize integrate .29 .04 .31 .43 (•) .02 .04 -.06 .37 General research background .07 .08 .05 -.06 -son .03 .08 .67 Active involvement in scientific field .14 .13 .19 .36 •65 n -.02 .04 .62 Basic academic preparation -.07 .02 .17 .15 .01 .77(*) -.05 .65 Advanced academic preparation .16 .51 -.11 .13 .11 56 (•) .19 .67 Minimal career experience .02 -.05 .0 -.04 .04 .04 .79 O .63 Considerable career experience -.10 .46 .38 .04 .24 -.01 .48C) .66 Eigenvalue 3.27 2.06 2.05 1.94 1.60 1.39 1.33 13.64 Percent variance explained 14% 9 % 9 % 8 % 7 % 6 % 6 % 5 9 % 1 Values greater than R M S 0.29 have been flagged by an (*). ... Continued TABLE 5-17 (Cont'd) Results of Stage II Factor Analysis (b) Mean Factor Construct Scores and Differences Between Groups Group (N) Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Total sample (238) 3.65 3.26 3.68 3.94 2.33 3.83 3.85 Plant scientists (112) 3.34 Others (126) 3.14 Less experience (69) 3.90 Others (169) 3.70 Location: Canada (56) 2.96 3.47 U S A (164) 3.32 3.76 Others (10) 3.33 3.76 Researchers (180) 3.33 2.29 Others (58) 2.91 2.61 Administrators (55) 3.47 Others (183) 3.17 Teachers (45) 3.64 3.87 4.04 4.11 Others (193) 3.14 3.65 3.92 3.77 University employees (89) 3.55 3.86 4.04 Others (149) 3.05 3.59 3.89 Consultants (54) 2.48 Others (184) 2.32 Government employees (98) 2.20 Others (140) 2.48 Experts (25) 2.01 Others (213) 2.40 The mean score for this attribute is 3.65, indicating that normative expertise is a moderately to very important attribute of the expert. There was general con-sensus between groups regarding the importance of this attribute. Factor 2: Number of Publications Only two factors loaded highly on this construct, both involving publications. While the mean factor score is 3.26, there were several differences in opinion between groups regarding the importance of this dimension. Plant scientists (3.34), administrators (3.47), teachers (3.64) and university employees (3.55) considered it to be more important. Canadians (2.96) and non-researchers (2.91) considered it less important. Factor 3: External Credibility Four variables loaded highly on this construct: (i) career orientation, (ii) intellectual honesty, (iii) interpersonal relations, and (iv) academic achievement. This dimension encompasses various personal or professional qualities or characteristics that are related to the credibility of experts as viewed by others. The mean score for this dimension is 3.68, with minor differences between groups. Canadians considered it to be less important (3.47), while teachers (3.87) and university employees (3.86) considered it more important. 160 Factor 4: Substantive Expertise Four variables that loaded on this construct include: (i) plant or botanical sciences education, (ii) conducted experiments and produced results, (iii) reviews up-to-date scientific data and results, and (iv) can synthesize and integrate. This dimension consists of various skills that are derived through a combination of education and career experience. This dimension has the highest importance score (3.94) of the six Stage II factors, reflecting its overall importance. There is general consensus between groups regarding this opinion, with teachers (4.04) and university employees (4.04) rating it slightly more important. Factor 5: General Networking and Communication Two Stage I factors loaded highly on this construct. These include: (i) possession of a general research background, and (ii) active involvement and communication in the scientific field. This construct was difficult to interpret since the two Stage I factors combined have no obvious meaning. It carries a relatively low mean factor score (2.33), indicating it is not particularly important. Non-researchers (2.61) and consultants (2.48) considered this dimension more important, while government employees (2.20) and experts (2.01) considered it less important. 161 Factor 6: Level of Education This construct includes: (i) basic academic preparation, and (ii) advanced academic preparation. It was regarded as being very important (3.83) by most groups; teachers considered it more important (4.11). Factor 7: Length of Career Experience Length of experience is very important (3.85) overall. Individuals with less experience considered it more important (3.90). 5.3.4 Construct Validity of the Questionnaire The second objective of the factor analysis was to provide evidence of the construct validity of the questionnaire. The factor analysis was evaluated in order to determine whether the factors constructed represented theoretically realistic underlying attributes. For example, the original variables "Ph.D. degree" and "post-doctorate" were actually indicators of the same theoretical construct "advanced academic preparation", while the variables "B.S. degree" and "M.S. degree" were found to represent "basic academic preparation". The Stage II factor analysis showed that "advanced academic preparation" and "basic academic preparation" were highly correlated and both represented the underlying attribute "level of education". While these results are not particularly interesting, they do provide evidence that the questionnaire is measuring what it was supposed to be measuring; i.e., it provides evidence of construct validity. 162 Further evidence is provided by the fact that all Stage I and Stage II factors were easily interpretable and represented sound theoretical constructs. 5.4 Comparison of Expert Profile with Experts Nominated by Peers 5.4.1 Purpose and Method To test the validity of the expert profile, a regression model was estimated relating expert nominations to the various dimensions of expertise. Number of nominations for individual experts was the dependent variable, and independent variables included several attributes of expertise, shown in Table 5-18. The independent variable information was obtained from persons who completed a questionnaire—thus providing the necessary biographical information—and who were also nominated as experts. Number of publications and number of times cited in the literature were obtained through a library computer search. The independent variables given in Table 5-18 were intended to be surrogates of the following four Stage II expert factor constructs: Factor 2: Number of publications Factor 4: Substantive expertise Factor 6: Education Factor 7: Length of career experience 163 TABLE 5-18 Independent Variables Used in Regression Analysis Education (degree): 1 = Ph.D., 0 = otherwise Discipline: 1 = Plant science or Botany, 0 = otherwise Experience: 1 = 5 yrs to 10 yrs, 0 = otherwise Experience: 1 = 11 yrs to 20 yrs, 0 = otherwise Experience: 1 > 20 yrs, 0 = otherwise Present location: 1 = Canada, 0 = U.S.A. Research job: 1 = yes, 0 = no Consulting job: 1 = yes, 0 = no Policy making job: 1 = yes, 0 = no Enforcement job: 1 = yes, 0 = no Administrative job: 1 = yes, 0 = no Teaching job: 1 = yes, 0 = no University employee: 1 = yes, 0 = no Government employee: 1 = yes, 0 = no Consulting employee: 1 = yes, 0 = no Number of publications (continuous variable) 164 Not represented are Factor 1 (normative expertise), Factor 3 (external credibility) and Factor 5 (general involvement and communication). These constructs encompass elusive and difficult-to-measure attributes. The sample consisted of 58 persons who were nominated as experts in the survey, and for whom the necessary respondent information was available (Table 5-19). 5.4.2 Results An examination of the dependent variable data, number of nominations, indicated that the data was not normally distributed. For this reason the dependent variable data was transformed to logarithmic units prior to regression analysis. The predicted values of the dependent variable from the model must be transformed back to number of nominations by taking the anti-logarithm of the predicted values. The results (Table 5-20) provide evidence of the validity of the expert profile. An examination of the independent variable data showed that primary attributes explaining number of nominations included having a Ph.D. degree, ten to twenty years experience conducting research, working at a university, and publishing frequently. Experts are not likely to have a policy-making, teaching or consulting job and are not likely to be employed by government. The adjusted coefficient of determination (R2) for the model is 0.565. The correlation matrix and collinearity diagnostics are given in Table 5-21. To test for collinearity, condition numbers were calculated, as recommended by Belsley, Kah and Welsh (1980). The condition indices are the square roots of the ratio of the largest eigenvalue to each individual eigenvalue (SAS User's Guide: Statistics, 1986). The highest obtained value 165 TABLE 5-19 Nominated Individuals in Expert Attributes Model1 (in alphabetical order) Dr. Brian Amiro Dr. Keith Jensen Pinawa, Manitoba Delaware, OH. Dr. Robert Amundson Mr. J. Johnston Ithaca, NY. Oak Ridge, TN. Dr. Ann Bartuska Dr. Herb Jones Raleigh, NC. Muscle Shoals, AL. Dr. D.W. Beckerson Dr. G. Kats Guelph, Ontario Riverside, CA. Dr. James Bennett Dr. Ron Kickert Denver, CO Corvallis, OR. Dr. Jeff Brandt Dr. Robert Kohut Washington, DC. Ithaca, NY. Dr. Eileen Brennan Dr. Sagar Krupa New Brunswick, NJ. St. Paul, MN. Dr. Boris Chevone Dr. Lance Kress Blacksburg, VA. Research Triangle Park, NC. Dr. Lance Evans Dr. John Laurence Bronx, NY. Ithaca, NY. Dr. Allen Heagle Dr. Allen Lefohn Raleigh, NC. Helena, MT. Dr. Walter Heck Dr. Allan Legge Raleigh, NC. Calgary, Alberta Dr. Howard Heggestad Dr. Ida Leone Silver Spring, MD. New Brunswick, NJ. Dr. Gerit Hofstra Dr. Sam Linzon Guelph, Ontario Toronto, Ontario Dr. Donald Holt Dr. William Manning Urbana, IL. Amherst, MA. Dr. Patricia Irving Dr. Beverly Marie Washington, DC. Guelph, Ontario Dr. Jay Jacobson Ithaca, NY. ... Continued 166 TABLE 5-19 (Cont'd.) Nominated Individuals Used in Expert Attributes Model1 (in alphabetical order) Dr. Delbert McCune Ithaca, NY. Dr. Sandy McLaughlin Oak Ridge, TN. Dr. Joe Miller Raleigh, NC. Mr. Paul Moskovitz Upton, NY. Dr. Robert Musselman Riverside, CA. Dr. R. Noble Bowling Green, OH. Dr. David Olszyk Riverside, CA. Dr. Douglas Ormrod Guelph, Ontario Mr. Ronald Oshima Sacramento, C A Mr. Ron Pearson Toronto, Ontario Dr. Eva Pell University Park, PA Dr. Peter Reich Madison, WI. Dr. David Reid Calgary, Alberta Dr. Richard Reinert Raleigh, NC. Dr. Vic Runeckles Vancouver, B.C. Dr. John Seiler Blacksburg, V A Dr. David Shriner Oak Ridge, TN. Dr. John Skelly University Park, PA Dr. Cliff Taylor Riverside, C A Dr. George Taylor, Jr. Oak Ridge, TN. Dr. Pat Temple Riverside, C A Mr. Robert Teso Riverside, C A Dr. Ray Thompson Riverside, C A Dr. David Tingey Corvallis, OR. Dr. Craig Weidensaul Woosfor, OH. Dr. Leonard Weinstein Ithaca, NY. Dr. William Winner Blacksburg, V A Addresses and education were current at time of survey. 167 TABLE 5-20 Results of Regression Analysis Predicting Logarithms of Number of Nominations (N = 58) Independent Variable Parameter t-statistic Probability Estimate Constant -2.231 -4.03 0.001 Education (1 = Ph.D., 0 = otherwise) 0.627 2.04 0.041 Discipline (1 = Plant science or Botany, 0 = otherwise) 0.0663 0.12 0.900 Experience (1 = 5yrsto 10 yrs, 0 = otherwise) 0.283 0.53 0.593 Experience (1 = 11 yrs to 20 yrs, 0 = otherwise) 0.579 1.80 0.072 Experience (1 > 20 yrs, 0 = otherwise) 0.280 0.75 0.453 Present Location (1 = Canada, 0 = U.S.A.) -0.292 -0.72 0.475 Research job (1 = yes, 0 = no) 1.452 4.70 0.000 Consulting job (1 = yes, 0 = no) -0.452 -1.78 0.074 Policy making job (1 = yes, 0 = no) 0.553 2.03 0.043 Enforcement job (1 = yes, 0 = no) 0.406 0.65 0.513 Administrative job (1 = yes, 0 = no) -0.203 -0.80 0.423 Teaching job (1 = yes, 0 = no) -0.794 -3.22 0.001 University employee (1 = yes, 0 = no) 1.168 3.27 0.001 Government employee (1 = yes, 0 = no) 0.782 2.31 0.021 Consulting employee (1 = yes, 0 = no) 1.562 2.27 0.002 Number of publications 0.0204 5.95 0.000 168 TABLE 5-21 Analysis of Collinearity in Regression Model Variable Eigenvalue Condition Number Education 7.6957 1.0 Discipline 1.5502 2.2 Experience (5 - 10 yrs) 1.2463 2.5 Experience (11 - 20 yrs) 1.1065 2.6 Experience (> 20 yrs) 1.0543 2.7 Location 0.8425 3.0 Research job 0.6510 3.4 Consulting job 0.5146 3.9 Policy job 0.4328 4.2 Enforcement job 0.3257 4.9 Administrative job 0.2367 5.7 Teaching job 0.1244 7.9 University employee 0.0915 9.2 Government employee 0.0622 11.1 Consulting employee 0.0431 13.4 Publications 0.0226 18.5 169 of 18.46 is above the value of 15 suggested by Belsley et al. (1980) as a threshold before further consideration of collinearity is required. An examination of the correlation matrix showed there is some correlation between independent variables, but that correlation coefficients are quite low. For example, only three r-values in the 16 x 16 matrix exceed 0.40 and all are less than 0.49 (consulting employee -discipline; number of publications - experience greater than 20 years; and number of publications - teaching job). On this basis it was determined that collinearity was not a problem in this model. The prediction ability of this model is quite good, considering that two important dimensions of expertise, normative expertise and external credibility, were not represented by the independent variables. Neither was a third less important dimension, related to general involvement and communication. It is thus concluded that the expert profile developed from the survey portrays attributes of the individuals who were identified as experts by their peers. This provides evidence of the validity of the expert profile. 5.5 Predicting the Probability of Being Nominated as an Expert Logit models (Maddala, 1983) were estimated which predict the probability of individual experts receiving at least 13, or at least 20, nominations. For these models: y = 1 if the individual receives at least 13 or 20 nominations, respectively, and y = 0 if the individual receives less than the above number of nominations. 170 The parameter estimates and performance of the two logit models are summarized in Table 5-22. It was assumed that an individual would receive the critical number of nominations if the estimated probability from the logit models was 0.50 or higher. The models correctly predicted 91% to 98% of the observations (i.e., whether specific individuals would receive at least 13 or at least 20 nominations). However, these statistics are somewhat misleading since several individuals within the sample received considerably less than the critical number of nominations. The logit models correctly predicted that these individuals would not be selected as experts. A more useful evaluation of the performance of the models is to determine how well they predict the probability of nomination of those individuals actually receiving the critical number of nominations. This comparison is summarized in Table 5-23. Actual nominations received are compared with the logit's predicted probability of nomination. The "top" experts were correctly predicted to receive the critical number of nominations 71% to 73% of the time. However, due to the limited data base and the lack of independent variables representing normative expertise and external credibility, the logit models were not able to correctly predict nomination of all experts. Thus, these logit models may be used to assist in the selection of experts, but other criteria should also be employed (representing normative expertise and external credibility). For this purpose, peer judgment is required, which means that nomination by peers is the preferred method for selecting experts. 171 TABLE 5-22 Logit Model Parameters and Results for Predicting the Probability of Receiving a Specified Number of Expert Nominations > 13 Nominations > 20 Nominations Independent Parameter t-statistic Probability Parameter t-statistic Probability Variable Estimate Estimate Constant -90.067 -0.00 0.993 -54.807 -0.00 0.996 Education 10.443 0.00 0.997 10.394 0.00 0.997 Discipline 42.817 0.01 0.994 9.115 0.00 0.998 Experience (5-10 yrs) 5.422 0.00 0.999 -0.784 0.00 0.999 Experience (11-20 yrs) 15.936 0.00 0.997 10.519 0.00 0.998 Experience (> 20 yrs) 17.389 0.00 0.997 11.092 0.00 0.998 Present Location -2.129 -1.30 0.194 0.951 0.33 0.745 Research job 11.494 0.00 0.998 13.817 0.00 0.998 Consulting job -4.697 -1.79 0.074 -6.972 -1.28 0.202 Policy making job 0.324 0.18 0.857 2.178 0.52 0.600 Enforcement job -5.081 -0.00 0.999 -2.670 -0.00 0.999 Administrative job 0.0567 0.03 0.974 2.854 0.67 0.503 Teaching job -1.636 -0.98 0.326 -2.051 -1.01 0.313 University employee 8.048 2.31 0.021 3.8334 0.74 0.458 Government employee 5.140 2.03 0.042 1.211 0.468 0.640 Consulting employee 35.725 0.00 0.993 -0.709 0.00 0.999 Number of publications 0.0426 1.82 0.069 0.0850 2.33 0.020 172 TABLE 5-23 Comparison of Nomination Probabilities from Logit Model to Actual Nominations Expert ID # Actual Number of Nominations Predicted Probability of Nomination > 13 Nominations > 20 Nominations 223 13 .08 — 7 13 <.01 -159 15 .95 — 117 17 .73 ~ 136 19 .61 ~ 155 19 .93 -226 19 1.0 — 309 19 .10 ~ 600 20 .62 <.01 235 23 .06 .03 308 26 .73 .61 225 31 .84 .94 113 36 .99 .89 316 52 .74 .94 116 82 .98 .96 Number correctly selected: 11/15 5/7 (73%) (71%) 173 CHAPTER SIX EXPERT JUDGMENTS OF OZONE EFFECTS ON FRASER VALLEY CROPS In this chapter, the rationale for the provision of expert judgments is discussed. This is followed by a discussion of how the experts were selected, identity of the experts, and their judgments regarding potential crop losses due to 0 3 pollution in the Fraser Valley. 6.1 Rationale for Use of Experts Judgments by experts are not intended to be a substitute for scientific research. The primary objectives are to generate risks (or probabilities or likelihoods) of crop loss based on existing scientific knowledge, as well as to explicitly characterize the magnitude of uncertainty that is associated with these estimates. As described in Chapter Two, the exposure-response model represents our best estimate of the relationship between various levels of 0 3 (the independent variable) and associated crop losses for individual crops. An exposure-response model is a simplified model of reality; it is not sufficiently comprehensive to provide for all the factors that can affect crop yield. Exposure-response models thus allow projections (predictions) of crop loss that are inevitably uncertain. Statistical methods may be used to characterize uncertainty due to random variability, measurement error and sampling error associated with the fitted model. However, the use of such a model assumes that the situation under which the prediction is made is identical to the situation under which the experiment was undertaken. 174 Even when making predictions from models that are directly representative (same cultivar, same agricultural region), the likelihood of this assumption holding is practically zero. This is especially the case with outdoor field experiments where there will always be differences in uncontrolled background variables that are known to confound plant response to 0 3. As von Winterfeldt and Edwards (1986) stated, "the most intellectually demanding problem of collecting statistics is to find a way about their topic that permits the world that generates them to look stationary." Where indirect exposure-response information is used (i.e., models which are only indirectly representative because they involve different cultivars or species, or a different agricultural region), the major sources of uncertainty are variation in the way different species respond to 0 3, and/or the extent to which local environmental conditions affect crop yield. Persons in the best position to judge the usefulness and applicability of indirect exposure-response information are those who have considerable experience and knowledge regarding the topic (i.e., the experts). When directly representative data are missing and when time and research resources are limited, experts are frequently asked to provide risk estimates based on the available evidence. Probabilistic exposure-response relationships based on expert judgments have been developed for effects of various air pollutants on human health (Keeney et al., 1984; Whitfield and Wallsten, 1984; Morgan et al., 1985; Amaral, 1988), and for the effects of acid deposition on fisheries (Peterson and Violette, 1985), forests and aquatic resources (North et al., 1985). 175 6.2 The Concept of Subjective Probability Probabilities are numbers between 0 and 1 attached to propositions that are uncertain. A probability of 0 indicates that an event will definitely not occur; a probability of 1 means an event is certain to occur. There are two major schools of philosophical thought associated with the concept of probability (von Winterfeldt and Edwards, 1986). The frequentist school relates to objective probabilities, compiled from observations of repetitive events. Subjectivists, on the other hand, maintain that there are no truly repetitive situations and, for this reason, assert that probabilities are characteristics of an individual's state of mind. The subjectivist interpretation recognizes explicitly that use of objective evidence such as empirical data requires subjective judgments, reflecting the knowledge of the individual using them. The use of subjective probabilities is a natural and intuitive way to communicate when outcomes are not known with certainty. For example, a physician might state the probability of a full recovery following an operation is 0.8 (or alternatively, the chance of full recovery is 80%, or the odds are 4:1). Because subjective probabilities are opinions, they cannot be wrong (unless they are 0 or 1); there is always the acknowledgment that the unknown event may, or may not, occur. When forming a judgment, an expert performs an informed subjective weighting of all the relevant scientific data and knowledge that he or she is aware of at the time. The subjective probability for the same event may be quite different for different people, particularly if they possess different information. For this reason, and ideally, the range of responsible opinion associated with different experts should be sampled. 176 6.3 Risk Assessment Methods Methods for eliciting probabilities from experts have been developed primarily within the area of decision analysis, for business and policy applications. The methods were initially developed by Spetzler and von Holstein (1975), but have recently been applied to environmental assessments (e.g., Morgan et al., 1979; Feagans and Biller, 1981; Ruckelhaus, 1983; Morgan et al., 1984; Fraser et al., 1985; Morgan et al., 1985; North et al., 1985; Peterson and Violette, 1985; Amaral, 1988). Expert judgments regarding the probabilities of crop loss levels may be modelled by eliciting crop loss estimates at lower (.05), median (.50) and upper (.95) fractiles (von Winterfeldt and Edwards, 1986). The lower (.05 fractile) crop loss estimate is provided such that the probability that the true value falls below the estimate is .05 (or 5%). The "best" estimate corresponds to the .50 fractile, or median of the distribution. There is an equal (50:50) chance that the true loss is higher or lower than this best estimate. The upper (.95 fractile) loss estimate is selected such that there is only a 5% chance that the true crop loss value exceeds the estimate. These fractiles correspond to three points on the expert's subjective probability distribution. If the expert is relatively confident of the exposure-response relationship, the .05 and .95 fractiles will be relatively close to the median. If the expert is very uncertain about the exposure-response relationship, the .05 and .95 fractiles will be widely separated from the median. 177 Protocols that encode probability in the environmental risk assessments previously cited varied significantly in procedure and sophistication. In general, the elicitors emphasized the need to ensure that the experts carefully review relevant information and data prior to encoding, and that they understand the exact definitions of response and air pollution measures. Probability encoding generally involves personal interviews with no feedback of the opinions of others, thus preserving independence of opinion. 6.4 Selection of Experts 6.4.1 Selection of Experts in Previous Studies Helmer (1983) suggested that selection of experts should initially be made on the basis of past performance in the area being investigated; that is, an expert's past judgments should have been shown to have been reliable and accurate (i.e., well calibrated). Secondly, expert choice should be made on the basis of personal qualifications and achievements, such as education, experience, publications, and status among peers. Regarding Helmer's first criterion, an a posteriori or calibration type of evaluation generally cannot be conducted because actual empirical data are generally not available to compare with judgmental data. (If empirical data are available, there is generally no reason to obtain judgmental data). Thus, for initial expert selection, we are generally restricted to the use of a priori type criteria related to qualifications and achievements. Once again, however, it is not clear how to proceed since the relationship between measures of these attributes and the 178 ability to make reliable, accurate predictions does not appear to have been systematically investigated and reported. In recent environmental risk assessments, a common means of identifying experts has been to obtain a list of recommended individuals from an appropriate agency or establishment. For example, health experts recommended by the U.S. Environmental Protection Agency were used by Keeney et al. (1984), Whitfield and Wallsten (1984) and, it would appear, Morgan et al. (1985). This expert selection method is, of course, susceptible to the bias of the particular agency and/or individual within the agency. A more objective approach was used by researchers in two recent environmental risk assessments. Peterson and Violette (1985) assessed the potential effects of acid deposition on fisheries in the Adirondacks. They compiled a list of authors from recent relevant scientific literature. Weighting of "expertise" was based on the individual's number of recent publications and the amount of work performed in the Adirondack Mountain region. In an assessment of the impact of air pollution on Canadian forests, Fraser et al. (1985) selected experts using a nomination procedure. The experts used were those nominated the most number of times by their peers. This is the method that was used to select experts in the present research project, as described in Section 6.4.2. Another question related to the selection of experts concerns how many experts should be contained in the sample. Work by Ashton (1986) showed that only five individuals were needed to achieve much of the improvement available from combining the forecasts of up to 13 individuals. In a risk assessment of health effects from air pollution by Morgan et al. (1985), the results showed that only a few atmospheric scientists' predictions were required due to the 179 consensus of opinion of this group. One or two scientists predicted nearly the same probabilities as the aggregate opinion of the larger group. On the other hand, due to disagreement among health experts, a considerably larger sample would be required to determine the range of opinion of this group. 6.4.2 Selection of Experts in Present Research As described in Chapter Five, experts were identified in the present research through nomination by their peers in the expert survey. The majority of questionnaire respondents nominated up to five experts, based on the following instructions: Please list the names and affiliations (if known) of up to five persons who in your opinion qualify as top experts in 0 3 impacts on crops. In your opinion these persons should be the most qualified experts you are aware of, capable of providing risk estimates (judgments) of probable yield reductions of crops in the Fraser Valley of British Columbia, resulting from observed 0 3 levels. A total of 106 nominees were provided by 166 of the respondents, and a total of 648 nominations were received. The maximum number of nominations proposed for a nominee was 82, the minimum was one and the average was 6.1s. A minimum of five experts was considered necessary for the risk assessment project, based on work by Ashton (1986). Analysis of the data showed that 13 individuals were clearly perceived as highly qualified experts, based on the high number of nominations received by The number of nominations received by each nominee is considered confidential. 180 each of these individuals. Of these, Dr. V.C. Runeckles was not considered eligible due to his direct involvement with the project. The remaining twelve individuals were invited to participate as experts in the risk assessment project. A letter was sent to each expert describing the project and requesting their participation. Nine experts agreed to participate and to be identified; they are listed in Table 6-1. 6.5 Design of Expert Risk Assessment Questionnaire As indicated previously, the majority of environmental assessments involving expert judgments used a personal interview method to elicit judgments from experts. In the present survey, it was decided to design and distribute a standardized risk assessment questionnaire, for . the following reasons: (i) A questionnaire can be mailed out and is thus less expensive than personal interviews. (ii) Experts are not pressured to provide data on the spot with a questionnaire; they can take the necessary time to review background information before providing judgments, if they wish. (iii) It ensures that all experts receive identical questions. The risk assessment questionnaire, reproduced in Appendix B, contains information about seven important Fraser Valley crops, including: green snap bean, potato, processing pea, corn, broccoli, raspberry and forage. Also included are summary ambient 0 3 data based on actual measurements from ambient air quality monitors in the Fraser Valley. For this purpose, 181 TABLE 6-1 Experts Selected for Crop Loss Risk Assessment1 (in alphabetical order) Dr. Allen S. Heagle Air Quality Research Programs North Carolina State University 1509 Varsity Drive Raleigh, NC. 27606 Dr. Walter W. Heck United States Dept. of Agriculture North Carolina State University 1509 Varsity Drive Raleigh, NC. 27606 Dr. Howard E. Heggestad 3112 Castleleigh Rd. Silver Spring, MD. 20904 (Formerly of USDA Beltsville; retired) Dr. Jay S. Jacobson Boyce Thompson Institute Cornell University Tower Road Ithaca, NY. 14853 Dr. Sagar V. Krupa University of Minnesota Department of Plant Pathology 495 Borlang Hall St. Paul, MN. 55108 Dr. Douglas P. Ormrod University of Guelph Dept. of Horticultural Science Guelph, Ontario NIG 2W1 Dr. O. Cliff Taylor SAPRC University of California Riverside, CA. 92521 Dr. Patrick J. Temple SAPRC University of California Riverside, CA. 92521 Dr. David T. Tingey U.S. EPA Corvallis Environmental Research Lab 200 S.W. 35 Street Corvallis, OR. 97333 Addresses were current at time of survey (1988). 182 raw data collected at Greater Vancouver Regional District (G.V.R.D.) and B.C. Ministry of Environment air quality monitoring stations between the years 1978 and 1986 were analyzed. Several exposure indices relevant to crop loss assessment were computed for each station and each site for the four-month growing season starting May 1. Three scenarios were then selected representing average or typical conditions, extreme low conditions and extreme high 0 3 conditions. The "low" 0 3 scenario was based on data collected at Chilliwack in 1986. The "medium" 0 3 scenario was based on data from Abbotsford in 1985, and the "high" scenario was based on data from Anmore in 1981 (see Figure 3-2 for the map of these locations). Both Chilliwack and Abbotsford are located in important agricultural areas of the Fraser Valley. Anmore is not located in an agricultural area, but was selected because air quality is typically quite poor in this area. Details of the exposure indices for the three pollution scenarios are contained in the Risk Assessment questionnaire in Appendix B. It would have been desirable to administer the questionnaire twice, at two different times, to determine if responses were consistent and thus reliable. It was decided not to request this because of the precious extra time that experts would have to spend redoing the questionnaire. 6.6 Results of Expert Risk Assessment Survey 6.6.1 Expert Biographical Information All nine experts have Ph.D. degrees in Plant Science or equivalent (Table 6-2). One individual had less than ten years experience, four had 21 to 30 years of experience, and two 183 TABLE 6-2 Expert Respondent Information Variable Frequency (%) Highest degree obtained Ph.D. 9 100 9 Years Career Experience (since graduation) Less than 10 1 11 10 to 20 2 22 21 to 30 4 45 31 to 40 2 22 41 to 50 0 _0 9 100 Type of Research Experience 1 Research 9 100 Experiments 9 100 Administration 4 45 Teaching 5 56 Consulting 5 56 Employer 1 University 6 67 Government 4 45 Consulting 1 11 Industry 0 0 Number of Publications Greater than 40 9 100 9 100 1 Some individuals fit into more than one category 184 had more than 30 years of experience. All nine experts conducted experimental research in 0 3 effects in crops. In addition, four had administrative experience, five had teaching experience, and five had done consulting work. Five of the experts were employed by a university, three by government and one had a joint university-government appointment. One individual was also working as a consultant. All nine experts had published in excess of 40 scientific papers. 6.6.2 Relative Sensitivity of Crops Aggregate expert judgment regarding the relative sensitivity of the seven crops to 0 3 is summarized in Table 6-3. On average, green bean was considered to be the most sensitive of the seven crops. Potato and pea were considered to be moderately sensitive, and the remaining crops were considered to be only slightly sensitive. Consensus is greatest regarding the relative sensitivity of green snap bean (v = 0.19), and lowest regarding the sensitivity of broccoli, raspberry and forage (v = 0.54, 0.50 and 0.47, respectively). 6.6.3 Mean Crop Loss Estimates The expert crop loss data are summarized in Tables 6-4, 6-5 and 6-6 for the low, medium and high 0 3 scenarios, respectively. Most experts predicted there would be yield loss for most crops under all three 0 3 scenarios. Predicted average loss for all crops under the low 0 3 scenario at the 0.05 fractile was 0.22%, at the 0.50 fractile was 1.11% and at the 0.95 fractile was 2.54% (Table 6-4). At the median, the predicted average losses were highest for 185 TABLE 6-3 Expert Judgments of Relative Sensitivity of Crops to 03' Crop Mean 00 Standard Deviation (s) Coefficient of Variation2 (v) Minimum Value Maximum Value Number of Experts Green snap bean 4.11 0.78 0.19 3.0 5.0 9 Potato 3.25 1.28 0.39 2.0 5.0 8 Pea 3.00 1.12 0.37 1.0 4.0 9 Corn 1.88 0.83 0.44 1.0 3.0 8 Broccoli 2.22 1.20 0.54 1.0 4.0 9 Raspberry 2.00 1.00 0.50 1.0 3.0 7 Forage 2.00 0.93 0.47 1.0 3.0 8 Based on the following scale: 5 = extremely sensitive 4 = very sensitive 3 = moderately sensitive 2 = slightly sensitive 1 = not sensitive at all 2 Coefficient of variation (v) = s/x TABLE 6-4 Expert Aggregate Crop Loss Estimates: Low 0 3 Scenario Crop Loss (%)  Number of Standard Minimum Maximum Crop and Fractile Experts Mean Deviation Value Value Green bean .05 8 0.63 1.19 0.0 3.0 .50 8 2.63 2.97 0.0 8.0 .95 8 4.63 4.10 0.0 10.0 Potato .05 9 0.11 0.33 0.0 1.0 .50 9 1.11 1.45 0.0 4.0 .95 9 2.22 2.68 0.0 8.0 Pea .05 8 0.13 0.35 0.0 1.0 .50 8 0.88 1.73 0.0 5.0 .95 8 3.00 3.66 0.0 10.0 Corn .05 8 0.00 0.00 0.0 0.0 .50 8 0.38 0.52 0.0 1.0 .95 8 1.13 1.13 0.0 3.0 Broccoli .05 8 0.25 0.71 0.0 2.0 .50 8 1.00 1.78 0.0 5.0 .95 8 2.38 2.88 0.0 7.0 Raspberry .05 7 0.29 0.76 0.0 2.0 .50 7 1.29 2.21 0.0 6.0 .95 7 3.14 3.72 0.0 10.0 Forage .05 8 0.13 0.35 0.0 1.0 .50 8 0.50 1.07 0.0 3.0 .95 8 1.25 1.75 0.0 5.0 Average (All Crops) .05 - 0.22 0.52 0.00 0.63 .50 - 1.11 1.68 0.50 2.63 .95 - 2.54 2.85 1.13 4.63 187 TABLE 6-5 Expert Aggregate Crop Loss Estimates: Medium 0 3 Scenario Crop Loss (%) Number of Standard Minimum Maximum Crop and Fractile Experts Mean Deviation Value Value Green bean .05 8 2.00 2.27 0.0 5.0 .50 8 5.50 4.54 0.0 12.0 .95 8 9.63 5.37 0.0 15.0 Potato .05 9 0.89 1.05 0.0 3.0 .50 9 3.22 2.82 0.0 9.0 .95 9 7.33 4.56 0.0 16.0 Pea .05 8 0.38 0.74 0.0 2.0 .50 8 2.75 2.60 0.0 7.0 .95 8 5.13 4.22 0.0 10.0 Corn .05 8 0.75 1.04 0.0 3.0 .50 8 2.25 2.31 0.0 6.0 .95 8 4.88 4.36 0.0 11.0 Broccoli .05 8 0.50 1.07 0.0 3.0 .50 8 1.88 3.04 0.0 8.0 .95 8 4.25 4.80 0.0 10.0 Raspberry .05 7 0.71 1.11 0.0 3.0 .50 7 2.86 3.29 0.0 8.0 .95 7 5.71 5.74 0.0 13.0 Forage .05 8 0.63 1.06 0.0 3.0 .50 8 2.13 1.89 0.0 5.0 .95 8 4.38 2.83 0.0 8.0 Average (All Crops) .05 0.84 1.19 0.38 2.00 .50 - 2.96 2.93 1.88 5.50 .95 - 5.90 4.55 4.25 9.63 188 TABLE 6-6 Expert Aggregate Crop Loss Estimates: High 0 3 Scenario Crop Loss (%)  Number of Standard Minimum Maximum Crop and Fractile Experts Mean Deviation Value Value Green bean .05 9 4.78 4.35 0.0 10.0 .50 9 13.22 6.63 5.0 25.0 .95 9 23.00 11.22 10.0 40.0 Potato .05 9 3.11 3.62 0.0 10.0 .50 9 12.11 15.20 0.0 50.0 .95 9 21.89 27.37 0.0 90.0 Pea .05 7 1.33 1.50 0.0 4.0 .50 7 7.11 5.90 0.0 20.0 .95 7 14.22 12.33 0.0 40.0 Corn .05 9 1.44 1.51 0.0 4.0 .50 9 5.78 3.56 0.0 10.0 .95 9 11.00 8.20 1.0 25.0 Broccoli .05 8 1.13 1.89 0.0 5.0 .50 8 6.38 6.91 0.0 20.0 .95 8 13.25 12.98 0.0 40.0 Raspberry .05 7 1.43 1.99 0.0 5.0 .50 7 4.43 4.35 0.0 10.0 .95 7 11.71 12.27 0.0 35.0 Forage .05 8 1.13 1.64 0.0 4.0 .50 8 4.75 2.43 2.0 8.0 .95 8 10.63 4.78 5.0 20.0 Average (All Crops) .05 - 2.05 2.36 1.13 4.7£ .50 - 7.68 6.43 4.43 13.22 .95 - 15.10 12.74 10.63 23.0C 189 green bean (2.63%), raspberry (1.29%), potato (1.11%) and broccoli (1.0%). Predicted loss levels at the median were less than 1.0% for pea, corn and forage. A wider range of losses was predicted under the medium 0 3 scenario for all crop types (Table 6-5). Average predicted loss at the 0.05 fractile was 0.84%, at the 0.50 fractile 2.96% and at the 0.95 fractile 5.90%. At the median, predicted average losses were highest for green bean (5.50%) and potato (3.22%). Losses of 2.0% to 3.0% were predicted for pea, corn, raspberry and forage. Lowest loss at the median was predicted for broccoli (1.88%). Under the high 0 3 scenario (Table 6-6), average losses for all crops were predicted to be 2.05% at the 0.05 fractile, 7.68% at the 0.50 fractile and 15.10% at the 0.95 fractile. Average losses predicted at the median were highest for green bean (13.22%) and potato (12.11%), followed by pea (7.11%), broccoli (6.38%), corn (5.78%), forage (4.75%) and raspberry (4.43%). The degree of uncertainty associated with the crop loss estimates may be judged by the relative consensus between experts, represented by the standard deviations of the aggregate crop losses given in Tables 6-4, 6-5 and 6-6. Relative degree of consensus was highest under the low 0 3 scenario, less under the medium 0 3 scenario and lowest under the high 0 3 scenario. Within the low 0 3 scenario, less consensus among experts existed in the case of green bean and raspberry. This was also the case for the medium 0 3 scenario. Under the high 0 3 scenario, uncertainty was greatest regarding potato losses. 6.7 Individual and Aggregate Crop Loss Models The means and variances for each expert's cumulative distribution functions (cdf s), representing their overall judgments, were computed from the three fractiles using the 190 three-point approximations proposed by Pearson-Tukey (Keefer and Bodily, 1983), as follows: Mean = 0.63x (0.50) + 0.185 [ x(0.05) + x(0.95) ] Variance = ( [ x(0.95) - x(0.05) ] / 3.25 )2 where: x(p) is the p fractile of the random variable x The resulting individual as well as aggregate expert judgments for each crop type and under each 0 3 scenario are portrayed in Figures 6-1 to 6-7. Each of these figures corresponds to a particular crop type and contains four individual graphs. Three of the graphs in each figure portray individual expert crop loss models under the low, medium and high 0 3 scenarios, respectively. The fourth graph contains aggregate expert crop loss models under all three 0 3 scenarios. Additionally, in the fourth graph, unweighted and weighted aggregate models are portrayed. Aggregate unweighted opinion is the average predicted crop loss for all experts. Aggregate weighted opinion was computed by adjusting individual expert judgments on the basis of the number of nominations received by each expert, and then computing the average crop loss estimate. The result of this weighting has been, in many cases, to shift the loss curve to the right. This shows that individuals who were nominated most frequently generally predicted higher crop losses than their peers. Appendix C contains the computed parameter values (means and standard deviations) for the individual and aggregate crop loss models contained in Figures 6-1 through 6-7. Predicted crop losses in the Fraser Valley, based on the expert judgments described in this chapter, are contained in Chapter Seven. 191 Green Bean : Low Ozone Scenario Cumula t ive Probabi l i ty Percent of Crop Loss Green Bean : Medium Ozone Scenario Cu"iu>ot lv« Probob l l l ly o 4-- —i — - j 0 10 20 30 Percent of Crop Loss FIGURE B-1 CROP LOSS RISK ESTIMATES BY EXPERTS (SEE TEXT SECTION 6.7 FOR EXPLANATION) Green Bean : High Ozone Scenario Cumulat ive Probobl l l ly 0 | , p— ' •—I 0 10 20 30 Percent of Crop Loss Green Bean : Aggregate Opinion It'O'i'i'tfV H W i V V lyUfl'&UW C u m u l o l l . . P r .bob lMI , W I l C H U t t W {H M T » »HtHT» 0 I — 1 • 1 1 1 I 0 • 10 20 30 Percent of Crop Loss GREEN BEAN Pea : Low Ozone Scenario Pea : Medium Ozone Scenario Cumu'otlvt Proboblllty Percent Crop Loss FIGURE 6-2 CROP LOSS RISK ESTIMATES BY EXPERTS (SEE TEXT SECTION 6.7 FOR EXPLANATION) Pea : High Ozone Scenario Cumulollvt ProbobJItly Percent Crop Loss Pea : Aggregate Opinion UHWixju untt'svu utwaiiv CumulolU. f .bobl l l ty W d O H T C O . , WCICHTtD W{HHT» • 0 | , , , 1 , , , 1 0 . 10 20 50 Percent Crop Loss PEA Potato : Low Ozone Scenario Cumulative Probability 0.5 -! —i—>—'—•—•—1~ 10 20 Percent Crop Loss 30 Potato : High Ozone Scenario Cumutollvi Probability 0.5 — i — • — 1 — 1 — 1 i 10 20 Percent Crop Loss 30 Potato : Medium Ozone Scenario Cumu'ollvo Probability 0.5 -10 20 Percent Crop Loss 30 Potato : Aggregate Opinion nw'fi»u« utn'&vw itm'fitiv Cumulo.lv. Probability WCICHTCO WCICHTtO WMCHTCO 0.5 -10 20 Percent Crop Loss 30 FIGURE 6-3 CROP LOSS RISK ESTIMATES BY EXPERTS: POTATO (SEE TEXT SECTION 6.7 FOR EXPLANATION) Broccoli : Low Ozone Scenario Cunuigllvo Proboblllly 0 H — . — . — . — . — r — — . — • — . — r — • — • — • — • — 0 10 20 30 Percent Crop Loss Broccoli : Medium Ozone Scenario CumuiallM Proboblllty o -I—,—i ,—,—,—i—,—,—,—,——.—.—I 0 10 . 2 0 30 Percent Crop Loss FIGURE 6-4 CROP LOSS RISK ESTIMATES BY EXPERTS (SEE TEXT SECTION 6.7 FOR EXPLANATION) Broccoli : High Ozone Scenario 0 10 20 30 Percent Crop Loss Broccoli : Aggregate Opinion ILttt'&iUU U I M f i t W ItWX'&UU Cumuioii,. Probobiiiir W H 8 H T » , M I W H r , . wtiGHTto Percent Crop Loss BROCCOLI Raspberry : Low Ozone Scenario Percent Crop Loss Raspberry : Medium Ozone Scenario Percent Crop Loss FIGURE 6-5 CROP LOSS RISK ESTIMATES BY EXPERTS (SEE TEXT SECTION 6.7 FOR EXPLANATION) Raspberry : High Ozone Scenario Cumulollvo Probablllty Percent Crop Los Raspberry : Aggregate Opinion IL'U'£ ,£KJV Utyi'&UU UtPi'Stltf Cumuloll , . Probability w c i C H T t D wj^jS-. WliiBIUL, 0 I , i 1 1 1 1 I 0 10 20 30 Percent Crop Loss RASPBERRY Corn : Low Ozone Scenario Percent Crop Loss Corn : Medium Ozone Scenario Cumulative Proboblllty Percent Crop Loss FIGURE 6-6 CROP LOSS RISK ESTIMATES BY EXPERTS (SEE TEXT SECTION 6.7 FOR EXPLANATION) Corn : High Ozone Scenario Cumulotlvt Proboblllty Percent Crop Loss Corn : Aggregate Opinion HHM ,iMJW tt^l'SHIV aVUTfitJU Cumulollv. Probability W " C H T C D wctCMTCO W [ I Y H T » Percent Crop Lost CORN Forage : Low Ozone Scenario Cumulallvt Probability 0 -4- , r — r 0 10 20 Percent Crop Los* forage : Medium Ozone Scenario Cumulative Probability 0 -I—.—, , — i — , — — . — . — i ,—,—,— 0 10 20 30 Percent Crop Loss Forage : High Ozone Scenario Cumulative Probability 0 I I 0 10 20 30 Percent Crop Loss Forage : Aggregate Opinion lityi'initf itwivv UHMfiUtf Cumulatlv. Probability w C l t " T » - » C H " T [ p _ " . H U T U , Percent Crop Loss FIGURE 6-7 CROP LOSS RISK ESTIMATES BY EXPERTS: FORAGE (SEE TEXT SECTION 6.7 FOR EXPLANATION) CHAPTER SEVEN GENERAL DISCUSSION AND APPLICATION: PREDICTED CROP LOSSES IN THE FRASER VALLEY 7.1 Introduction Two kinds of judgment.are involved in environmental decision making: (i) cognitive judgment based on scientific information and analyses, and (ii) evaluative judgment, based on policy considerations (National Academy of Science, 1977). The identification and assessment of risks to ecosystems is the responsibility of science, but the determination of the acceptability of a risk is a task for the political process. Governments are often required to act in the absence of full knowledge about risks, and because limited resources do not allow detailed evaluation of every situation where risk may exist (Russell and Gruber, 1987). Indeed, as pointed out by Orians (1986), sufficient information to permit vigorous predictions of the consequences of most policy and management decisions will never be available. Substantial uncertainty exists regarding the potential impact of 0 3 on agricultural crops in the Fraser Valley of British Columbia. Very limited financial assistance has been available to support the experimental scientific research that is needed to generate exposure-response information. And even if sufficient research funds should become available, the generation and verification of data for the several economically-important Fraser Valley crops could take several years. The preferred method for improving decisions is by reducing uncertainty. There is no substitute for scientific research; the strength and integrity of modern science fundamentally rest on disciplined objectivity in an effort to decipher a world that is often full of apparent 199 contradictions. However, in the real world, decisions often have to be made before sufficient scientific information is available. When risk judgments depend on incomplete or indirect information, the policy maker is best served by being informed both of the estimates of the probabilities of adverse consequences, and of how firm those estimates are (Feagans and Biller, 1981). The firmness of the estimates depends on the completeness of the information base. For example, policy makers should recognize that ambient air quality standards, as they presently exist, do not allow the magnitude of impacts to crops to be accurately estimated in situations where air quality conditions are known to be elevated above "background" levels. The challenge is to communicate such information to policy makers in a manner that will be useful in formulating policies and making decisions. Chapter Two of this thesis described numerous sources of uncertainty associated with assessment of crop losses from air pollution. In summary, several sources of uncertainty presently contribute to the difficulty in assessing the effects of air pollutants on crops: (i) The available data base is small and thus is not fully representative of North America. (ii) Potential effects of different experimental methods have not been fully addressed. (iii) Exposure-response models are empirical and not based on mechanistic considerations. (iv) Effect of soil moisture on crop response to air pollutants is not well understood. (v) Effects of biotic stresses on crop response to air pollutants are not well understood and are not included in predictive models. (vi) The importance of exposure dynamics (i.e., peak 0 3 values) occurring throughout the growing season is not well understood. (vii) Ozone exposures used in experiments were frequently not realistic. 200 (viii) Exposure-response models used may be a product of the specific experimental design. (ix) The use of long-term seasonal averages as the basis of predictive models has serious flaws related to exposure dynamics and duration. (x) Potential synergistic effects between 0 3 and other common pollutants such as sulphur dioxide (S02) and nitrogen oxides (NO,) are not well understood. Ozone is just one constituent of photochemical smog, a complex mixture of gases that probably has • synergistic effects (Simpson, 1988). A prime objective of vegetation effects research is to provide sound quantitative evidence regarding the relationships between air quality and vegetation effects. A basic requirement for the understanding of such relationships is the determination of a threshold (or thresholds) below which negative effects do not occur, or are so minor as to be indisting-uishable from effects caused by other factors (Runeckles, 1987). For many air pollutants, there is some justification for this threshold assumption (Simpson, 1988). Several pollutants, including 0 3, arise naturally from nonanthropogenic sources, so adaption by the environment to these pollutants was already taking place before anthropogenic emissions began.6 This argument is the basis for the "natural standards" concept discussed by Fischoff et al. (1981), which states that "the ambient level of pollution that existed during the development of a species is the level to which the species is best suited, and the level to be sought when setting future tolerances." Fischoff et al. (1981) indicated that such an approach is insensitive to economic issues, and is thus politically unrealistic. Certain carcinogenic pollutants, such as polyaromatic hydrocarbons, are new to the environment, and for these chemicals no threshold is assumed. 201 Political attitudes toward environmental protection have recently changed, however, and the modern concept of "sustainable development" may mean that future air quality goals will be significantly more stringent than existing environmental standards, as society is made to "pay the costs of pollution" (World Commission on Environment and Development, 1987). Notwithstanding these political considerations that represent an additional (external) source of uncertainty, judgment is needed both in the establishment of air quality standards and in the prediction of crop losses based on the application of empirical exposure-response information. Although scientific research is improving our knowledge of the relationships between air pollution and vegetation response, a complete understanding of the inter-relationships is "a distant hope" (Runeckles, 1987); thus the need for "cognitive judgment". Evaluative judgment is also required, particularly in the case of formulation of air quality standards, because of the tradeoffs that are involved between risks and social and economic values. 7.2 Fraser Valley Crop Loss Assessment The complicated issues described in the previous section are unlikely to be resolved in the near future. When combined with the lack of empirical data regarding effects of 0 3 on Fraser Valley crops, it is clear that predictions of impact that utilize the existing information base will be highly uncertain. Additional information related to the potential impact of 0 3 on Fraser Valley crops was obtained through biomonitoring with tobacco Bel W-3 during three growing seasons, and through provision of crop loss estimates by experts. The value and limitations of these two 202 methods relate to the modes of Beanlands and Duinker's (1983) environmental decision making model, involving both empirical experimental research and the analysis of expert judgment (Figure 1-2). 7.2.1 Biomonitor Project The biomonitoring program was conducted during a three-year period, 1985 through 1987. Figure 7-la shows the monthly mean 0 3 concentration for all stations for the period of record (1978-1988) and shows that the maximum 0 3 concentrations are generally observed in the spring and early summer months. During the years 1985 through 1987, as shown in Figure 7-lb, there were no exceedances of the maximum tolerable 1-hour average 0 3 objective (0.15 ppm) throughout the entire Greater Vancouver Regional District (G.V.R.D.) air quality monitoring network. There were exceedances in most other years in the period from 1978 to 1988. Note that the ordinate in Figure 7-lb is logarithmic between the values of 1 and 1000. The frequency of exceedances of the maximum acceptable objective (0.08 ppm) was also low during the biomonitoring program, as shown in Figure 7-lb. The fewest exceedances of the maximum acceptable objective occurred in 1987. In 1988, there was a sharp increase in the number of exceedances. According to Concorde Scientific Corporation and B.H. Levelton and Associates Ltd. (1989), the improved air quality between 1982 and 1987 is not the result of reduced oxidant precursor emissions. Rather, it is likely due to prevailing climatological and meteorological conditions over this period. Evidence for this is provided by the sharp increase in the number 203 0.025 0.000 JAN FEB MAR APR MAY JUN JUL AUG SEP MONTH la) MONTHLY MEAN OZONE CONCENTRATION (FOR YEARS 1978 TO 1988) OCT NOV DEC 1000 in U J z < o U J — U J C/1 cj cn X ZD U J o ^ 4-a 2 o U J t-m < 2 < r-O 100 10 7B 79 w MAXIMUM ACCEPTABLE (0.08 ppmj * HOURLY OZONE OBJECTIVE A MAXIMUM TOLERABLE (0.15 ppm) HOURLY OZONE OBJECTIVE B5 B6 lb) NUMBER OF HOURLY OZONE EXCEEDANCES BY YEAR SOUHCE: CONCOROE SCIENTIFIC CORPORATION AND B.H. LEVELTON S ASSOCIATES LTD (1989) F IGURE 7-1 OZONE A IR QUAL ITY TRENDS IN THE LOWER MAINLAND OF B R I T I S H COLUMBIA (BASED ON AVERAGE OF 20 MONITORING STATIONS) 204 of exceedances in 1988. The implication is that future 0 3 conditions are unlikely to improve, unless there is a significant decrease in the magnitude of precursor emissions. In the present research, a relatively low correlation (r < 0.35, p < 0.05) was demon-strated between Fraser Valley 0 3 levels and tobacco injury response. While the strength of this relationship is lower than that observed by other researchers (e.g., Ashmore et al., 1978; Naveh et al., 1978; Manning and Feder, 1980; Horsman, 1981), it demonstrates the existence of phytotoxic 0 3 levels in the Fraser Valley. The relatively low observed correlation between 0 3 exposures and biomonitor injury is partially due to the atypically low 0 3 conditions during the monitoring program. In 1986 and 1987 in particular, 0 3 levels were frequently insufficient to induce plant injury. While 0 3 levels would have exhibited normal diurnal and seasonal variation, the threshold exposure for biomonitor injury was seldom exceeded. Indeed, it was only possible to detect a significant relationship between injury and 0 3 when the entire data set from the three years of monitoring was aggregated, and when data were included only where injury was observed. It is perhaps unfortunate that 0 3 levels were atypically low during the research period, which affected the potential usefulness of the results. Figure 7-lb provides evidence of the limitations associated with data that has been collected during a limited time period. It confirms the importance of the inherent stochasticity of the environment, and the difficulty in predicting impacts based on a limited data base. The limited data resulting from the biomonitor injury-crop yield calibration experiment provide some evidence for the existence of a potentially useful relationship in crop effects research. The concave relationship observed was the result of the lowest level of biomonitor injury being associated with substantial crop losses. This observation alone is potentially 205 important since tobacco Bel W-3, considered to be extremely sensitive to 0 3, has been frequently used as an "early warning system" of vegetation impacts. The present results indicate that chronic yield losses of agricultural crops may result under 0 3 concentrations that are too low to result in acute injury to the biomonitor plant. Although evidence of this phenomenon is admittedly limited, data collected by Wright (1988) showed that 0 3 episodes above 25 ppb can result in crop losses. The 0 3 exposure threshold for injury to tobacco Bel W-3, on the other hand, is known to be approximately 40 ppb for four hours (Ashmore et al., 1978). Additionally, research will be required to determine whether "respite periods" (Kohut et al., 1988) significantly influence the potential relationship between biomonitor foliar injury and crop yield losses. The results of the present biomonitor research have strengthened the conclusions of others, in particular, that: (i) Biomonitor plants can be used to provide information about potential vegetation impacts at locations where ambient air quality instruments do not exist. (ii) Biomonitors used in conjunction with ambient air quality monitors can provide information related to both air quality conditions and associated biological effects. In future research it would be desirable to "fine-tune" the relationship between biomonitor injury and crop losses. Until this is done, the absence of biomonitor injury does not provide evidence of the existence of some environmental condition protective of plants in general, as previously thought. 206 7.2.2 Crop Loss Risk Assessment by Experts An alternative method of providing information related to impacts of 0 3 on crops is through risk assessment by crop loss experts. Risk assessments by experts are not intended to substitute for much needed scientific research; rather, such assessments may be used to help summarize the existing state of scientific knowledge, and thus used to help set priorities for policy analysis and future research. This may be done through explicit communication, by scientists to policy makers, of existing scientific knowledge related to critical cause and effect relationships. It is important that policy makers make use of all available information, because their decisions frequently affect millions of people. While environmental risk assessment has been proposed by academics and bureaucrats as a vehicle for improving the communication of scientific information between scientists and policy makers (e.g., Ruckelhaus, 1983), various methodological and ethical problems are still associated with its use. One problem stems from the basic differences in opinion between the two major schools of philosophical thought regarding the concept of probability: the frequentists and subjectivists (von Winterfeldt and Edwards, 1986). A second problem regarding the use of experts and their judgments for policy formulation is that there is likely to be disagreement and misunderstanding about the credentials an expert should possess. The perception of "expertise" involves dimensions other than just calibration (i.e., the goodness of their predictions). Some external certification other than scientific knowledge, in particular credibility, is recognized as playing an important role. With the exception of relatively recent work by Shanteau and his colleagues at the University of Kansas, and certain checklists of expert witness attributes in the law literature, 207 there is a surprising lack of information regarding the necessary and desirable attributes of expertise. This not only applies to experts in 0 3 effects on crops — the attributes of expertise have not been well defined in any discipline. The reason for this may be that many highly qualified individuals in most professions feel that the credentials of an expert are fairly obvious. This in turn may lead to the view that it should be a relatively straightforward task to select a group of experts for risk assessment. The work by Shanteau, which essentially describes strategies used by experts to help them avoid cognitive limitations such as representativeness, anchoring and adjustment, availability, and overconfidence (Tversky and Kahneman, 1974), does not assist with an understanding of the desirable attributes of a crop loss expert. Hence, it does not make the selection of crop loss experts any easier. The results of this research demonstrate that there is generally a common view regarding the desirable attributes of an expert in 0 3 effects on crops. This implies that persons possessing the attributes shown to be important in the expert survey will probably conform to the expectations of most groups of stakeholders. This finding is somewhat surprising, because it shows, for example, that members of environmental conservation groups and environmental consultants have approximately the same expectations regarding an expert's credentials. Individual variation within the sample may be the result of different interpretations of the importance scale, but also could be representative of personal preferences reflecting factors such as educational background, geographic location and acquaintances. A rational method for the selection of experts is through nomination by a large group of their peers, as in the present research. It was shown that experts nominated in this manner reflect the desirable profile of the expert held by the scientific community. The primary 208 disadvantage of selection of experts by peer nomination is that it is a somewhat tedious and time-consuming procedure. With this in mind, actuarial models were estimated, using logit analysis, which could be used to predict the probability of being nominated as an expert. The specific attributes of an individual would be used to predict his or her relative degree of expertise. The prediction ability of the logit models is hampered somewhat by the lack of variables representing "normative expertise" and "external credibility". Thus, while the logit models may be useful in assisting with the selection of experts, it must be noted that two highly important dimensions of expertise, related to cognitive skills and public credibility, are not represented. Since these criteria can only be obtained through peer judgment, the peer nomination method is clearly the most reliable method for selecting top experts. The use of experts for predicting crop losses in the Fraser Valley seems particularly justified, considering the lack of exposure-response information for local crops. Although considerable scientific evidence related to 0 3 effects is available from studies in North America and Europe, the vast majority of this information relates to species and/or growing conditions which differ from those in the Fraser Valley. Experts are in the best position to judge the relative applicability of this (indirect) exposure-response information to the Fraser Valley situation. Their judgments regarding the relative applicability of these data were represented in the form of subjective probability distributions, representing probable crop losses under specific 0 3 scenarios. The recognition of the potential usefulness of expert judgment led to the popularity of the Delphi method in the 1960's and early 1970's. The Delphi approach involves feedback of judgments to experts and promotes revision of judgments in an effort to achieve consensus. 209 The Delphi approach was literally abandoned after a review by Sackman (1975), who stated it "rewards conformity and penalizes individuality, and proffers non-independent iterative results as authentic expert consensus." The risk assessment methodology utilized in the present research did not involve feedback of results to experts in an effort to achieve consensus. It was thus possible to judge to what extent independent judgments concur, as well as to judge the real degree of uncertainty that is involved. Another concern related to the Delphi approach is the anonymity assured to experts. Sackman (1975) believes "this guarantees protection against individual accountability, which may promote vested interests or other biases." In the present research, all nine experts participating in the Fraser Valley Risk Assessment Project agreed to be identified. Although the majority of the experts agreed that their judgments could be made public, some of the experts desired that their judgments remain anonymous. For this reason, all experts were identified, but names were not attached to specific judgments. An indirect indication of the accuracy of judgments is related to the extent to which independent assessments concur. In a risk assessment of health effects from air pollution (Morgan et al., 1985), it was found that the independent judgments by atmospheric scientists of sulphur oxidation rates were in good agreement. Conversely, health experts' predictions of health effects resulting from sulphur oxides differed dramatically. Although there was no guarantee that the meteorologists' predictions were accurate (they may have all been wrong), a decision maker would probably have more confidence in the median meteorologists' prediction than in the median health effects prediction. 210 In this research, a higher degree of consensus between experts was observed under the low and medium 0 3 scenarios, than under the high 0 3 scenario. Of interest was the observation that aggregate crop loss estimates, weighted on the basis of number of nominations by peers, exceeded the unweighted judgments for most crops (green bean, pea, potato, raspberry and corn). In other words, for these crops, the experts nominated most frequently predicted higher crop losses than their peers. This might indicate that these individuals possess some information, knowledge or bias that the others do not. An attempt was made through the questionnaire to determine some of the sources of information individual experts used when making judgments. These may assist in explaining some of the differences in judgments between experts. An effort was made to minimize judgment measurement error through inclusion of specific instructions for provision of loss probabilities, and through checks to ensure that the estimates provided were consistent with actual beliefs. Experts were provided with summarized information regarding the commonly observed cognitive bias "anchoring and adjustment" (Tversky and Kahneman, 1974) in an effort to reduce unwarranted overconfidence in the judgments provided. Additionally, prior to distribution of the questionnaire, experts were asked which summary measures of 0 3 exposure they would like to use. Experts were provided with all summary exposure indices, and were asked to comment on which indices were used in formulating their individual judgments. They were also asked to describe other information that was used in making their judgments. Although the responses regarding sources of information used by the various experts were relatively incomplete, it was evident that different information sources were frequently used by different experts, and this probably constituted a major source of variation between 211 individual judgments. For example, with regard to the 0 3 exposure index used, one expert used several monthly indices, including the M7, PI, M12, P7, Hours 90, Hours 100, Sum 90 and Sum 100. Emphasis was "generally placed on June and July exposures". Another expert used the monthly PI index exclusively. Another assumed "average exposure values of 0.08, 0.14 and 0.18 ppm for the low, medium and high 0 3 scenarios" (these values were not provided in the questionnaire, and it is unknown how they were arrived at). Another expert used the "M7 primarily", while another used the "seasonal M7, seasonal Hours 80 . . . 100, and monthly M7". One expert requested the raw 0 3 data be transferred to him by computer, because he was "not a student of the type of 0 3 statistics" that were provided. This expert generated his own 0 3 exposure indices — it is not known what indices were used (although this expert acknowledged "that the literature primarily followed the type of 0 3 exposure indices that were provided in the questionnaire"). Little additional information was provided by experts that could be used to provide insight into why the risk estimates might have varied. Some experts provided the major references used (e.g., U.S. EPA, 1976; NCLAN, "California Air Resources Board Data", National Research Council, 1985; U.S. EPA, 1986; "various publications"). It was impossible to determine to what extent the same literature was used by the various experts. Three of the experts indicated they used the tobacco injury information that was provided for the low (Chilliwack, 1986) and medium (Abbotsford, 1985) 0 3 scenarios. Several experts indicated that the climatic information provided was used. The majority of experts stated they were unfamiliar with the cultivars listed in the questionnaire. One expert indicated he assumed that Fraser Valley cultivars were "more sensitive" than those found in the eastern U.S. Another indicated that provision of crop loss 212 estimates was a "difficult and uncertain task", due to unfamiliarity with the cultivars. This expert indicated that the "Fraser Valley 0 3 conditions were quite low, but local moisture conditions may make crops quite sensitive". Another expert indicated that his crop loss estimates were based on "(i) assumed inherent susceptibility, (ii) capacity to recover during low 0 3 intervals, and (iii) assumed non-drought conditions." It was interesting to note that several experts indicated an unfamiliarity with cultivars named in the questionnaire, but that there was general consensus regarding the assumed sensitivity of the cultivars (Table 6-3). However, considering the apparent variety of information used by the various experts, it is perhaps surprising that reasonable consensus was generally observed regarding probable losses under the low and medium 0 3 scenarios. On the other hand, considering that the general consensus was that impacts would be relatively minor under these two 0 3 scenarios, the results may not be surprising. Fraser Valley 0 3 levels are apparently relatively low compared with other regions where crop loss research is being undertaken. The type of 0 3 conditions typically observed in the Fraser Valley have been shown to result in minor impacts in these other regions. The important question that remains unanswered relates to the actual sensitivity of the Fraser Valley cultivars, growing under Fraser Valley conditions. 7.2.3 Comparison of Crop Loss Assessment Models A direct measure of the accuracy or reliability of judgments is to compare expert judgments with actual empirical results, once the latter are obtained. Although such "calibration 213 studies" have been conducted for certain disciplines (e.g., weather forecasters), no information was previously available which compared crop loss judgments with empirical results. Limited exposure-response information for Fraser Valley crops is available for comparison with the biomonitor and expert risk models. Exposure-response information is available for potato (cv. Russet Burbank) and pea (cv. Puget) (Runeckles and Wright, 1988; Wright, 1988). Linear and reciprocal exposure-response models were fitted for these two crops. The linear exposure-response models are: Yield (pea) = 36.97 - 0.59 X Yield (potato) = 11.83 - 16.88 X where: X = M7 0 3 concentration (ppb) The reciprocal exposure-response models are: Yield (pea) = 612 / (1 + 1.21 X) Yield (potato) = 1678 / (1 + 0.0509 X) where: X = M7 0 3 concentration (ppb) A comparison of crop losses predicted by the various models under the low and medium 0 3 scenarios is presented in Table 7-1. The comparison did not include the high 0 3 scenario because: (i) the associated 0 3 levels are well out of the range of the exposure-response models, (ii) biomonitor injury data (for Anmore in 1981) were not available, and (iii) the high 214 TABLE 7-1 Crop Losses Predicted by Various Models Predicted Crop Losses (%) Crop Loss Model Low Medium 0 3 Scenario1 0 3 Scenario1 (%) (%) (a) Pea Models Biomonitor injury-crop response models2: linear model 2 10 exponential model 4 16 reciprocal model 11 33 Weighted expert risk models3: .05 fractile <1 <1 .50 fractile <1 3 .95 fractile 4 6 Linear exposure-response model 9 36 Reciprocal exposure-response model 11 35 (b) Potato Models Biomonitor injury-crop response models2: linear model <1 3 exponential model 1 4 reciprocal model 1 5 Weighted expert risk models3: .05 fractile <1 1 .50 fractile 2 5 .95 fractile 4 10 Linear exposure-response model 7 30 Reciprocal exposure-response model 7 23 M7 0 3 exposures and biomonitor injury levels used were: Background Low O, Scenario Medium O, Scenario 0 3 exposure 25 ppb 28.2 ppb 38.6 ppb Biomonitor injury (%) 0 % 0.726 % 3.020 % Described in Chapter Three. Described in Chapter Six and Appendix C. 215 0 3 scenario is very atypical and such conditions are not expected to be observed in the future in the rural Fraser Valley. There is a lack of consensus among the three types of models in terms of their predictions. For pea, the expert risk models consistently predicted lower impacts than the biomonitor models, and the biomonitor models predicted lower impacts than the exposure-response models. In the case of potato, the magnitude of impacts predicted by the expert risk models and biomonitor models was similar. Exposure-response model predictions under the low 0 3 scenario were within the same order of magnitude, while exposure-response model predictions under the medium 0 3 scenario were substantially higher than biomonitor or expert risk model predictions. It is not known why the exposure-response models predict higher losses than the other models, but there are several possibilities: (i) The cultivars of the two crops in question may be significantly more sensitive than the experts believe. Several of the experts acknowledged they were not familiar with these particular cultivars. Several researchers, in previous work, have shown that certain potato cultivars are highly sensitive to 0 3, but there is no published data on cv. Russet Burbank, other than that described in this chapter. Ormrod et al. (1971) showed that certain cultivars of potato (Norland) experienced 40% yield losses, while yields of others (Kennebec, Sebago, Superior) were not significantly affected. It was shown by Foster et al. (1983) that yields of Centennial Russet can be reduced by 45% under 0 3 conditions typically experienced in California. Heggestad (1970, 1973) showed that yield reductions 216 of potato due to 0 3 in the eastern U.S. were minor in some cultivars, but were estimated at close to 50% in others. The exposure-response models used for predicting losses in Table 7-1 were based on the M7 exposure index, even though models based on other 0 3 indices clearly outperformed those using the M7 (Wright, 1988). It was necessary to use models based on the M7 index because a standard background level for 0 3 of 25 ppb could be assumed (Heck et al., 1984a). Background levels for other 0 3 exposure indices have not been reported. The assumption that the background level for 0 3 is 25 ppb is highly uncertain. The U.S. Congress, Office of Technology Assessment (1989) assumed a background level of 30 ppb for the continental U.S. The use of 30 ppb background values would result in much lower impact predictions from the exposure-response models. The experts may not have placed enough weight upon the influence of environmental variables unique to the Fraser Valley (e.g., relative humidity) that can increase the susceptibility of vegetation to damage from 03. The experimental crop loss data used in the exposure-response models and the biomonitor models was, in each case, based on a single experiment. The wide variation in response of certain cultivars observed during the NCLAN experiments led Heagle (1989) to question how much confidence should be placed in loss estimates based on three experiments or less. Hence, the accuracy of the exposure-response and biomonitor data used in this research must be verified by additional experimentation. 217 At this time, it is premature to accept or reject any of the crop loss predictions in Table 7-1. In order to resolve the discrepancy between the various models, it will be necessary to conduct additional experiments and generate additional exposure-response information. 7.2.4 Predicted Crop Losses in the Fraser Valley For the purpose of predicting potential crop losses in the Fraser Valley at this time, the weighted expert risk models at the median (0.50 fractile) were used, for the following reasons: (i) The expert models are based on the collective opinion and experience of a large number of experts in 0 3 effects on crops, while the exposure-response and biomonitor models are based on data from a single experiment. The latter models are thus considered to be less reliable than the expert risk models. (ii) Expert risk models are available for seven important Fraser Valley crops, while exposure-response and biomonitor models are available for pea and potato only. The crop production volumes and gross value of these seven crops in 1985 and 1986 (the years of the low and medium 0 3 scenarios) is summarized in Table 7-2. Based on these figures, potential gross crop losses for the seven crops in 1985 and 1986 were estimated using the expert risk models, and are summarized in Table 7-3. The losses included in Table 7-3 were calculated based on the "traditional" method of determining economic effects associated with crop losses from air pollution. The yield losses were simply multiplied by the present 218 TABLE 7-2 Crop Production Volumes and Value: Fraser Valley Crop and Year Production Value Value per Tonne (tonnes) ($ million) ($) Broccoli 1985 7.94 3.24 408,060 1986 6.79 2.64 388,810 Corn 1985 37.61 2.96 78,700 1986 32.68 2.30 70,380 Green bean 1985 7.77 1.22 157,010 1986 6.87 1.14 165,940 Pea 1985 17.12 3.68 214,950 1986 13.84 2.94 212,430 Potato 1985 67.24 6.47 96,220 1986 91.52 9.20 100,530 Raspberry 1985 26.17 20.48 782,580 1986 21.87 26.35 1,204,850 Forage 1985 446,430 50.00 112 1986 446,430 50.00 112 Source: Based on data supplied by Mr. Don Bates, B.C. Ministry of Agriculture and Food. 219 TABLE 7-3 Potential Gross Crop Losses in Fraser Valley Low O, Scenario Medium O, Scenario Crop Fractile % Loss Tonnes Value Fractile % Loss Tonnes Value Lost Lost Broccoli .05 0.14% 0.01 t $ 3,888 .05 0.51% 0.04 t $ 16,332 .50 1.01% 0.07 t 27,217 .50 2.18% 0.17 t 69,370 .95 2.77% 0.19 t 73,874 .95 5.26% 0.42 t 171,385 Corn .05 0.00% 0.00 t $ 0 .05 1.61% 0.61 t $ 48,007 .50 0.54% 0.18 t 12,668 .50 2.94% 1.11 t 87,357 .95 1.52% 0.53 t 37,301 .95 12.44% 4.68 t 368,316 Green bean .05 0.33% 0.03 t $ 4,978 .05 1.89% 0.15 t $ 23,551 .50 2.67% 0.18 t 29,870 .50 5.96% 0.46 t 72,225 .95 5.24% 0.36 t 59,738 .95 10.70% 0.83 t 130,318 Pea .05 0.07% 0.01 t $ 2,124 .05 0.20% 0.03 t $ 6,449 .50 0.71% 0.10 t 21,243 .50 3.12% 0.53 t 113,924 .95 3.54% 0.49 t 104,091 .95 5.95% 1.02 t 219,249 Potato .05 0.07% 0.06 t $ 6,032 .05 1.25% 0.08 t $ 7,698 .50 1.71% 1.56 t 156,827 .50 4.99% 3.35 t 322,337 .95 3.55% 3.24 t 325,717 .95 10.12% 6.80 t 654,296 Raspberry .05 0.13% 0.03 t $ 36,146 .05 0.64% 0.17 t $133,039 .50 1.18% 0.26 t 313,261 .50 3.37% 0.88 t 688,670 .95 3.16% 0.69 t 831,347 .95 6.88% 1.80 t 1,408,644 Forage .05 0.07% 312.50 t $ 35,000 .05 0.47% 2098.20 t $ 234,998 .50 0.32% 1428.60 t 160,000 .50 2.17% 9688.00 t 1,085,056 .95 0.90% 4017.87 t 450,000 .95 4.59% 20491.14 t 2,294,992 Total (above crops) .05 $ 88,168 .05 $ 470,064 .50 721,086 .50 2,438,939 .95 1,882,068 .95 5,247,200 220 market prices of the particular crops (Adams et al., 1984). This method ignores potential price changes and thus represents gross (as opposed to net) revenue changes. Losses computed by this method are usually 10% to 50% higher than losses computed by more sophisticated models which estimate changes in economic surplus (Adams et al., 1984). Loss estimates in Table 7-3 have been estimated based on weighted aggregate expert opinion at three fractiles (0.05, 0.50, 0.95) in order to represent the range of uncertainty that is associated with these estimates. Additional sources of uncertainty not included within these estimates relate to future pollution levels (which are likely to be higher in many years, based on the trend shown in Figure 7-lb), future production volumes and future crop prices. Total losses of all crops from 0 3 in the Fraser Valley would, of course, be significantly higher than losses for the seven crops included in Table 7-3. Under the low ozone scenario, which relates to the type of 0 3 pollution conditions observed at Chilliwack in 1986, the crop loss estimate based on aggregate expert opinion at the 0.50 fractile (the median), for the seven crops, is approximately $721,000. However, there is a 50:50 chance that losses could be lower or higher than this amount. The loss estimate for the seven crops at the 0.05 fractile is approximately $88,000. There is only a 5% chance that losses would be less than this amount under the low 0 3 scenario. The loss estimate for the seven crops at the 0.95 fractile is approximately $1.9 million. There is only a 5% chance that losses would exceed this amount, based on the aggregate opinion of the experts. Under the medium 0 3 scenario, gross crop loss probabilities range from approximately $470,000 at the 0.05 fractile, to $2.4 million at the median, to approximately $5.2 million at the 0.95 fractile. These values represent the range of potential gross crop losses that are predicted to occur based on aggregate expert opinion, and represent the level of uncertainty that is associated with these estimates. 221 CHAPTER EIGHT RESEARCH SUMMARY AND CONTRIBUTIONS Very limited exposure-response information presently exists for Fraser Valley agricultural crops and growing conditions. In this research, two methods were used to provide alternate information related to the potential magnitude of crop losses from 0 3 in the Fraser Valley. These methods included biomonitoring with an 03-sensitive plant, and provision of probabilistic crop loss estimates by experts. The main findings of the research were as follows: 1. A biomonitoring project involving tobacco Bel W-3 provided spatial and temporal information regarding the biological impact of 0 3 throughout the Fraser Valley in the summers of 1985, 1986 and 1987. A weak (r < 0.35) but significant (p < 0.05) correlation was obtained between ambient 0 3 exposure and biological injury to the biomonitor plant. 2. While the weak relationship observed was partially due to atypically low prevailing O, conditions during the project, it was primarily due to the influence of unmeasured environmental conditions that can affect plant uptake of 0 3. As a result, ambient 0 3 concentrations could not be accurately predicted on the basis of vegetative injury, in accordance with the findings of other researchers. 3. The value of biomonitors relates to the information provided by the biological response of the plant to local air pollution conditions, as it is modified by prevailing 222 environmental conditions. Injury is a direct and quantitative measure of impact. Ambient air quality data provide only indirect evidence of injury or damage to plants, and the degree of biological response of the plant (the impact) frequently will not exhibit a close relationship with prevailing pollution conditions. 4. Biomonitor injury observed at different locations and at different times during the three years of the Fraser Valley survey suggested that impacts to other vegetation occurred, even under the atypically low 0 3 conditions that prevailed throughout most of the project. 5. The results of an experiment designed to calibrate the injury response of tobacco Bel W-3 with the losses of certain crops indicated that crop losses may be incurred even if biomonitor injury is negligible. It was shown that even the smallest level of biomonitor injury was associated with measurable crop losses. 6. It was concluded that the absence of biomonitor injury does not necessarily mean that crops have not been affected by 0 3. However, the existence of injury to the biomonitor plant is an indication that yields of certain agricultural crops may have been affected. 7. An alternative method of obtaining information regarding the potential impact of air pollution on crops is through expert judgment. Although risk estimates by experts are subjective and thus cannot be verified without objective scientific evidence, such information is useful to policy makers who must make decisions under uncertainty 223 related to protection of the environment. Expert judgment is valuable for several reasons: (i) It can be used to provide information regarding probable magnitudes of impacts, in the absence of directly applicable scientific data. (ii) It is possible to judge the degree of uncertainty associated with expert judgments, since independent judgments are obtained. (iii) Expert judgments can be obtained relatively quickly and inexpensively, thus avoiding the risk of impacts occurring undetected while the necessary scientific evidence is obtained. (iv) Research priorities may be established on the basis of risk assessment results. There is a paucity of information regarding the necessary and desirable attributes of expertise. As a result, there is likely to be disagreement and misunderstanding about the credentials an expert should possess. A survey was conducted of 238 environmental scientists to provide information regarding the desirable attributes of an expert in 0 3 effects on crops. The survey focused upon the weights and tradeoffs that people place on the various attributes of expertise. Factor analysis was used to group 76 indicators into seven independent dimensions of expertise. These dimensions of expertise and their relative importance scores (maximum = 5.0) were as follows: substantive expertise (3.94), length of career experience (3.85), 224 level of education (3.83), external credibility (3.68), normative expertise (3.65), number of publications (3.26) and involvement in the scientific community (2.33). 11. There was general consensus between different professional groups regarding the importance of the majority of the attributes and dimensions. Where significant differences (p < 0.05) existed between groups, there were generally only minor differences in importance scores. This demonstrates there is a generally common view regarding the desirable profile of an expert in 0 3 effects on crops, even between stakeholders with potentially divergent biases (for example, between consultants, members of conservation groups and research scientists). 12. It was demonstrated using regression analysis that experts selected through nomination by their peers possess many of the attributes depicted in the expert profile developed from the survey. 13. Logit models were developed to assist in the selection of experts in 0 3 effects on crops. The relative expertise of an individual can be predicted, with some confidence, based on specific attributes possessed by that individual. 14. The prediction ability of the models is limited by the lack of variables representing "normative expertise" and "external credibility". Because judgments regarding the degree of possession of these important dimensions can only be made by one's peers, 225 nomination by peers is clearly the most reliable method available for selecting experts at this time. 15. In the present research, nine experts in 0 3 effects on crops were selected based on nomination by a large sample (n = 166) of their scientific peers. Risks of crop losses from 0 3 were elicited from the nine experts under three 0 3 scenarios, based on actual ambient air quality data collected in the Fraser Valley. Crop loss estimates were elicited for each of seven crops at lower (.05), median (.50) and upper (.95) fractiles. Feedback was not provided regarding risk estimates provided, thus ensuring the independence of the various judgments. 16. The judgment data obtained indicates that consensus generally exists between experts regarding probable loss levels of most crops under the low and medium 0 3 scenarios. Less consensus was evident under the high 0 3 scenario. It was determined that the various experts used different information sources in formulating their judgments, in particular regarding 0 3 exposure index and source of exposure-response information. The majority of experts admitted they were "unfamiliar" with Fraser Valley cultivars, although there was reasonable consensus regarding the sensitivity of the various crops assessed. 17. Fraser Valley 0 3 levels are typically much lower than levels in other parts of the continent where exposure-response experiments are carried out. If other environmental conditions were similar, it seems likely that little impact would be associated with typical 226 Fraser Valley 0 3 conditions, as predicted by the experts. However, the potential for degree of impact to be higher or lower in the Fraser Valley than predicted by the experts does exist, based on the lack of information regarding the sensitivity of Fraser Valley crops growing under Fraser Valley conditions. 18. A comparison was made of crop losses predicted by exposure-response, biomonitor and expert crop loss models. The exposure-response models consistently predicted higher impacts than the other models. The reasons for the lack of consensus are unknown, but may be due to: (i) greater sensitivity of Fraser Valley crops; (ii) the use of an unrealistic exposure index or background 0 3 assumption in the empirical models; (iii) the influence of environmental variables unique to the Fraser Valley; or, (iv) the need to verify existing empirical data with additional experiments. 19. Based on weighted aggregate expert opinion, gross losses for the seven crops ranged from $88,000 to approximately $1.9 million under the low 0 3 scenario, and from $470,000 to $5.2 million under the medium 0 3 scenario. The range of potential crop losses represents the degree of uncertainty that is associated with the crop loss estimates under these two scenarios. Median loss estimates are $721,000 and $2.4 million for the low and medium 0 3 scenarios, respectively. Monetary losses of all crops grown in the Fraser Valley would be significantly higher than these figures. 227 Epilogue By reviewing the findings of this research, policy makers will be able to evaluate potential impacts to agriculture in the Fraser Valley and, combined with other information (for example, regarding potential health effects), will be able to make better decisions regarding the appropriate action to take at this time. Such action may take the form of legislating more stringent pollution controls, and it may involve funding of well focused scientific research to establish more clearly the impacts and costs of air pollution. 228 BIBLIOGRAPHY Aaker, D.A. 1980. "Factor Analysis: An Exposition," in Marketing Research, D.A. Aaker and G.S. Day. John W. Wiley and Sons, New York, NY., pp. 163-171. Adams, S. 1986. "Research Assesses Ozone Damage to Crops," Agricultural Research 34, 10, 6-7. Adams, R.M.; Ledeboer, M.V.; McCarl, B.A. 1984. The Economic Effect of Air Pollution on Agriculture: An Interpretive Review of the Literature. Oregon State University, Report 202. Adelman, L.; Mumpower J. 1979. "The Analysis of Expert Judgment," Tech. Fore. Soc. Ch. 15, 191-201. Altshuller, A.P. 1987. "Estimation of the Natural Background of Ozone Present at Surface Rural Locations," JAPCA 37, 12, 1409-1417. Amaral, D. 1988. "Including Uncertainty in Assessments of Sulfur Oxide Health Risks," JAPCA 38, 399-405. Amiro, B.D.; Gillespie, T.J.; Thurtell, G.W. 1984. "Injury Response of Phaseolus vulgaris to Ozone Flux Density," Atmos. Environ. 18, 1207-1215. Angle, R.P.; Sandhu, H.S. 1986. "Rural Ozone Concentrations in Alberta, Canada," Atmos. Environ. 20, 1221-1228. Arkes, H.R. 1981. "Impediments to Accurate Clinical Judgment and Possible Ways to Minimize Their Impact," Cons, and Clin. Psych. 49, 3, 323-330. Arkes, H.R.; Hammond, K.R. (eds.) 1986. Judgement and Decision Making, An Interdisciplinaiy Reader. Cambridge University Press. Ashmore, M.R.; Bell, J.N.B.; Reily, C L . 1978. "A Survey of Ozone Levels in the British Isles Using Indicator Plants," Nature 276, 813-815. Ashmore, M.R.; Bell, J.N.B.; Reily, C L . 1980a. "The Distribution of Phytotoxic Ozone in the British Isles," Environ. Pollut. Ser. B, 195-216. Ashmore, M.R.; Bell, J.N.B.; Dalpra, C; Runeckles, V.C 1980b. "Visible Injury to Crop Species by Ozone in the United Kingdom," Environ. Pollut. Ser. A, 21, 209-215. Ashmore, M.R.; Bell, J.N.B.; Mimmack, A. 1988. "Crop Growth Along a Gladient of Ambient Air Pollution," Environ. Poll. 53, 99-121. 229 Ashton, R.H. 1986. "Combining the Judgments of Experts: How Many and Which Ones?" Organ. Beh. Hum. Dec. Proc. 38, 405-414. Ashton, A.H.; Ashton, R.H. 1985. "Aggregating Subjective Forecasts: Some Empirical Results," Mgmt. ScL 31, 12, 1499-1507. B.C. Environmental Network 1988. The B.C. Environmental Directory: A Directory of Environmental, Peace, Union and Native Organizations. Vancouver, B.C. Bambawale, O.M. 1986. "Evidence of Ozone Injury to a Crop Plant in India," Atmos. Environ. 20, 7, 1501-1503. Beanlands, G.E.; Duinker, P.N. 1983. An Ecological Framework for Environmental Impact Assessment in Canada. Institute for Resource and Environmental Studies, Dalhousie University, Halifax, N.S. Belsley, D.A; Kah, E.; Welsch, R.E. 1980. Regression Diagnostics. John Wiley and Sons, Inc., New York, NY. Bennett, J.H. 1979. "Foliar Exchange of Gases," in Methodology for the Assessment of Air Pollution Effects on Vegetation: A Handbook from a Specialty Conference, W.W. Heck, S.V. Krupa and S.N. Linzon (eds.). Air Pollution Control Association, Pittsburgh, PA. Bennett, J.P.; Resh, H.M.; Runeckles, V.C. 1974. "Apparent Stimulations of Plant Growth by Air Pollution," Can. I. Bot. 52, p. 35. Bennett, J.P.; Runeckles, V.C. 1977. "Effects of Low Levels of Ozone on Plant Competition," /. Appl. Ecol. 14, 877-880. Bicak, C.J. 1978. "Plant Response to Variable Ozone Regimes of Constant Dosage." M.Sc. Thesis, Department of Plant Science, University of British Columbia. Black, V.J.; Unsworth, M.H. 1979. "Resistance Analysis of Sulphur Dioxide Fluxes to vicia fabia," Nature (London), 282, 68-69. Bollen, K.A. 1989. Structural Equations with Latent Variables. John Wiley and Sons, Inc., New York, NY., pp. 313-315. Bradley, M.D. 1983. The Scientist and Engineer in Court. American Geophysical Union, Water Resources Monograph 8, Washington, DC. Brehmer, B. 1986. "In One Word: Not From Experience," in Judgement and Decision Making, An Interdisciplinary Reader, H.R. Arkes and K.R. Hammond (eds). Cambridge University Press, pp. 705-709. Carnahan, J.E.; Jenner, E.L.; Wat, E.K.W. 1978. "Prevention of Ozone Injury to Plants by a New Protectant Chemical," Phytopathology 69, 1225-1229. 230 Clarke, B.B.; Smith, G.; Greenhalgh-Weidman, B.; Brennan, E. 1988. "Assessing the Impact of Ambient Ozone on Field-Grown Potato and Soybean Crops in New Jersey Using the E D U Method." Paper 88-69.3, presented at the 81st Annual Meeting, Air Pollution Control Association, Dallas, TX., June 19-24, 1988. Clark, W.C. 1985. "Technical Uncertainty in Quantitative Policy Analysis," (comment) Risk Analysis 4, 3, 217. C M U 1984. A Conceptual Framework for Integrated Assessments of the Acid Deposition Problem. Draft Final Report. Centre for Energy and Environmental Studies, Carnegie - Mellon University, Pittsburg, PA. Concorde Scientific Corporation and B.H. Levelton and Associates Ltd. 1989. GVRD Air Management Plan - Stage I: Assessment of Current and Future Air Quality. Prepared for Greater Vancouver Regional District, Pollution Control Division, Burnaby, B.C. Cooley, W.W.; Lohnes, P.R. 1971. Multivariate Data Analysis. John Wiley and Sons, Inc., New York, NY. pp. 129-166. Craker, L.E.; Berube, J.L.; Fredrickson, P.B. 1974. "Community Monitoring of Air Pollution with Plants," Atmos. Environ. 8, 845-853. Cure, W.W.; Sanders, J.S.; Heagle, A.S. 1986. "Crop Yield Response Predicted with Different Characterizations of the Same Ozone Treatments," /. Environ. Qual. 15, 3, 251-254. Dunning, J.A.; Heck, W.W. 1973. "Response of Pinto Bean and Tobacco to Ozone as Conditioned by Light Intensity and/or Humidity," Environ. Set Technol. 37, 824-826. Dunning, J.A.; Heck, W.W. 1977. "Response of Bean and Tobacco to Ozone: Effect of Light Intensity, Temperature and Relative Humidity," JAPCA 27, 882-886. Edwards, Mr. G , Head, Compliance and Enforcement, Conservation and Protection, Western and Northern Region, Environment Canada. Personal communication, February, 1989. Einhorn, H J . 1986. "Expert Judgment: Some Necessary Conditions and an Example", in Judgment and Decision Making: an Interdisciplinary Reader. Cambridge University Press, pp. 480-491. Environment Canada. 1981. The Clean Air Act - Compilation of Regulations and Guidelines. Report EPS l-AP-81-1. Ottawa. Evans, G.; Finkelstein, P.; Martin, B.; Possiel N . 1983. "Ozone Measurements from a Network of Remote Sites," JAPCA 33, 4, 291-296. Feagans, T.B.; Biller, W.F. 1981. "Risk Assessment: Describing the Protection Provided by Ambient Air Quality Standards," Environ. Prof. 3, p. 235. 231 Feder, W.A; Kelleher, T.J.; Riley, W.D.; Perkins, I.; Moeller, W.K. 1975. "Ozone Injury on Tobacco Plants on Nantucket Island Is Caused by Long Range Transport of Ozone from the Mainland," Proc. Amer. Phytopath. Soc. Division Meeting Abstracts, NE-22, p. 97. Feder, W.A; Manning, W.J. 1979. "Living Plants as Indicators and Monitors," in Methodology for the Assessment of Air Pollution Effects on Vegetation: A Handbook from a Specialty Conference, W.W. Heck, S.V. Krupa and S.N. Linzon (eds.). Minneapolis, MN., April 1978. Air Pollution Control Association, Pittsburgh, PA. pp. 9-1 - 9-14. Fischoff, B.; Slovic, P.; Lichtenstein, S. 1977. "Knowing With Certainty: The Appropriateness of Extreme Confidence," /. Exper. Psych. 3, 4, 552-564. Fischoff, B.; Lichtenstein, S.; Slovic, P.; Derby, R.L.; Keeney, R.L. 1981. Acceptable Risk. Cambridge University Press, New York, NY. 185 pp. Fisheries and Environment Canada 1976. Criteria for National Air Quality Objectives: Sulphur Dioxide, Suspended Particulates, Carbon Monoxide, Oxidants (Ozone), and Nitrogen Dioxide. Report to the Federal-Provincial Committee on Air Pollution by the Subcommittee on Air Quality Objectives. Ottawa, Ontario. Fong, F. 1984. "Mechanisms of Acute and Chronic Effects of Ozone Injury," in Evaluation of the Scientific Basis for Ozone I Oxidants Standards, S.D. Lee (ed.). Air Pollution Control Association, Pittsburgh, PA. pp. 107-114. Foster, K.W.; Timm, H.; Labanauskas, C.K.; Oshima, RJ. 1983. "Effects of Ozone and Sulfur Dioxide on Tuber Yield and Quality of Potatoes," /. Environ. Qual. 12, 75-80. Fowler, D.; Cape, J.N. 1982. "Air Pollutants in Agriculture and Horticulture," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 3-26. Fraser, G.A.; Phillips, W.E.; Lamble, G.W.; Hogan, G.D.; Teskey, A G . 1985. The Potential Impact of the Long Range Transport of Air Pollution on Canadian Forests. Canadian Forestry Service Information Report E-X-36. Edmonton, Alberta. Godzik, S.; Krupa, S.V. 1982. "Effects of Sulphur Dioxide on Growth and Yield of Crops," in Effects of Gaseous Air Pollution In Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 247-266. Goldberg, L.R. 1968. "Simple Models or Simple Processes? Some Research on Clinical Judgments," Amer. Psych. 23, 7, 483-496. Guderian, R.; Tingey, D.T.; Rabe, R. 1985. "Effects of Photochemical Oxidants on Plants," in Air Pollution by Photochemical Oxidants: Formation, Transport, Control, and Effects on Plants, R. Guderian (ed.). Ecological Studies 52, Springer-Verlag, Berlin, Germany, pp. 129-333. 232 Hammond, KR.; Adelman, L. 1976. "Science, Values and Human Judgment," Science 194, 389-396. Heagle, AS. 1989. "Ozone and Crop Yield," Annu. Rev. Phytopathol. 27, 397-423. Heagle, AS.; Body, D.E.; Heck, W.W. 1973. "An Open-Top Field Chamber to Assess the Impact of Air Pollution on Plants," /. Environ. Qual. 2, 365-368. Heagle, AS.; Heck, W.W.; Lesser, V.M.; Rawlings, J.O. 1987. "Effects of Daily Ozone Exposure Duration and Concentration Fluctuation on Yield of Tobacco," Phytopathology 77, 856-862. Heck, W.W. 1966. "The Use of Plants as Indicators of A r Pollution," Air Water Pollut. 10, 99-111. Heck, W.W. 1982. "Future Directions in A r Pollution Research," in Effects of Gaseous Air Pollution In Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 441-436. Heck, W.W. 1988. "Assessment of the Impacts of Ozone on Crop Production," in Emerging Priorities for Agricultural Research in the 1990's. 37th Annual Meeting, Agricultural Research Institute, Washington, DC, October 12-14, 1988. pp. 37-53. Heck, W.W.; Dunning, J.A; Hindawi, I. J. 1966. "Ozone: Nonlinear Relation of Dose and Injury to Plants," JAPCA 17, 112-114. Heck, W.W.; Heagle, A.S. 1970. "Measurement of Photochemical A r Pollution with a Sensitive Monitoring Plant," JAPCA 20, 97-99. Heck, W.W.; Taylor, O.C.; Adams, R.; Bingham, G ; Miller, J.; Preston, E.; Weinstein, L. 1982. "Assessment of Crop Loss from Ozone," JAPCA 32, 353-361. Heck, W.W.; Adams, R.M.; Cure, W.W.; Heagle, AS.; Heggestad, H.E.; Kohut, R.J.; Kress, L.W.; Rawlings, I.O.; Taylor, O.C. 1983. "A Reassessment of Crop Loss from Ozone," Environ. Set Technol. 17, 573A-581A Heck, W.W.; Cure, W.W.; Rawlings, J.O.; Zaragoza, L.J.; Heagle, AS.; Heggestad, H.E.; Kohut, R.J.; Kress, L.W.; Temple, PJ. 1984a. "Assessing Impacts of Ozone on Agricultural Crops: I - Overview," JAPCA 34, 729-735. Heck, W.W.; Cure, W.W.; Rawlings, J.O.; Zaragoza, L.J.; Heagle, AS.; Heggestad, H.E.; Kohut, R.J.; Kress, L.W.; Temple, PJ. 1984b. "Assessing Impacts of Ozone on Agricultural Crops: II - Crop Yield Functions and Aternative Exposure Statistics," JAPCA 34, 810-817. 233 Heck, W.W.; Taylor, O.C.; Tingey, D.T. (eds.) 1988. Assessment of Crop Loss From Air Pollutants: Proceeding of the International Conference: Raleigh, NC. Elsevier Applied Science. Heggestad, H.E. 1970. "Variation in Response of Potato Cultivars to Air Pollution," Abstr. Phytopathology 60, p. 1015. Heggestad, H.E. 1973. "Photochemical Air Pollution Injury to Potatoes in the Atlantic Coastal States," Am. Potato J. 50, 315-328. • Heggestad, H.E.; Bennett, J.H. 1981. "Photochemical Oxidants Potentiate Yield Losses in Snap Beans Attributable to Sulfur Dioxide," Science 213, 1008-1010. Heggestad, H.E.; Bennett, J.H. 1984. "Impact of Atmospheric Pollution on Agriculture," in Air Pollution and Plant Life, M. Treshow (ed.). John Wiley and Sons, Inc., New York, NY. pp. 357-397. Heggestad, H.E.; Menser, H.A. 1962. "Leaf Spot-Sensitive Tobacco Strain Bel W-3, a Biological Indicator of the Air Pollutant Ozone," Phytopathology 52, p. 735. Helmer, O. 1983. Looking Forward: A Guide to Futures Research. Sage Publications, Inc. Hogarth, R.M. 1980. Judgment and Choice. Wiley-Interscience, John Wiley and Sons, Inc., New York, NY. Hogarth, R.M. 1981. "Beyond Discrete Biases: Functional and Dysfunctional Aspects of Judgmental Heuristics," Psych. Bull. 90, 2, 197-217. Hogsett, W.E.; Tingey, D.T.; Holman, S.R. 1985. "A Programmable Exposure Control System for Determination of the Effects of Pollutant Exposure Regimes on Plant Growth," Atmos. Environ. 19, 1135-1145. Hogsett, W.E.; Tingey, D.T.; Lee, E.H. 1988. "Exposure Indices: Concepts for Development and Evaluation of their Use," in Assessment of Crop Loss from Air Pollutants: Proceedings of the International Conference, W.W. Heck, O.C. Taylor and D.T. Tingey (eds.). Raleigh, NC, Elsevier Applied Science, pp. 107-138. Horowitz, J.L. 1982. Air Quality Analysis for Urban Transportation Planning. MIT Press, Cambridge, MA. Horsman, D.C. 1981. "A Survey of Ozone in Melbourne Using Tobacco as an Indictor Plant," Environ. Pollut. Ser. B, 69-77. 234 Jacobson, J.S. 1982. "Ozone and the Growth and Productivity of Agricultural Crops," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 293-304. Jacobson, J.S.; Feder, W.A. 1974. "A Regional Network for Environmental Monitoring: Atmospheric Oxidant Concentrations and Foliar Injury for Tobacco Indicator Plants in the Eastern United States." Massachusetts Agricultural Experiment Station, College of Food and Natural Resources, Amherst, M A Bulletin N. 604. Johnson, D. 1963. "Characteristics of Productive Scientists," in Judgement and Choice, R.M. Hogarth, 1980. Wiley-Interscience, John Wiley and Sons, Inc., New York, NY., p. 117. Jones, H.C.; Lacasse, N.L.; Liggett, W.S.; Weatherford, F. 1977. "Experimental A r Exclusion System for Field Studies of S02 Effects on Crop Productivity." Report No. EPA-600/7-77-122, U.S. Environmental Protection Agency, Washington, DC. Joreskog, K G . 1979. "Basic Ideas of Factor and Component Analysis," Advances in Factor Analysis and Structural Equation Models, K G . Joreskog and D. Sorbon (eds.). Cambridge Mass.: Abt Books, pp. 5-20. Kahneman, D.; Slovic, P.; Tversky, A. (eds.) 1982. Judgment Under Uncertainty: Heuristics and Biases. Cambridge University Press. Keefer, D.L.; Bodily, S.E. 1983. "Three-Point Approximations for Continuous Random Variables," Mgmt. Set 29, 5, 595-609. Keeney, R.L.; Sarin, R.K.; Winkler, R.L. 1984. "Analysis of Aternative National Ambient Carbon Monoxide Standards," Mgmt. Sci 30, 4, p. 518. Keeney, R.L.; von Winterfeldt, D.V. 1986. "Improving Risk Communication," Risk Anal. 6, 4, 417-424. Keeney, R.L.; von Winterfeldt, D. 1988. Use of Expert Judgment on Complex Technical Problems. Systems Science Dept., University of Southern California, Los Angeles, CA. Kelleher, T.J.; Feder, W.A 1978. "Phytotoxic Concentrations of Ozone on Nantucket Island: Long Range Transport from the Middle Atlantic States over the Open Ocean Confirmed by Bioassay with Ozone-Sensitive Tobacco Plants," Environ. Pollut. 17, 187-194. Kelly, N.A; Wolff, G.T; Ferman, M.A 1984. "Sources and Sinks of Ozone in Rural Aeas," Atmos. Environ. 18, 1251-1266. Kinnear, T.C.; Taylor, J.R. 1983. Marketing Research: An Applied Approach. McGraw-Hill, Inc. Kohut, R.J.; Laurence, J.A.; Colavito, L.J. 1988. "The Influence of Ozone Exposure Dynamics on the Growth and Yield of Kidney Bean," Environ. Pollut. 53, 79-88. 235 Kosuslco, M.; Nolen, S.L. 1989. "Area Sources of VOC Emissions and their Contribution to Tropospheric Ozone Concentrations." Paper 89-3.2, presented at the 82nd Annual Meeting and Exhibitionof the Air and Waste Management Association, Anaheim, CA., June 25-30, 1989. Krupa, S.V.; Teng, P.S. 1982. "Uncertainties in Estimating the Ecological Effects of Air Pollutants." Paper 82-6.1, presented at the 75th Annual Meeting, Air Pollution Control Association. Krupa, S.V.; Kickert, R.N. 1987. "An Analysis of Numerical Models of Air Pollutant Exposure and Vegetation Response," Environ. Pollut. 44, 2, 127-158. Krupa, S.V.; Manning, W.J. 1988. "Atmospheric Ozone: Formation and Effects on Vegetation," Environ. Pollut. 50, 101-137. Krupa, S.V.; Nosal, M. 1989. "Effects of Ozone on Agricultural Crops," in Atmospheric Ozone Research and its Policy Implications, T. Schneider et al. (eds.). Elsevier Science Publishers, Amsterdam, pp. 251-260. Larsen, R.I.; Heck, W.W. 1984. "An Air Quality Data Analysis System for Interrelating Effects, Standards, and Needed Source Reductions: Part 8 - An Effective Mean 03 Crop Reduction Mathematical Model," JAPCA 34, 1023-1034. Larsen, R.I.; McCurdy, T.R.; Johnson, P.M.; Heck, W.W. 1988. "An Air Quality Data Analysis System for Interrelating Effects, Standards, and Needed Source Reductions: Part 10 -Potential Ambient 0 3 Standards to Limit Soybean Crop Reduction," JAPCA 38, 12, 1497-1503. Laurence, J.A; Greitner, C.S. 1980. "A Literature Review: Biological Indicators of Air Pollutants," in Development of a Biological Air Quality Indexing System. Report SEO 184B, Minnesota Environmental Quality Board, January 1984. pp. 13-41. Lee, E.H.; Tingey, D.T.; Hogsett, W.E. 1987. "Selection of the Best Exposure-Response Model Using Various 7-Hour Ozone Exposure Statistics." U.S. EPA, Office of Air Quality Planning and Standards, Research Triangle Park, NC, 1987. Lee, E.H.; Tingey, D.T.; Hogsett, W.E. 1988. "Evaluation of Ozone Exposure Indices in Exposure-Response Modelling," Environ. Pollut. 53, 43-62. Lee, J.J.; Lewis, R.A.; Body, D.E. 1975. "A Field Experimental System for the Evaluation of the Bioenvironmental Effects of S02," in Fort Union Coal Field Symposium, W.S. Clark (ed.). E. Montana College, Billings, MT. pp. 608-620. Lee, U.B.; Comrey, A.L. 1979. "Distortions in a Commonly Used Factor Analytic Procedure," Mult. Beh. Res. 14, 301-321. 236 Lefohn, AS. 1984. "A Comparison of Ambient Ozone Exposures for Selected Non-Urban Sites," presented at the 77th Annual Meeting, Air Pollution Control Association, San Francisco, CA, June 24-29, 1984. Lefohn, AS.; Benedict, H.M. 1982. "Development of a Mathematical Index that Describes Ozone Concentration, Frequency, and Duration," Atmos. Environ. 16, 2529-2532. Lefohn, AS.; Benedict, H.M. 1985. "Exposure Considerations Associated with Characterizing Ozone Ambient A r Quality Monitoring Data," in Evaluation of Scientific Basis for Ozone/Oxidants Standards, S.D. Lee (ed.). A r Pollution Control Association, Pittsburgh, PA pp. 17-31. Lefohn, AS.; Runeckles, V.C. 1987. "Establishing Standards to Protect Vegetation-ozone Exposure/Dose Considerations," Atmos. Environ. 21, 3, 561-568. Lefohn, AS.; Hogsett, W.E.; Tingey, D.T. 1987. "The Development of Sulfur Dioxide and Ozone Exposure Profiles that Mimic Ambient Conditions in the Southeastern United States," Atmos. Environ. 21, 3, 659-669. Lefohn, AS.; Laurence, J.A; Kohut, R.J. 1988. "A Comparison of Indices that Describe the Relationship Between Exposure to Ozone and Reduction in Yield of Agricultural Crops," Atmos. Environ. 22, 6, 1229-1240. Lefohn, AS.; Runeckles, V.C; Krupa, S.V.; Shadwick, D.S. 1989. "Important Considerations for Establishing a Secondary Ozone Standard to Protect Vegetation," JAPCA 39, 8, 1039-1945. Lopes, L.L. 1982. "Doing the Impossible: A Note on Induction and the Experience of Randomness," /. Exper. Psych. 8, 6, 626-636. MacDowall, F.D.H.; Vickery, L.S.; Runeckles, V.C; Patrick, Z.A 1963. "Ozone Damage to Tobacco in Canada," Can. PL Dis. Sutv. 43, 131-151. MacDowall, F.D.H.; Mukammal, E.I.; Cole, AF.W. 1964. "Direct Correlation of Ar-Polluting Ozone and Tobacco Weather Reck," Can. J. PI. Sci 44, 410-417. Maddala, G.W. 1983. Limited Dependent and Qualitative Variables in Econometrics. Umbridge University Press, New York, NY. pp. 22-27. Male, L.M. 1981. Critique of Heck-Heagle-Rawlings Experimental Design. Internal memorandum, Corvallis Environmental Research Laboratory. Male, L.M. 1982. "An Experimental Method for Predicting Plant Yield Response to Pollution Time Series," Atmos. Environ. 16, 9, 2247-2252. 237 Manning, W.J. 1988. "EDU: A Research Tool for Assessment of the Effect of Ozone on Vegetation." Paper 88-69.2, presented at the 81st Annual Meeting, Air Pollution Control Association, Dallas, TX., June 19-24, 1988. Manning, W.J.; Feder, W.A. 1980. Biomonitoring Air Pollutants with Plants. Applied Science Publishers, London, England. Martino, J.P. 1972. Technological Forecasting for Decision Making. American Elsevier Publishers, New York, NY. McCool, P.M.; Musselman, R.C; Younglove, T.; Teso, R.R. 1988. "Response of Kidney Bean to Sequential Ozone Exposures," Environ. Exper. Bot. 28, 4, 307-313. MEQB 1984. Development of a Biological Air Quality Indexing System. Minnesota Environmental Quality Board, Power Plant Siting Program, St. Paul, MN. Mildred, R.H. 1977. The Expert Witness. George Godwin, London, England and New York, NY. Miller, J.E.; Heagle, AS.; Vozzo, S.F.; Philbeck, R.B.; Heck, W.W. 1989. "Effects of Ozone and Water Stress, Separately and In Combination, on Soybean Yield," /. Environ. Qual 18, 3, 330-336. Morgan, M.G.; Henrion, M.; Morris, S.C. 1979. Expert Judgments for Policy Analysis. Brookhaven National Laboratory, Upton, New York, NY. Morgan, M.G.; Morris, S.C; Henrion, M.; Amaral, D.; Rish, W.R. 1984. "Technical Uncertainty in Quantitative Policy Analysis: A Sulfur Air Pollution Example," Risk Anal. 4, 3, p. 201. Morgan, M.G.; Morris, S.C; Amaral, D. 1985. "Uncertainty in Risk Assessment: A Case Study Involving Sulfur Transport and Health Effects," Environ. Set Technol. 19, 8, p. 662. Morganstern, O.; Knorr, K.; Heiss, K.P. 1973. Long Term Projections of Power: Political, Economic and Military Forecasting. Ballinger Publishing Company, Cambridge, MA. Mukammal, E.I. 1965. "Ozone as a Cause of Tobacco Injury," Agric. Met. 2, 145-165. Murphy, AH.; Winkler, R.L. 1974. "Probability Forecasts: A Survey of National Weather Services Forecasters," Bull. Amer. Met. Soc. 55, 1449-1453. Musselman, R.C; Oshima, R.J.; Gallavan, R.E. 1983. "Significance of Pollutant Concentration Distribution in the Response of 'Red Kidney' Beans to Ozone," /. Am. Soc. Hortic. Sci. 108, 347-351. National Academy of Sciences 1977. Perspectives on Technical Information for Environmental Protection. Washington, DC. 238 National Research Council 1977. Ozone and Other Photochemical Oxidants. National Academy of Sciences, Washington, DC. pp. 437-585. Naveh, Z.; Chaim, S.; Steinberger, E.H. 1978. "Atmospheric Oxidant Concentrations in Israel as Manifested by Foliar Injury in Bel W-3 Tobacco Plants," Environ. Pollut. 16, 249-262. Noggle, J.C. 1982. Sulphur Accumulation by Plants: The Role of Gaseous Sulphur in Crop Nutrition," in Atmospheric Sulfur Deposition: Environmental Impact and Health Effects, D.S. Shriner, CR. Richmond and S.E. Lindberg (eds.). Ann Arbor Science. North, D.W.; Balson, W.E.; Colville, G. 1985. Analysis of Sulfur Dioxide Control Strategies Related to Acid Deposition in Wisconsin: Volume 1 - Application of Decision Analysis. Prepared for Wisconsin Utilities Acid Deposition Task Force. Nosal, M. 1983. Atmosphere-Biosphere Interface: Probability Analysis of an Experimental Design for Studies of Air-Pollutant Induced Response. Alberta Environment, Edmonton, Alberta. Nosal, M. 1984. "Statistical Models for Air Pollutant Dose and Plant Response." Paper 84-104.5, presented at the August 27th Annual Meeting, Air Pollution Control Association, San Francisco, CA., June, 1984. Nouchi, I.; Aoki, K. 1979. "Morning Glory as a Photochemical Oxidant Indicator," Environ. Pollut. 18, 289-303. Ontario Ministry of the Environment 1984. Ozone Effects on Crops in Ontario and Related Monetary Values. Report ARB-13-84-Phyto. Orians, G.H. 1986. "The Place of Science in Environmental Problem Solving," Environment 28, 9, 12-17. Ormrod, D.P.; Adedipe, N.O.; Hofsta, G. 1971. "Responses of Cucumber, Onion and Potato Cultivars to Ozone," Can. J. Plant Set 51, 283-288. Oshima, R.J. 1974. "A Viable System of Biological Indicators for Monitoring Air Pollutants," JAPCA 24, 576-578. Oshima, R.J.; Taylor, O.C; Braegelmann, P.K.; Baldwin, D.W. 1975. "Effect of Ozone on the Yield and Plant Biomass of a Commercial Variety of Tomato," /. Environ. Qual. 4, 463-464. Oshima, R.J.; Poe, M.P.; Braegelmann, P.K.; Baldwin, D.W.; Van Way, V. 1976. "Ozone Dosage-Crop Loss Function for Alfalfa: A Standardized Method for Assessing Crop Losses from Air Pollutants," JAPCA 26, 861-865. Peake, E.; Maclean, M.A. 1983. "Surface Ozone and Peroxyacetyl Nitrate (PAN) Observations at Rural Locations in Alberta, Canada," JAPCA 33, 9, 881-883. 239 Peterson, D.C; Violette, D.M. 1985. Dimensioning Uncertainty in Estimates of Regional Fish Population Damage Caused by Acidification in Adirondack Ponded Waters. Prepared for the Office of Policy Analysis, U.S. Environmental Protection Agency. Politt, D.F.; Hungler, B.P. 1983. Research Principles and Methods. 2nd edition, J.B. Lippincott Co., Toronto, Ontario. Posthumus, A.C. 1982. "Exposure to Gaseous Pollutants and Uptake by Plants," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 43-65. Pratt, G.C.; Hendrickson, R.C; Chevone, B.I.; Christopherson, D.A.; O'Brien, M.V.; Krupa, S.V. 1983. "Ozone and Oxides of Nitrogen in the Rural Upper-Midwestern U.S.A.," Atmos. Environ. 17, 10, 2013-2023. Price, P.B.; Taylor, C.W.; Nelson, D.E.; Lewis, E.G.; Loughmiller, G.C.; Mathieson, R.; Murray, S.L.; Maxwell, J.G. 1971. Measurement and Predictors of Physician Performance: Two Decades of Intermittently Sustained Research. Department of Psychology, University of Utah. Rafiq, M. 1986. Ozone Impact on Fraser Valley Crops: A Preliminary Assessment. Ministry of Environment and Parks, Waste Management Branch, Surrey, B.C. Ravetz, J. 1978. "Scientific Knowledge and Expert Advice in Debates about Large Scale Technological Innovations," Minerva 16, 2, 273-282. Rawlings, J.O.; Cure, W.W. 1985. "The Weibull Function as a Dose-Response Model to Describe Ozone Effects on Crop Yields," Crop Sci 25, 807-814. Rawlings, J.O.; Lesser, V.M.; Heagle, AS.; Heck, W.W. 1988. "Alternative Ozone Dose Metrics to Characterize Ozone Impact on Crop Yield Loss," /. Environ. Qual. 17, 2, 285-291. Reich, P.B.; Amundson, R.G. 1986. "Ambient Levels of Ozone Reduce Net Photosynthesis in Tree and Crop Species," Science 230, 566-570. Roose, M.L.; Bradshaw, A D . 1982. "Evolution of Resistance for Gaseous Air Pollutants," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth and D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 379-410. Rowe, W.D. 1977. An Anatomy of Risk. Wiley-Interscience Publication, John Wiley and Sons, New York, NY. Ruckelhaus, W.D. 1983. "Science, Risk and Public Policy," Science 221, 1026-1028. 240 Runeckles, V.C. 1974. "Dosage of Air Pollutants and Damage to Vegetation," Environ. Cons. 1, 4, 305. Runeckles, V.C. 1986. "Terrestrial Vegetation-Air Pollutant Interactions: Gaseous Pollutants-Photochemical Oxidants," in Air Pollutants and Their Effects on Terrestrial Ecosystems, S.V. Krupa and A H . Legge (eds.). John Wiley and Sons, Inc., New York, NY. pp. 265-304. Runeckles, V.C. 1987. "Exposure, Dose, Vegetation Response and Standards: Will They Ever Be Related?" Paper 87-33.2, presented at the 80th Annual Meeting, Air Pollution Control Association, New York, NY., June 21-26, 1987. Runeckles, V.C. 1988. Comment on AS. Lefohn, J.A Laurence, and RJ. Kohut, 1988, Atmos. Environ.. 22, 6, 1241. Runeckles, V.C; Rosen, P.M. 1977. "Effects of Ambient Ozone Pretreatment on Transpiration and Susceptibility to Ozone Injury," Can. J. Bot. 55, 193-197. Runeckles, V.C; Staley, L.M.; Bulley, N.R. 1978. "A Downdraft Chamber for Studying the Effects of A r Pollutants on Plants," Can. J. Bot. 56, 768-778. Runeckles, V.C; Palmer, K T ; Trabelsi, H. 1981. "Effects of Field Exposures to S0 2 on Douglas Fir," Silva Fennica 15, 4, 405-415. Runeckles, V.C; Brown, G.L. 1986. "Risk Assessment and Management of the Ecological Effects of Ozone." Paper 86-92.2, presented at the 79th Annual Meeting, A r Pollution Control Association, Minneapolis, MN., June 22-27, 1986. Runeckles, V.C; Palmer, K 1987. "Pretreatment with Nitrogen Dioxide Modifies Plant Response to Ozone," Atmos. Environ. 21, 3, 717-719. Runeckles, V.C; Wright, E.F. 1988. "Crop Response to Pollutant Exposure - The "ZAPS" Approach." Paper 86-92.2, presented at the 81st Annual Meeting, A r Pollution Control Association, Dallas, TX., June 19-24, 1988. Runeckles, V.C; Wright, E.F. 1989. "Exposure-Response Models for Crops." Paper 89-89.5, presented at the 82nd Annual Meeting, Ar and Waste Management Association, Los Angeles, C A Runeckles, V.C; Wright, E.F.; White, D. 1990. "A Chamberless Field Exposure System for Determining the Effects of Gaseous A r Pollutants on Crop Growth and Yield," Environ. Poll. 63, 61-77. Russel, M.; Gruber, M. 1987. "Risk Assessment in Environmental Policy-Making," Science 236, 286-290. 241 Sackman, H. 1975. Delphi Critique: Expert Opinion, Forecasting and Group Process. Lexington Books, MA. Saks, M.J.; Hastie, R. 1985. "Social Psychology in Court: The Judge," in Judgment and Decision Making, An Interdisciplinary Reader, H.R. Arkes and KR. Hammond (eds.). Cambridge University Press, pp. 255-275. Saks, M.J.; Kidd, R.F. 1986. "Human Information Processing and Adjudication: Trial by Heuristics," in Judgment and Decision Making, An Interdisciplinary Reader, H.R. Arkes and K.R. Hammond (eds.). Cambridge University Press, pp. 213-243. SAS User's Guide: Statistics 1986. SAS Institute Inc., Cary, NC. Shanteau, J. 1984. "Some Unasked Questions About the Psychology of Expert Decision Makers," in Proceedings of the 1984 IEEE Conference on Systems, Man, and Cybernetics, M.E. Elhawary (ed.). New York, NY. Shanteau, J. 1987. Psychological Characteristics and Strategies of Expert Decision Makers. Dept. of Psychology, Kansas State University, Manhattan, KA. Sigal, L.L.; Suter, G.W. 1987. "Evaluation of Methods for Determining Adverse Impacts of Air Pollution on Terrestrial Ecosystems," Environ. Manag. 11, 5, 675-694. Simon, H.A. 1986. "Alternative Visions of Rationality," in Judgment and Decision Making, An Interdisciplinary Reader, H.R. Arkes and K.R. Hammond (eds.). Cambridge University Press, pp. 97-114. Simpson, R.W. 1988. "A Human Ecological Assessment of Air Quality Management: A Convergence in Economic and Ecological Thinking?" Environ. Manag. 12, 3, 285-295. Slovic, P. 1987. "Perception of Risk," Science 236, 280-285. Smith, G.; Greenhalgh, B.; Brennan, E.; Justin, J. 1987. "Soybean Yield in New Jersey Relative to Ozone Pollution and Antioxidant Application," Plant Disease 71, 121-125. Spetzler, C.S.; von Holstein, C.S. 1975. "Probability Encoding in Decision Analysis," Mgmt. Sci. 22, 3, 340. Suter II, G.W.; Barnthouse, L.W.; O'Neill, R.V. 1988. "Treatment of Risk in Environmental Impact Assessment," Environ. Manag. 11, 3, 295-303. Taylor, G.E. 1984. "The Significance of Elevated Levels of Ozone on Natural Ecosystems in North America," in Evaluation of the Scientific Basis for Ozone I Oxidants Standards, S.D. Lee (ed.). Air Pollution Control Association, Pittsburg, PA. pp. 152-175. 242 Taylor, G.E., Jr.; McLaughlin, S.B.; Shriner, D.S. 1982. "Effective Pollutant Dose," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth, D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 458-460. Temple, P.J.; Benoit, L.P. 1988. "Effects of Ozone and Water Stress on Canopy Temperature, Water Use and Water Use Efficiency of Alfalfa," Agron. J. 80, 3, 439-447. Teng, P.S. 1982. "A Report Evaluating the Possible Use of Biological Systems to Assess the Economic Effects of Air Pollutants on Crops and Forests," in Development of a Biological Air Quality Indexing System: A Report to the Minnesota Environmental Quality Board, (SE0184B), 1984. Tingey, D.T. 1989. "Bioindicators in Air Pollution Research," in Biologic Markers of Air Pollution Stress and Damage in Forests. National Academy Press, Washington, DC. pp. 73-80. Tingey, D.T.; Taylor, G.E., Jr. 1982. "Variation in Plant Response to Ozone: A Conceptual Model of Physiological Events," in Effects of Gaseous Air Pollution in Agriculture and Horticulture, M.H. Unsworth, D.P. Ormrod (eds.). Butterworth Scientific, London, England, pp. 113-138. Tingey, D.T.; Hogsett, W.E.; Lee, E.H. 1988. "Analysis of Crop Loss for Alternative Ozone Exposure Indices." Paper 88-121.4, presented at the 81st Annual Meeting, Air Pollution Control Association, Dallas, TX., June 19-24, 1988. Tingey, D.T.; Hogsett, W.E.; Lee, E.H. 1989. "Analysis of Crop Loss for Alternative Ozone Exposure Indices," in Atmospheric Ozone Research and its Policy Implications, T. Schneider et al. (eds.). Elsevier Science Publishers, Amsterdam, pp. 219-227. Tonneijck, A.E.G.; Posthumus, A C . 1987. "Use of Indicator Plants for Biological Monitoring of Effects of Air Pollution: The Dutch Approach," in VDI Berichte NR. 609, pp. 205-216. Turner, H.E. 1985. Environmental Risk Management: An Introduction. Prepared for the Federal-Provincial Advisory Committee on Air Quality Meeting, Quebec City, Quebec, May, 1985. Tversky, A ; Kahneman. D. 1974. "Judgement Under Uncertainty: Heuristics and Biases," Science 185, 1124-1131. U.S. Congress, Office of Technology Assessment. 1989. "Effects of Ozone on Crops and Forests," in Catching Our Breath: Next Steps for Reducing Urban Ozone. OTA-0-412. Washington, DC. pp. 79-93. U.S. EPA 1978. Air Quality Criteria for Ozone and Other Photochemical Oxidants. Report No. EPA-600/8-78-004, U.S. Environmental Protection Agency, Environmental Criteria and Assessment Office, Research Triangle Park, NC. 243 U.S. EPA 1985. List of Conservation Groups Within North America. Washington, DC. U.S. EPA 1986. Air Quality Criteria for Ozone and Other Photochemical Oxidants, Volume I. Report No. EPA-600/8-84-02aF, Research Triangle Park, NC. U.S. EPA 1987. Air Pollution Exposure Systems and Experimental Protocols. Report No. EPA-600/3-87-037b, Environmental Research Laboratory, Corvallis, OR. Vaartnou, M. 1988. EPR Investigation of Free Radicals in Excised and Attached Leaves Subjected to Ozone and Sulphur Dioxide. Ph.D. Thesis, Dept. of Plant Science, University of British Columbia. von Winterfeldt, D.V.; Edwards, W. 1986. Decision Analysis and Behavioural Research. Cambridge University Press. Walmsley, L.; Ashmore, M.L.; Bell, J.N.B. 1980. "Adaptation of Radish Raphanus sadvus L. in Response to Continuous Exposure to Ozone," Environ. Pollut. 23, 165-177. Wallsten, T.S.; Forsythe, B.H. 1985. "On the Usefulness, Representation, and Validation of Non-Additive Probability Judgments for Risk Assessment." (in press). Weinstein, L.H.; Laurence, J.A. 1989. "Indigenous and Cultivated Plants as Bioindicators," in Biologic Markers of Air Pollution Stress and Damage in Forests. National Academy Press, Washington, DC. pp. 195-204. Whitfield, R.G.; Wallsten, T.S. 1984. Estimating Risks of Lead-Induced Haemoglobin Decrements Under Conditions of Uncertainty: Methodology, Pilot Judgments, and Illustrative Calculations. Prepared for the Office of Air Quality Planning and Standards, U.S. EPA. Wilson, R.A.; Mills, J.B.; Wituschek, E.P. 1984. A Report on the Assessment of Photochemical Oxidants in the Lower Mainland. B.C. Ministry of the Environment, Victoria, B.C. Winkler, R.L. 1981. "Combining Probability Distributions from Dependent Information Sources," Mgmt. Sci. 27, 4, 479. World Comission on Environment and Development 1987. Our Common Future, Oxford University Press. 400 pp. Wright, E.F. 1988. The Effect of Ozone on Horticultural Crops Important in the Fraser Valley of British Columbia. M.Sc. Thesis, Dept. of Plant Science, University of British Columbia. Zar, J.H. 1984. Biostatistical Analysis , Prentice-Hall, Inc., Englewood Cliffs, NJ. 244 APPENDIX A Questionnaire to Determine Attributes of an Expert in Ozone Effects on Crops 2 4 5 Q u e s t i o n n a i r e | FRASER VALLEY OZONE RISK ASSESSMENT PROJECT (BRITISH COLUMBIA, CANADA) WHAT IS AN EXPERT? WHAT ARE THE DIMENSIONS OF SCIENTIFIC EXPERTISE? WHAT CONDITIONS ARE IMPORTANT TO THE ABILITY TO MAKE INFORMED JUDGMENTS? T h i s q u e s t i o n n a i r e has been d e s i g n e d t o h e l p us d e f i n e the necessary and d e s i r a b l e dimensions of expertise i n a p a r t i c u l a r t o p i c area: ozone a i r p o l l u t i o n e f f e c t s and impacts on a g r i c u l -t u r a l crop p r o d u c t i o n . T h i s i n f o r m a t i o n w i l l be used i n the s e l e c t i o n of e x p e r t s c i e n t i s t s who w i l l be asked t o p r o v i d e r i s k e s t i m a t e s (judgments of ozone impacts on crop p r o d u c t i o n ) f o r p u b l i c p o l i c y making. Although the b a s i c c o n d i t i o n s of e x p e r t i s e i n v o l v e m a s t e r i n g a s p e c i f i c knowledge base through e d u c a t i o n and c a r e e r expe-r i e n c e , o t h e r i m p o r t a n t dimensions i n c l u d e c e r t a i n c o g n i t i v e s k i l l s , and c e r t a i n p e r s o n a l or p r o f e s s i o n a l q u a l i t i e s , s k i l l s or s t y l e s . Your response w i l l a s s i s t i n d e f i n i n g the v a r i o u s dimensions of e x p e r t i s e as w e l l as the r e l a t i v e importance of each. In a d d i t i o n , you are requ e s t e d t o p r o v i d e the names of persons who i n your o p i n i o n q u a l i f y as e x p e r t s i n ozone e f f e c t s on a g r i c u l t u r a l c r o p s . THANK YOU FOR YOUR COOPERATION AND OPINION ! 246 - 1 -1. _ What l e v e l of e d u c a t i o n s h o u l d an e x p e r t have who p r o v i d e s r i s k e s t i m a t e s o f ozone i m p a c t s on c r o p y i e l d s ? P l e a s e r a t e e ach o f t h e f o l l o w i n g l e v e l s o f e d u c a t i o n a c c o r d i n g t o t h e f o l l o w i n g s c a l e : E = VT = MI = e s s e n t i a l v e r y m o d e r a t e l y i m p o r t a n t i m p o r t a n t S I = s l i g h t l y i m p o r t a n t NI = n o t i m p o r t a n t P l e a s e c i r c l e one r e s p o n s e f o r each i t e m : B.S. d e g r e e o r e q u i v a l e n t E V I MI S I NI M.S. d e g r e e o r e q u i v a l e n t E V I MI SI NI Ph.D. d e g r e e o r e q u i v a l e n t E V I MI S I NI Completed p o s t - d o c t o r a t e o r e q u i v a l e n t E V I MI SI NI 2. What m a j o r d i s c i p l i n e s o r f i e l d s b a c k g r o u n d f o r an e x p e r t who p r o v i d e s i m p a c t s on c r o p y i e l d s ? o f s t u d y a r e an r i s k e s t i m a t e s i m p o r t a n t o f ozone P l e a s e r a t e t h e r e l a t i v e i m p o r t a n c e o f f o l l o w i n g d i s c i p l i n e s : competency i n e a ch o f t h e P l a n t S c i e n c e E V I MI SI NI B o t a n y E VI MI SI NI A t m o s p h e r i c S c i e n c e E V I MI SI NI M a t h e m a t i c s E V I MI SI NI S t a t i s t i c s E V I MI SI NI P h y s i c s E V I MI SI NI C h e m i s t r y E V I MI SI NI O t h e r E V I MI SI NI 3. I s t h e r e a s i n g l e a r e a of academic or t e c h n i c a l s p e c i a l i z a t i o n t h a t i s i m p o r t a n t t o e x p e r t i s e r e g a r d i n g ozone i m p a c t s on c r o p p r o d u c t i o n ? I f s o , p l e a s e d e s c r i b e t h i s a r e a of s p e c i a l i z a t i o n and r a t e i t s r e l a t i v e i m p o r t a n c e : E V I MI SI NI 247 TABLE 3-8 Comparison of LII and Crop Yields in 1986 UBC Experiment Site (Block) Potato Weight (grams) Pea Weight (grams) Seasonal Cumulative LLI (%) Seasonal Cumulative LLI (%) Control 1022 38 0.00 0.00 1-3 593 19 1.87 0.17 2-2 767 15 3.41 0.31 3-1 661 22 5.61 0.51 4-3 731 17 14.41 1.31 248 - 2 -4. How i m p o r t a n t i s academic p e r f o r m a n c e (as measured by g r a d e s ) w h i l e i n c o l l e g e o r u n i v e r s i t y ? P l e a s e r a t e t h e i m p o r t a n c e of a c h i e v i n g each of t h e f o l l o w i n g a v e r a g e g r a d e s : ( b e f o r e r a t i n g t h e s e i t e m s , p l e a s e r e f e r t o Box A below) E = V I = MI = S I = NI = e s s e n t i a l very m o d e r a t e l y s l i g h t l y not important important important important P a s s (50% - 64%, C) Second C l a s s ( 6 5 % - 79-F i r s t C l a s s (>80%, A) E V I MI SI NI B) E V I MI SI NI E VI MI SI NI BOX A A p a r t i c u l a r v a r i a b l e o r i t e m of s p e c i f i c i m p o r t a n c e c a n be p r e c e d e d by, b u t s h o u l d n o t be f o l l o w e d by, a v a r i a b l e o f g r e a t e r i m p o r t a n c e . Example: V a r i a b l e 1 @ VI MI SI NI ) V a r i a b l e 2 E ^ MI S I NI ) i s OK V a r i a b l e 3 E VI (MTJ SI NI ) W h i l e : V a r i a b l e 1 E VI © SI NI ) V a r i a b l e 2 E © MI SI NI ) i s not V a r i a b l e 3 @ VI MI S I NI ) l o g i c a l 5. How i m p o r t a n t i s length of career experience ( i n c l u d i n g g r a d u -a t e w o r k ) ? P l e a s e r a t e t h e r e l a t i v e i m p o r t a n c e of s p e n d i n g t h e f o l l o w i n g number of c a r e e r y e a r s i n v e s t i g a t i n g t h e e f f e c t s o f ozone on c r o p s : ( b e f o r e r a t i n g t h e s e i t e m s p l e a s e r e f e r t o Box A above) 5 y e a r s o r l e s s . 6 t o 10 y e a r s 11 t o 20 y e a r s 21 y e a r s o r more E V I MI SI NI E VI MI SI NI E V I MI SI NI E V I MI SI NI 249 - 3 -6. What types of c a r e e r experience s h o u l d an expert have who p r o v i d e s r i s k e s t i m a t e s of ozone impacts on crop p r o d u c t i o n ? Please r a t e the r e l a t i v e importance of each of the f o l l o w i n g types of c a r e e r e x p e r i e n c e : (Please note: Box A does not apply to t h i s q u e s t i o n which c o n t a i n s independent items) e s s e n t i a l VI = v e r y important MI = moderately important P a r t i c i p a t e d i n experiments d e a l i n g w i t h ozone e f f e c t s on crops Supervised experiments d e a l i n g with ozone e f f e c t s on crops Studied the m a j o r i t y of p u b l i s h e d a r t i c l e s d e a l i n g with ozone e f f e c t s on crops Served as a peer reviewer i n e v a l u a t i n g r e s e a r c h r e g a r d i n g ozone e f f e c t s on crops SI = s l i g h t l y i m p ortant NI = not important VI MI SI NI VI MI SI NI VI MI SI NI VI MI SI NI Conducted s t a t i s t i c a l a n a l y s i s r e l a t e d to d e f i n i n g p l a n t dose-response models Authored s c i e n t i f i c papers d e a l i n g with ozone e f f e c t s on crops VI MI SI NI VI MI SI NI Peer-reviewed s c i e n t i f i c papers submitted to j o u r n a l s d e a l i n g w i t h ozone e f f e c t s E VI MI SI NI Held a t e a c h i n g appointment at a post-secondary e d u c a t i o n a l i n s t i t u t i o n R e g u l a r l y attended conferences and workshops d e a l i n g w i t h ozone e f f e c t s R e g u l a r l y gave s p e e c h e s / l e c t u r e s d e a l i n g w i t h ozone e f f e c t s on crops Appeared as an expert witness t o pr o v i d e testimony d e a l i n g w i t h ozone e f f e c t s E VI MI SI NI E VI MI SI NI E VI MI SI NI E VI MI SI NI Conducted r e s e a r c h p r i m a r i l y i n a u n i v e r s i t y s e t t i n g Conducted r e s e a r c h p r i m a r i l y f o r p r i v a t e i n d u s t r y Conducted r e s e a r c h p r i m a r i l y f o r government VI MI SI NI VI MI SI NI VI MI SI NI 250 - 4 -E = VI = MI = SI = NI = e s s e n t i a l very moderately s l i g h t l y • not important important important important Other types of career experience (please l i s t ) : E VI MI SI NI E VI MI SI NI E VI MI SI NI 7. How important i s frequency of c o n t r i b u t i o n s to the l i t e r a t u r e ? Please rate the r e l a t i v e importance of c o n t r i b u t i n g to the f o l l o w i n g number of p u b l i c a t i o n s as author or co-author i n the l a s t ten (10) years. Note: Before r a t i n g these items, please r e f e r to Box A on page 2. 1 - 5 p u b l i c a t i o n s E VI MI SI NI 6 - 1 0 p u b l i c a t i o n s E VI MI SI NI 11 - 20 p u b l i c a t i o n s E VI MI SI NI more than 20 p u b l i c a t i o n s E VI MI SI NI 8. Are there any important p r o f e s s i o n a l o r g a n i z a t i o n s or t i o n s that an expert i n ozone impacts on crops should belong a s s o c i a -te? Please l i s t these below and rate the r e l a t i v e importance of each: E VI MI SI NI E VI MI SI NI E VI MI SI NI E VI MI SI NI 251 - 5 -9.' Is a f f i l i a t i o n with a p a r t i c u l a r r e s e a r c h f a c i l i t y or program important to the development of ex p e r t i s e i n ozone impacts on crops? Please l i s t s p e c i f i c , outstanding research f a c i l i t i e s or programs de a l i n g with ozone e f f e c t s on crops, that are l i k e l y to produce top experts i n t h i s f i e l d : F a c i l i t y / . Program #1 F a c i l i t y / Program #2 F a c i l i t y / Program §3 F a c i l i t y / Program #4 F a c i l i t y / Program #5 10. Are there any a d d i t i o n a l symbols of r e c o g n i t i o n , c r e d e n t i a l s , d i s t i n c t i o n s , appointments or t i t l e s that an expert i n ozone impacts on crops should hold? Please l i s t these below and r a t e the r e l a t i v e importance of each: E = VI = MI = S I = NT = e s s e n t i a l very moderately s l i g h t l y not important important important important E VI MI SI NI E VI MI SI NI E VI MI SI NI E VI MI SI NI 252 - 6 -11. For p u b l i c p o l i c y making, experts sometimes are asked to provide judgments i n v o l v i n g u ncertain s c i e n t i f i c r e l a t i o n s h i p s . How important are c e r t a i n c o g n i t i v e or mental s k i l l s to the a b i l i t y to provide r i s k estimates (judgments) of ozone impacts on crops? Please r a t e the importance of an expert possessing an above average  l e v e l of each of the f o l l o w i n g c o g n i t i v e s k i l l s or s t y l e s . (Please note: Box A on page 2 does not apply to t h i s question.) E = VT = MT = SI = NI = e s s e n t i a l v e r y moderately s l i g h t l y not important important important important General i n t e l l i g e n c e and/or i n t e l l e c t • E VI MI SI NI Mental quickness E •VI MI SI NI Memory E VI MI SI NI Imagination and c r e a t i v i t y E VI MI SI NI A b i l i t y to ignore i r r e l e v a n c e E VI MI SI NI A b i l i t y to do i n d u c t i o n ( i n f e r a general c o n c l u s i o n from p a r t i c u l a r instances) E VI MI SI NI A b i l i t y to focus a t t e n t i o n E VI MI SI NI Decisiveness E VI MI SI NI Consistency i n judgments and d e c i s i o n s E VI MI SI NI A b i l i t y to express i n t e r n a l u n c e r t a i n t y (ie 'think p r o b a b i l i s t i c a l l y ' ) E VI MI SI NI Foresightedness E VI MI SI NI Independent, autonomous t h i n k i n g a b i l i t y E VI MI SI NI Independent, autonomous l e a r n i n g a b i l i t y E VI MI SI NI Alertness i n terms of observing relevant information or cues E VI MI SI NI Judgmental a b i l i t y (as rated by peers) E VI MI SI NI A b i l i t y to i n t e g r a t e diverse opinion E VI MI SI NI M a t h e m a t i c a l / s t a t i s t i c a l a b i l i t y E VI MI SI NI 253 - 7 -E = V I = M I = S I = NI = e s s e n t i a l v e r y moderately s l i g h t l y not important important important important C u r i o s i t y , i n q u i s i t i v e n e s s E VI MI SI NI L i k i n g f o r method, p r e c i s i o n , exactness E VI MI SI NI L i k i n g f o r a b s t r a c t t h i n k i n g E VI MI SI NI L i k i n g f o r i n t e l l e c t u a l argument and debate E VI MI SI NI Other c o g n i t i v e or mental s k i l l s (please l i s t ) : E VI MI SI NI E VI MI SI NI VI MI SI NI 254 - 8 -1 2 . Whether or not risk estimates by experts are accepted for public policy making may depend upon the c red i b i l i t y ' o f the experts, i . e . , whether the expert i s trustworthy or believable. Please rate the importance of an expert possessing an above average level of each of the following personal qual it ies as i t relates to c r ed i b i l -i t y . (Box A does not apply to this question.) E= VI = MI = S I= NI = essential very moderately s l i gh t l y not important important important important Forthrightness (to be direct and straight forward) E VI MI SI NI Willingness to admit ignorance of a specif ic topic or problem E VI MI SI NI Honesty E VI MI SI NI Time spent working overtime E VI MI SI NI Ab i l i t y to speak (articulateness) E VI MI SI NI Ab i l i t y to convey knowledge or information (communicate) E VI MI SI NI Satisfaction and pride in career E VI MI SI NI Desire to discover s c ien t i f i c truth E VI MI SI NI Sympathy or concern for problems underlying work/research E VI MI SI NI Interest in work/research E VI MI SI NI Conscientiousness E VI MI SI NI Open-mindedness E VI MI SI NI Willingness to reject group pressures toward conformity in thinking E VI MI SI NI Emotional s t ab i l i t y E VI MI SI NI Conservativeness (in data interpretation) E VI MI SI NI Confidence E VI MI SI NI Modesty E VI MI SI NI Friendliness E VI MI SI NI Constructiveness (in reviewing work of others) E VI MI SI NI 255 - 9 -E = VI = MI = SI = NI = e s s e n t i a l very moderately s l i g h t l y not important important important important Other personal q u a l i t i e s (please l i s t ) : E VI MI SI NI E VI MI SI NI E VI MI SI NI 25fi - 10 -13. RELATIVE IMPORTANCE OF MAJOR DIMENSIONS You have rated the importance of s e v e r a l items w i t h i n each of se v e r a l major dimensions i n the questionnaire. At t h i s time you are asked to rate the importance between these major dimensions, so that weights can be d e r i v e d f o r each. Please a l l o c a t e 100 point s between the 5 major dimensions given below i n Box B. BOX B Dimension Points Assigned Education ( l e v e l and d i s c i p l i n e ) Career experience (eg. type, number of experiments) S c i e n t i f i c r e c o g n i t i o n ( p u b l i c a t i o n s ) Cognitive and mental s k i l l s (eg. judgmental a b i l i t y ) Personal q u a l i t i e s r e l a t e d t o c r e d i b i l i t y (eg. communication a b i l i t y ) TOTAL 100 257 - 11 -14. NOMINATION OF EXPERTS P l e a s e l i s t the names and a f f i l i a t i o n s ( i f known) of up t o 5 persons who i n your o p i n i o n q u a l i f y as top e x p e r t s i n ozone impacts on crop s . I n your o p i n i o n these persons should be the most q u a l i f i e d e x p e r t s you are aware o f , capable of p r o v i d i n g r i s k e s t i m a t e s (judgments) of probable y i e l d r e d u c t i o n s of crops i n the F r a s e r V a l l e y of B r i t i s h Columbia, r e s u l t i n g from observed ozone l e v e l s . You may nominate y o u r s e l f as an ex p e r t . Expert #1: Name: A f f i l i a t i o n E xpert #2: Name: A f f i l i a t i o n E xpert #3: Name: A f f i l i a t i o n E xpert H : Name: A f f i l i a t i o n E x p e r t #5: Name: A f f i l i a t i o n 258 - 12 15. RESPONDENT INFORMATION ^ Please provide a few items of information about yourself Please check (V) the appropriate items below: Education: BS degree MS degree PhD degree Other (see note 1) Years Experience: < 5 years 6 to 10 years 11 to 20 years more than 20 years Work Description: Sc ient i f ic Research Consulting Policy/Regulations Enforcement Administration Teaching Other (see note 4) Discipl ine: Botany, Plant Science Other Biology Other Science Social Science Other (see note 2) Location: Canada U.S.A. Other (see note 3) Employer: University Government Consulting Industry Other (see note 5) Notes: Please print other education status below: Please print other d isc ip l ine below: Please print other country below: Please print other work description below: Please print other employer below: *************************************************************** If you have any suggestions or comments, please write them on the back of this page. 259 APPENDIX B Crop Loss Questionnaire 260 Fraser V a l l e y Risk Assessment Project Crop Loss Questionnaire (Part A) Instructions and Information Prepared by Gordon L. Brown Resource Ecology U n i v e r s i t y of B r i t i s h Columbia 2204 Main Mall Vancouver, B.C. V6T 1W5 261 Fraser Valley Risk Assessment Project Crop Loss Questionnaire Instructions and Information (a) Crops to be Assessed The crops, c u l t i v a r s and growing p e r i o d s to be assumed i n the • Fraser V a l l e y Risk Assessment P r o j e c t are as f o l l o w s : Crop C u l t i v a r ( s ) l Green snap bean Bush Blue Lake 290 (Provider) (Executive) Potato Pea Corn B r o c c o l i Raspberry Forage Russet Burbank Aka Netted Gem (Norgold Russet) Olympia (Progress No. 9) (Green Arrow) J u b i l e e Emperor (Premium crop) W i l l i a m e t t e (Skeena) Mixture of orchard grass (60-70%) - p e r e n n i a l rye grass (30-40%) Approximate P l a n t i n g Date l a t e May -e a r l y June l a t e A p r i l -e a r l y May l a t e A p r i l -e a r l y May l a t e A p r i l -e a r l y May mid-May end A p r i l e a r l y May P e r e n n i a l Approximate Harvest Date l a t e J u l y l a t e August -e a r l y September e a r l y August l a t e J u l y -e a r l y August l a t e August l a t e J u l y -e a r l y August e a r l y J u l y mid Throughout season Less popular c u l t i v a r s are enclosed w i t h i n brackets. 262 - 2 -(b) R e l a t i v e S e n s i t i v i t y of Crops You are i n i t i a l l y asked to judge the r e l a t i v e s e n s i t i v i t y to ozone p o l l u t i o n , i n terms of p o t e n t i a l y i e l d r e d u c t i o n , of the F r a s e r V a l l e y crops l i s t e d above. The r a t i n g s heet f o r t h i s t a s k i s i n c l u d e d i n P a r t B: Answer and Assessment B o o k l e t , as " P a r t I : R e l a t i v e S e n s i t i v i t y of Crops to Ozone". Space i s l e f t a t the bottom of the s e n s i t i v i t y r a t i n g sheet to s t a t e assumptions or c o n d i t i o n s r e l a t e d t o your judgments r e g a r d i n g r e l a t i v e s e n s i t i v i t y of the c r o p s to ozone, i f a p p r o p r i a t e . (c) Ozone S c e n a r i o s and A s s o c i a t e d Crop Y i e l d Loss E s t i m a t e s In " P a r t I I : Crop Y i e l d Loss E s t i m a t e s " , you are asked to e s t i m a t e p r o b a b i l i t i e s of y i e l d r e d u c t i o n s f o r each of the above crops under t h r e e l e v e l s (low, medium, high) of ozone p o l l u t i o n . Summarized ozone and c l i m a t o l o g i c a l s t a t i s t i c s and d e f i n i t i o n s f o r each of these t h r e e ozone s c e n a r i o s are a t t a c h e d i n Appendix A: Ozone S c e n a r i o s . P r o d u c t i o n guide recommendations f o r each crop (growing season, f e r t i l i z e r , e t c . ) are a t t a c h e d as Appendix B: P r o d u c t i o n Guide f o r Commercial Growers. C o n s i d e r a b l e r e s e a r c h has been conducted by p s y c h o l o g i s t s and d e c i s i o n a n a l y s t s on v a r i o u s methods of e l i c i t i n g s u b j e c t i v e p r o b a b i l i t i e s and t h e r e i s no consensus on which method i s s u p e r i o r . G e n e r a l r u l e s of thumb, however, i n c l u d e sugges-t i o n s t h a t the method used s h o u l d be e a s i l y comprehended by those p r o v i d i n g the judgments, and t h a t the method u t i l i z e d s h o u l d a l l o w r e a s o n a b l y a c c u r a t e a p p r o x i m a t i o n of the con-t i n u o u s s u b j e c t i v e p r o b a b i l i t y d i s t r i b u t i o n s r e p r e s e n t i n g the judgments. On the b a s i s of the above, the t h r e e - p o i n t a p p r o x i m a t i o n of Pearson and Tukey, as proposed by K e e f e r and B o d i l y (1983) has been s e l e c t e d f o r the F r a s e r V a l l e y Ozone R i s k Assessment P r o j e c t . You are asked to p r o v i d e t h r e e c r o p l o s s e s t i m a t e s due to ozone f o r each crop type and each ozone s c e n a r i o . 263 - 3 -The three estimates you provide for each crop and ozone scenario should correspond to the .05, .50 and .95 f r a c t i l e s of your subjective p r o b a b i l i t y d i s t r i b u t i o n of crop loss. The meaning of " t r a c t i l e " , and the method by which crop loss estimates should be provided, i s described below. The .5 f r a c t i l e i s simply the median of your subjec-tive probability distribution. You are asked to provide an estimate of crop loss due to ozone (say you estimate 20 units out of a hypothetical maximum yie l d of 100 units) such that the true value of crop loss i s equally l i k e l y to f a l l below or above this value. In other words, there i s an equal chance that the true y i e l d loss due to ozone i s higher, or lower, than 20 units. The two other f r a c t i l e s are estimated i n a s i m i l a r fashion. For the .05 f r a c t i l e you are asked to provide a crop loss estimate such that the probability that the true value f a l l s below the selected one i s .05 (or five percent). For example, you might select a loss value of 2 units for the .05 f r a c t i l e . What this means i s that, i n your opinion, the chance of y i e l d loss being less than or equal to 2 units i s 5%. Conversely, the chance of yiel d loss being greater than 2 units, i n your opinion, i s 95%. For the .95 f r a c t i l e you might choose a loss estimate of 30 units, meaning that in your opinion, there i s a 95% chance that y i e l d loss w i l l be 30 units or less, while there i s a 5% chance that y i e l d loss w i l l exceed 30 units. Summary of F r a c t i l e Method In the example above various crop loss estimates were provided using the f r a c t i l e method, for one crop type and one ozone condi-t i o n . For the purpose of t h i s example, assume these estimates were provided for green snap bean under the high ozone scenario. Similar estimates would be provided for the low and medium ozone scenarios, and the r e s u l t s would be reported on one s i n g l e answer sheet contained i n the Part B booklet. The next page i s an example of how t h i s answer sheet should be f i l l e d out, based on the above example. Any notes, conditions or assumptions associated with provision of these judgments should be written on the back of the answer sheet. 264 - 4 - ' , . PART I I ; Crop Y i e l d Loss E s t i m a t e s Crop Type (C i rc le one): ^ Green bean Potato Pea Corn Brocco l i Raspberry Forage Low Ozone S c e n a r i o 1 7 1, F rac t i l e Crop Loss Estimate V a l i d i t y Check •05 / .95 • 5 0 f . 50 .95 ?j .05 Medium Ozone S c e n a r i o 1 2 3 F rac t i l e Crop Loss Estimate" V a l i d i t y Check • 0 5 / S -95 .50 -50 .95 : ' . 05 High Ozone S c e n a r i o 1 2 3 F rac t i l e Crop Loss Estimate V a l i d i t y Check .05 A .95 .50 ^ 0 .50 • i .95 ' ~' .05 Probab i l i ty that true crop loss i s less than or equal to crop loss estimate. 2 Given i n un i t s , based on a maximum hypothetical y i e l d of 100 units . 3 Probab i l i ty that true crop loss i s greater than crop loss estimate. Note: P l e a s e i n c l u d e notes, assumptions o r c o n d i t i o n s on r e v e r s e s i d e of t h i s s heet. 2 6 5 - 5 -(d) Information Used i n Decision-Making In Part III of the answer booklet, "Dose S t a t i s t i c s and Other Cues U t i l i z e d " , you are asked to provide the primary informa-t i o n you used f o r your crop loss estimate decisions. These w i l l be p r i m a r i l y , and perhaps e x c l u s i v e l y , r e l a t e d to one or more of the ozone dose s t a t i s t i c s summarizing the three ozone scenarios. For example, i f you used the Mean7 (M7) s t a t i s t i c p r i m a r i l y ; please write "M7" i n the space provided. If you used more than one s t a t i s t i c for your crop loss d e c i s i o n , please i n d i c a t e which s t a t i s t i c s were used. A d d i t i o n a l l y , i f you used d i f f e r e n t information for loss estimates of d i f f e r e n t crops, i t would be appreciated i f t h i s information was provided. (e) Respondent Information In "Part IV: Respondent Information", you are asked to provide summary information about y o u r s e l f , which w i l l help us to characterize your expert a t t r i b u t e s , and compare them with others. This information w i l l , of course, be treated c o n f i d e n t i a l l y , unless you indicate otherwise (space for t h i s i s provided at the end of the questionnaire). F i n a l Notes and Precautions (1) Several v a r i a b l e s i n addition to ozone can a f f e c t crop loss estimates. T y p i c a l farming p r a c t i c e s may d i f f e r i n the Fraser Valley from p r a c t i c e s common i n other regions. This i s a d i f f i c u l t problem to deal with i n the r i s k assessment project, as wide ranging background conditions that i n t e r a c t with ozone's e f f e c t on crop y i e l d may be assumed by d i f f e r e n t judges (experts). I t i s hoped that by i n c l u d i n g summarized c l i m a t o l o g i c a l data and recommended agronomic methods (produc-t i o n guide information), judges w i l l assume s i m i l a r background conditions. Space i s provided i n the questionnaire to describe any assumptions or conditions associated with your crop loss estimates. (2) Research i n subjective p r o b a b i l i t y e l i c i t a t i o n has shown that e l i c i t e d d i s t r i b u t i o n s are often too " t i g h t " , meaning an unduly large proportion of true values (e.g. actual crop losses) f a l l i n t o the extreme t a i l s of the assessed d i s t r i b u -tions . These r e s u l t s may be due to a well recognized cognitive bias referred to as "anchoring and adjustment" (Tversky and Kahneman, 1974). These cognitive psychologists hypothesize that, when a subject i s asked for values corresponding to s p e c i f i c f r a c t i l e s , he or she f i r s t "anchors" on the value 266 - 6 -f i r s t estimated (e.g. the median), and then "adjusts'" that value i n the d i r e c t i o n appropriate f o r the given f r a c t i l e . The adjustment process w i l l , however, usually be i n s u f f i c i e n t , thus leading to "too t i g h t " d i s t r i b u t i o n s . I t i s , again, d i f f i c u l t to deal with t h i s problem, however, i t i s hoped that, by making you aware of t h i s cognitive bias, you may i n f a c t p a r t i a l l y compensate for i t when providing subjective p r o b a b i l i t i e s . (3) On the other hand, your uncertainty regarding the dose-response r e l a t i o n s h i p w i l l be portrayed by the "tightness" of • your crop loss d i s t r i b u t i o n s . I f you are r e l a t i v e l y confident of the dose-response r e l a t i o n s h i p , the .05 and .95 f r a c t i l e s w i l l be r e l a t i v e l y close to the median. Conversely, i f you are very uncertain about the dose-response r e l a t i o n s h i p , the .05 and .95 f r a c t i l e s w i l l be r e l a t i v e l y widely separated from the median. Thus, although you may f e e l very unsure about the r e l a t i o n -ship, you are s t i l l encouraged to attempt to portray your uncertainty, through provision of the f r a c t i l e s . Demonstra-t i o n of s i g n i f i c a n t uncertainty w i l l c l e a r l y demonstrate the need for a d d i t i o n a l s c i e n t i f i c experimentation by plant s c i e n t i s t s to determine the actual dose-response r e l a t i o n -ships . (4) As a f i n a l check to ensure your judgments are consistent, the r e l a t i v e s e n s i t i v i t y ratings provided i n Part I of the questionnaire should be compared with the judgments provided i n Part I I . Presumably the greatest y i e l d reductions should be predicted for the most s e n s i t i v e crops, and so on. I f t h i s i s not the case, adjustments should be made to ensure the judgments are consistent. References Keefer, D.L. and S.E. Bodily (1983). Three-point approximations for continuous random va r i a b l e s . Management Science Vol. 29 No. 5 May 1983. Tversky, A. and D. Kahneman (1974). Judgment under uncertainty: h e u r i s t i c s and biases. Science 185, 1124-1131. 267 Appendix A: Ozone Scenarios Note: This is Appendix A of the crop loss questionnaire. 268 Definition of Summary Ozone St a t i s t i c s Mean24: season-long mean of d a i l y mean Meanl2: season-long mean of d a i l y 12-hr mean (9 am to 9 pm) Mean7: season-long mean of d a i l y 7-hr mean (9 am to 4 pm) Meanl: season-long mean of d a i l y 1-hr max. (9 am to 4 pm) Peakl: peak 1-hr mean observed i n season Peak7: peak 7-hr mean observed i n season Total ozone dose: t o t a l ozone dose i n ppb-hr observed i n season, followed by an adjusted value computed by mul t i p l y i n g by a factor to account for hours of missing data i n the season (e.g. x 1.026 for Chilliwack, x 1.156 for Abbotsford, and x 1.124 for Anmore) Days 80, 100, 120: number of days i n the season i n which ozone exceeded 80, 100 and 120 ppb one-hour average, respec-t i v e l y . Hours 40 ... Hours 100: number of hours i n the season when the ozone concentration was greater than or equal to 40 ppb, ... 100* ppb. Sum 40 ... Sum 100: sums of the f r a c t i o n s of a l l ozone one-hour concentrations exceeding thresholds of 40 ppb ... 100 ppb. (Designed to give greater weight to peak events (e.g. Oshima, 1974)). E f f e c t i v e Mean7 and 12: also formulated i n an attempt to give greater weight to peak concentrations. Computed by r a i s i n g the hourly ozone concentration to a power p r i o r to summing the ozone concentrations and d i v i d i n g by the number of observations as follows (Larson and Heck 1984): EFFMEAN = [(E ozone(t)**2.6)/N]**0.376 A - 1 269 LOW OZONE CONDITION CHILLIWACK, 1986 Ozone Monitor T12 Summary ozone s t a t i s t i c s f o r the period May 1 to August 31: Mean24 (M24) 17. 1 (Ppb) Meanl2 (M12) 26. 4 (PPb) Mean7 (M7) 28. 2 (Ppb) Meanl (Ml) 38. 2 (PPb) Peakl (PI) 84. 0 (PPb) Peak7 (P7) 67. 0 (PPb) Total ozone dose 49255 ppb-hr Days80 2 days DayslOO 0 days Days120 0 days Hours40 213 hours Hours50 106 hours Hours60 40 hours Hours70 13 hours Hours80 2 hours Hours90 0 hours HourslOO 0 hours Sum40 11053 ppb-hr Sum50 6311 ppb-hr Sum60 2750 ppb-hr Sum70 990 ppb-hr Sum80 165 ppb-hr Sum90 0 ppb-hr SumlOO 0 ppb-hr E f f e c t i v e mean-7 15. 3 PPb E f f e c t i v e mean-12 14. 9 PPb 50538 (adjusted for missing data) Summary ozone s t a t i s t i c s by month: Mean24 Meanl2 Mean7 Meanl Peakl Peak7 (M24) (M12) (M7) (Ml) (PI) (P7) May June July August 20.2 19.8 12.7 15.8 28.0 29.8 20.0 28.1 30.0 31.4 21.4 30.3 38.4 41.3 29.5 43.9 84.0 81.0 56.0 75.0 54.9 67.0 45.9 48.3 A - 2 270 17 AUG 88 CHILL 86 14:48:55 U n i v e r s i t y of B r i t i s h Columbia OZONE COUNT MIDPOINT ONE SYMBOL EQUALS APPROXIMATELY 16.00 OCCURRENCES 686 .002 * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 336 .006 * * * * * * * * * * * * * . * * * * * * * 254 .010 * * * * * * * * * * * * * * * * 264 .014 * * * * * * * * * * * * * * * * * 223 .018 * * * * * * * * * * * * * * 201 .022 * * * * * * * * * * * * * 216 .026 * * * * * * * * * * * * * * 196 .030 * * * * * * * * * * * * < 143 .034 * * * * * * * * * 120 .038 * * * * * * . * 72 .042 * * * * . 43 .046 * * . 35 .050 * . 27 .054 . * 18 .058 : 10 .062 * 15 .066 * 5 .070 7 .074 4 .078 2 .082 I + 1 + ... .1 + I . . . . + .... I + I 0 160 320 480 640 800 HISTOGRAM FREQUENCY HISTOGRAM OF OZONE ONE-HOUR AVERAGE CONCENTRATIONS LOW OZONE CONDITION 271 MEDIUM OZONE CONDITION ABBOTSFORD, 1985 Ozone Monitor T i l Summary ozone s t a t i s t i c s f o r t h e p e r i o d May 1 t o August 31: Mean24 (M24) 24. 5 (ppb) Meanl2 (M12) 35. 8 (ppb) Mean7 (M7) 38. 6 (PPb) Meanl (Ml) 48. 1 (ppb) P e a k l ( P I ) 107. 0 (ppb) Peak7 (P7) 83. 4 (PPb) T o t a l ozone dose 62590 ppb-hr Days80 5 days Days100 1 day D a y s l 2 0 0 day Hours40 459 h o u r s Hours50 239 h o u r s Hours60 85 h o u r s Hours70 29 hour s Hours80 14 ho u r s Hours90 10 h o u r s HourslOO 1 hour Sum40 23973 ppb-hr Sum50 14287 ppb-hr Sum60 5928 ppb-hr Sum70 2404 ppb-hr Sum80 1274 ppb-hr Sum90 938 ppb-hr SumlOO 107 ppb-hr E f f e c t i v e mean-7 14. 87 PPb E f f e c t i v e mean- 12 14. 09 PPb 72324 ( a d j u s t e d f o r m i s s i n g d a t a ) Summary ozone s t a t i s t i c s by month: May June J u l y August Mean24 (M24) 27.9 25.2 24.5 16.1 Meanl2 (M12) 36.9 35.5 39.6 24.7 Mean7 (M7) 38. 5 37.7 44.2 27.8 Meanl (Ml) 47.8 45.9 56.4 38 . 0 P e a k l ( P I ) 92.0 75.0 107.0 57.0 Peak7 (P7) 71. 2 59.9 83.4 38.7 A - 3 272 17 AUG 88 ABBOTSFORD 85 15:02:09 U n i v e r s i t y of B r i t i s h Columbia OZONE COUNT MIDPOINT 0 -.0200 0 -.0125 0 -.0050 755 .0025 * * 406 .0100 * * 450 .0175 * * 410 .0250 * * 459 .0325 ** 295 .0400 * * 187 .0475 * * 116 • .0550 * * 62 .0625 * * 11 .0700 10 .0775 * 4 .0850 8 .0925 * 1 .1000 1 .1075 0 .1150 0 .1225 0 .1300 ONE SYMBOL EQUALS APPROXIMATELY 16.00 OCCURRENCES * * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * . . . .1 + I. ... + ... .1, 160 320 480 HISTOGRAM FREQUENCY , . . I 640 . . . I 800 VALID CASES 3175 MISSING CASES 0 HISTOGRAM OF OZONE ONE-HOUR AVERAGE CONCENTRATIONS MEDIUM OZONE CONDITION ^ / ^ ^ f JfJ 273 HIGH OZONE CONDITION ANMORE, 1981 Ozone Monitor T7 Summary ozone s t a t i s t i c s f or the period May 1 to August 31: Mean24 (M24) 31. 2 (ppb) Meanl2 (M12) 47. 3 (ppb) Mean7 (M7) 53. 8 (ppb) Meanl (Ml) 70. 8 (ppb) Peakl (PI) 217. 0 (ppb) Peak7 (P7) 160. 8 (ppb) Tot a l ozone dose 81863 ppb-hr Days80 33 days DayslOO 23 days Daysl20 16 days Hours40 646 hours Hours50 422 hours Hours60 295 hours Hours7 0 221 hours Hours80 161 hours Hours90 127 hours HourslOO 100 hours Sum40 44307 ppb-hr Sum50 34390 ppb-hr Sum60 27568 ppb-hr Sum70 22763 ppb-hr Sum80 18340 ppb-hr Sum90 15464 ppb-hr SumlOO 12917 ppb-hr E f f e c t i v e mean-7 17. 39 ppb E f f e c t i v e mean-12 17. 17 ppb 92013 (adjusted for missing data) Summary ozone s t a t i s t i c s by month: May June July August. Mean24 (M24) 28. 2 27.4 30.7 33 . 3 Meanl2 (M12) 37.8 37.9 46.8 52.9 Mean7 (M7) 40.5 49.9 53.7 64.2 Meanl (Ml) 50. 6 52.8 72.6 87. 4 Peakl (PI) 97.0 120.0 204.0 188.0 Peak7 (P7) 80.3 94.0 109.7 135.3 A - 4 274 17 AUG 88 ANMORE 81 15:02:27 U n i v e r s i t y of B r i t i s h Columbia COUNT MIDPOINT ONE SYMBOL EQUALS APPROXIMATELY 24.00 OCCURRENCES 0 -.020 108 -.005 * * * * * 1184 .010 * * * * * * * * * * * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * * * * * * * 928 .025 * * * * * * * * * * * * * * * * * * * * * * * * * * * ; * * * * * * * * * * * 524 .040 * * * * * * * * * * * * * * * * * * * * * * 244 .055 * * * * * * * * * * 125 .070 * * * * * 69 .085 * * * 43 .100 * . 41 .115 * * 20 .130 * 17 .145 * 11 .160 9 .175 1 .190 3 .205 2 .220 1 .235 0 .250 1 .265 0 .280 .. .1 + I + I. 240 480 720' HISTOGRAM FREQUENCY , . .1, 960 + .... I 1200 VALID CASES 3331 MISSING CASES HIGH OZONE CONDITION 275 C l i m a t o l o g i c a l Summaries Associated with Ozone Scenarios M e a n T e m p . M a y M e a n T e m p . J u n e M e a n T e m p . J u l y M e a n T e m p . A u g u s t L o w O z o n e C h i l l i w a c k 1 9 8 6 1 4 ° C 1 8 ° C 1 7 ° C 2 1 ° C M e d i u m O z o n e A b b o t s f o r d 1 9 8 5 1 2 ° C 1 4 " C 2 0 ° C 1 7 ° C H i g h O z o n e A n m o r e 1 9 8 1 1 2 ° C 1 3 ° C 1 7 ° C 2 0 ° C • T o t a l P r e c i p . M a y T o t a l P r e c i p . J u n e T o t a l P r e c i p . J u l y T o t a l P r e c i p . A u g u s t 1 4 2 mm 6 1 mm 9 1 mm 3 mm 7 9 mm 6 6 mm 3 mm 1 3 mm 1 0 2 mm 2 0 3 mm 4 3 mm 5 6 mm S u n s h i n e M a y S u n s h i n e J u n e S u n s h i n e J u l y S u n s h i n e A u g u s t 2 0 3 h r 2 2 6 h r 2 2 5 h r 3 1 3 h r 2 0 7 h r 2 5 8 h r 3 9 4 h r 2 8 4 h r 1 6 1 h r 1 3 3 h r 2 3 7 h r 3 1 9 h r S o u r c e : E n v i r o n m e n t C a n a d a , A t m o s p h e r i c E n v i r o n m e n t S e r v i c e , A - 5 276 Tobacco B e l W-3 I n j u r y Tobacco B e l W-3 biomonitoring s i t e s c o n s i s t i n g of four p l a n t s were e s t a b l i s h e d i n the v i c i n i t y of the low ozone c o n d i t i o n s i t e ( C h i l l i w a c k 1986) and the medium ozone c o n d i t i o n s i t e (Abbotsford 1985). Biomonitoring work was done under the d i r e c t i o n of Dr. V i c Runeckles, UBC Department of Plant Science. No biomonitoring with tobacco was done before t h i s time, thus no data i s a v a i l a b l e f o r the high ozone c o n d i t i o n (Anmore 1981). Percent l e a f i n j u r y due to ozone was assessed v i s u a l l y at approximately one week i n t e r v a l s . Tobacco p l a n t s were replaced approximately every 4 to 5 weeks. I n j u r y was summarized as f o l l o w s : T o t a l l e a f i n j u r y (TLI) = E a l l i n j u r y observed i n a l l leaves during the weekly exposure p e r i o d . Leaf i n j u r y index (LII) = 2 a l l i n j u r y observed i n weekly exposure p e r i o d Exposure periods and l e a f i n j u r c o n d i t i o n s s i t e s are as f o l l o w s : Abbotsford 1985 (Medium Ozone Condition) Exposure Period TLI(%) LH (%) J u l y 11 to J u l y 17 0.0 0. 0 J u l y 18 to J u l y 24 35.0 73 J u l y 25 to Aug. 1 4.0 07 Aug. 2 to Aug. 12 1.0 02 Aug. 13 to Aug. 19 5.0 07 Aug. 20 to Aug. 28 33.0 43 Aug. 29 to Sept. 2 0.0 0. 0 t o t a l it of leaves i e s f o r the low and medium ozone C h i l l i w a c k 1986 (Low Ozone Condition) Exposure P e r i o d TLI LII May 25 to May 31 10. 0 0. 31 June 1 to June 6 0. 0 0. 0 June 7 to June 12 0 . 0 0. 0 June 13 to June 20 0. 0 0. 0 June 21 to June 26 0 . 0 0 . 0 June 27 to J u l y 4 0 . 0 0. 0 J u l y 5 to J u l y 10 0. 0 0. 0 J u l y 11 to J u l y 25 0. 0 0. 0 J u l y 26 to J u l y 31 0 . 0 0. 0 Aug. 1 to Aug. 7 0. 0 0. 0 Aug. 8 to Aug. 14 6. 0 0. 15 Aug. 15 to Aug. 21 6. 0 0. 13 Aug. 22 to Sept. 3 0. 0 0. 0 A - 6 277 Appendix B: Production Guide Information Note: This is Appendix B of the crop loss questionnaire. 278 Vegetable Production Guide 1987 B/ilish Columbia Caialogumg m Publication Daia Mam enlry under lille: Vegutahie production guide. — 1972-Annual. f l c p o l for 1976-77 covers two ycai period. Cover Nile, 1985- : Vegetable production guide (or cmnmofoa i growers. ' Prepared by ollrfials ol Ihc C i n . i d a nnd Piovmcial Dcpadrnenls o' Agncullure," Issuing body vanes: 197?-1976-77. Dcpf of Agrtcul-lure; 19761980. Ministry ol Agriculture. 1961-1985. Ministry ol Agncullure and Food, I9A7- , Ministry ol Agriculture and Fisheries. Continues' Vegetable production recommendations ISSN03I8 -3G7X ISSN 0318OG61 Vegetable production g u » l o 1. Vegeiabie gardening • Qndsn Columbia • Periodicals. 2. Vegetables • Diseases and ucsls • Qniish Columbia • Periodicals 3. Pesi conirol • Bnl.sh Columbia Periodicals. I. UnhsM Columbia. Dcpl Ol Agncullure. II Unhsh Columbia. Ministry ol Agncullure. Ill British Columbia. Ministry o( Agncullure and f o o d . IV. British Columbia Ministry of Agncullure and Fisheries V. C a n a d a Dcpt of. Ay/iciilture. VI Canada Agriculture C a n a d a VII Tillc. Vegetable production guide lor commercial growers. SQ320 8 C3V-13 6.15'09711 Rev. Dec 1986 Province of Brit ish Co lumbia Victoria, B .C. Ministry of Agriculture and F i sher ies 1987 279 DRAINAGE AND IRRIGATION GENERAL Water management is an essential part of vegetable pro-duction. O p t i m u m yields can be obtained only where a reasonable measure of control of the water in the soil has been achieved. The effectiveness of most, if not all other management factors, varies in direct proport ion with the degree of control of soil water. Either too little or too much water adversely affects product ion and natural conditions rarely satisfy crop requirements. The B.C. Ministry of Agr iculture and Fisheries provides an advisory service on irrigation and drainage. DRAINAGE Remova l of excess water in spring, fall and winter is an essential requirement of soils in the humid, coastal regions of British C o l u m b i a and, to a lesser degree, of some interior areas. In the interior, drainage is frequently also required (or reclamation and control of soil salinity and alkalinity. Underdra inage is the most effective means of controll ing water (or the majority of soils. Lightweight continuous, flexi-ble perforated plastic drain tubing is used. The plastic tubing alone will work effectively in many soils, but slowly draining soils, clay soils and sloping land subject to erosion require porous materials (drain rock, pea gravel or wood shavings) immediately over the tubing. S andy soils require the use of a nylon mesh sleeve a round the tubing to prevent sand from clogging the drain tube. A filler should not be used on organic soils. Drain depth and spacing to be used depends on the soil type. It may be necessary to use pumps in low-lying areas which lack gravity outlets. Properly constructed drainage systems are considered a permanent improvement and the cost should be amortized over a number of years. IRRIGATION Loss of water by evaporat ion from the soil and use by the crop exceeds natural rainfall dur ing at least part of the grow-ing season almost anywhere in the province. Appl icat ion of addit ional water at the right time will, therelore, increase total yields. In the southern interior dry bell, irrigation is essential lor successful crop production. Under extreme conditions, irrigation will not only increase yield but will prevent possible crop failure. New plantings may (ail if addit ional water is not prov ided to the very .small and shal low root system when the soil is dry. There are various means and systems of applying water, including furrow, subsurface, trickle and overhead sprinkler irrigation, each having its own merits. MANAGEMENT OF COASTAL PEAT AND MUCK SOILS Peat and muck soils decompose and subside when they are drained and cultivated. This is a natural process result-ing from the oxidat ion of soil organic matter upon exposure to air. The average rale of decompos i t ion and subsidence in most organic soil is about 1 cm per year. Though f looding dur ing winter will reduce decomposit ion and subsidence, it is not recommended (or these soils due to compact ion and erosion problems. Ho ld ing the water table to within 10 c m below the soil surface during winter is a practical way to limit decompos i t ion and still maintain a porous surface soil layer. Exper imenta l work in Quebec and elsewhere has shown that decompos i t ion of organic soils can be effectively con-trolled by the addit ion of copper as follows: App l y 60 kg/ha (24 kg/acre) of copper sulfate each year (or three years, fo l lowed by up to 20 kg/ha every 2nd or 3rd year particularly where onions, carrots or lettuce are grown. C o p p e r may be added safely until the copper concentration in the soil reaches 3 00 ppm. Tillage contributes to decompos i t ion of muck and peat by increasing exposure to air. Excessive tillage with rolovators increases soil compact ion wh ich leads to reduced water and air movement and eventual ly to restricted workabi l i ty of the soil. Pond ing on a tiled field after rain is an indicator of poor soil structure imped ing drainage. Subsurface saturation to 10 cm. good tillage practices and min imal traffic should maintain the soil in good tilth. PROBLEM SOILS Vegetable crops require moderate ly to wel l -drained soils with at least 0.5 m unrestricted rooting depth, in order lor top yields to be obta ined. Most lowland soils in the Fraser Valley arc naturally poor ly drained with a high water table in the fall, winter and spring. These soils arc not well suited to vegetable produc-tion without the use of a tile drainage system to remove excess water from the rooting zone. (See the fol lowing sec-tion on Drainage and Irrigation.) Many up land soils in the coastal region have hardpan subsoil within 0.5 m of the surface. In most cases, this hard-pan will not a l low the soils to drain dur ing the (all, winter and spring. S u c h soils require a tile drainage system to remove excess water from the rooting zone. In the coastal region, soils with slopes over 3 % are subject to water erosion when cult ivated and left bare over the winter. Many up land soils have slopes from 5 to 1 0 % or more and on these, water erosion can be a serious problem. Valuable topsoil is r emoved from the upper slopes and may be deposited deep enough to bury plants on the lower slopes. Where water eros ion occurs, the soils require drain-age and other special management practices. (See section on Eros ion Control.) The B.C. Ministry of Agriculture and Fisheries has soil maps cover ing most of the farmland in B.C. These maps can give a good indication of potential soil problems for vegetable product ion. Growers planning to use new land should contact the ministry 's Soils Branch for recommendat ion [or soil suitability and management. 3 280 E R O S I O N C O N T R O L — C O A S T A L REG IONS Where wafer erosion is a problem, some erosion control practices should be used to reduce and minimize soil losses. Erosion damage is worst on long, sleep slopes where the crop rows run up-and-down the hill. Where practical, plant-ing should be done across the slope. A winter cover crop should be used. Barley or rye broad-cast seeded at 100-150 kg/ha (40-60 kg/acre) before Sep-tember 15 will provide good erosion control. The cover crop will also serve as green manure. Barley will usually be killed by winter frosts. Barley that survives the winter can be controlled the following spring wilh herbicides or cultivation. Rye will not be killed by frost and will grow again in the early spring. It can be tilled under in the spring or killed by herbicides. Wet spots in fields should be drained with an under-drainage system. Runoff water from adjacent fields should be controlled wilh an interceptor drain. SAL IN ITY O F I RR IGAT ION WATER In some coastal areas, during prolonged dry spells, ditch water used for Irrigation may become quite saline. Growers are advised to check the water for salts or bring samples into local district offices (or salt conductivity tests before using such water for irrigation. FERTILIZERS AND LIME SOIL TEST ING A soil analysis is the most accurate guide to fertilizer and lime requirements. It is especially important to determine soil fertility and pH levels before planting a crop, so that necessary lime and fertilizer can be applied to the soil. Soil testing as well as tissue testing are useful for determining fertilizer requirements in established crops. Soil and tissue sampling must be done accurately and carefully i( the sam-ples are to be representative of soil and crop conditions. (See B.C. Ministry of Agriculture and Fisheries publication "Soil Sampling" for proper soil sampling methods. Consult the local horticulture specialist for methods of tissue sam-pling.) SOIL A N A L Y S I S The B.C. Ministry of Agriculture and Fisheries has a soil and tissue testing laboratory for soil analysis and fertilizer recommendations. A "Soil Sample information Sheet" must be sent with each sample in order (or a proper inter-pretation of soil analysis leading to a useful soil recommen-dation. This sheet indicates two important things: 1. crop(s) to be grown and 2. field history — which includes previous crop, previous rates of fertilizer, lime and manure applied, drainage, soil type and topography. Information sheets, soil sample boxes and price informa-tion may be obtained from your local B.C. Ministry of Agri-culture and Fisheries office. Samples may be brought into district offices for shipment to the lab in Kelowna or they may be mailed or sent by bus to B.C. Ministry of Agriculture and Fisheries, Soils Branch, Soil Testing Laboratory, 1873 Spall Road, Kelowna, B.C. V1Y 4R2. FERTIL IZER R E Q U I R E M E N T S The fertilizer recommendations given in this guide are general guidelines only and are to be used if soils have not been tested. The amounts of fertilizer are slated as kilo-grams of nitrogen (N) or phosphate ( P 2 O 5 ) and potassium or potash (K2O) per hectare. No general recommendations are made for magnesium (Mg), sulfur (S) or trace elements such as boron (B), copper (Cu), zinc (Zn) or manganese (Mn). Some soils are known to be deficient in one or more of these elements and, therefore, soil testing is highly rec-ommended. C A L C U L A T I O N O F FERT IL IZER RATES Fertilizers have been labelled by percentage according to their guaranteed minimum analysis in terms of nitrogen (N), phosphate ( P 2 O 5 ) , potash (K^O), and other nutrients when these are present. Example: METRIC Five 20 kg bags (100 kg) ol 11 -48-0 contain 1 1 % nitrogen (11 kg N), 4 8 % phosphate (48 kg P2O0) and no potash (0 kg K^O). The rest of the material in Ihe five bags arc other ele-ments (oxygen and hydrogen) that are part of the fertilizer compounds carrying the nitro-gen, phosphate and potash. A. The amount of fertilizer required = recommended rate xlOO Example fertilizer analysis Recommended rate potash Fertilizer analysis Amount of fertilizer METRIC 135 kg/ha 0-0-60 135 kg/ha x 100 60 225 kg/ha Apply 225 kg/ha of 0-0-60. B. The amount of nutrient applied by a fertilizer = amount of fertilizer applied x fertilizer analysis 100 Example: METRIC Amount of fertilizer applied 225 kg/ha Fertilizer analysis 13-16-10 Amount of N supplied = 225 kg/ha x 13 100 29 kg/ha of N = 225 kg/ha x 16 , 100 36 kg/ha of P?O s = 225 kg/ha x 10 100 = 22.5 kg/ha of K,0 Amount of P/Oo supplied Amount of K2O supplied 281 M E T H O D OF FERTIL IZER A P P L I C A T I O N (TERMS DEF INED ) Broadcast ing and Incorporat ion Refers to spreading fertilizer on a soil surface before the crop has been planted, then incorporating the fertilizer into the soil with tillage. Top Dress ing Refers to spreading fertilizer on a field when a crop is growing. It is not incorporated, but sprinkler irrigation will wash fertilizer off the leaves to prevent burning of the leaves and move the nutrients into the surface few centimeters of soil. Banding Refers to the application of fertilizer at time of planting in continuous bands 2.5 cm or- more to the side of the plant and 5 cm or more deep, depending on the crop. S ide Dress ing Refers to the banding of fertilizer after plants are estab-lished. Care should be taken not to disturb the roots of the plants. Fert igation Refers to the application of fertilizer in irrigation water. Deep Banding Refers to banding fertilizer at a depth of 5 cm or more prior to planting. There is scientific evidence indicating that this results in greater fertilizer efficiency lhan surface broad-casting. L IME When the soil pH (acidity) is not known, have the soil tested. Soil acidity can be con-ecled by the application of lime. On extremely acid soils (low pH), most crops will not respond to fertilization or other management factors. Some soil limed heavily over a period of years may not require further applications. Some light-textured soils that have an adequate pH occasionally test very low in calcium, and therefore require lime. Some crops are more tolerant of acidity and may not require the addition of lime to the soil. Use of some dolomitic limestone is recommended since it contains a significant quantity of magnesium which is an essential and often deficient plant nutrient. Effects of L ime 1. Corrects soil acidity. 2. May improve the physical condition of the soil. 3. Provides the nutrient calcium and increases the avail-ability of other plant foods. 4. Favours bacterial action, thus hastening the decompo-sition of organic matter and releasing additional plant foods. 5. Improves conditions for availability of other nutrients, notably phosphorus and some minor elements. 6. High rates of lime may digest organic matter and release nitrogen for a short period after application. 7. Reduces toxicity of some elements such as manganese and aluminum. 8. Above 6.7 tonnes/ha may tie up some micro-nutrients such as boron. Magnesium deficiencies may be aggra-vated, especially in sandy soil. Where this is a problem, some dolomitic lime should be used. Forms of L ime Used 1. Calcium oxide — quicklime, caustic lime, burned lime. Not recommended on agricultural land. 2. Calcium hydroxide -hydrate or slaked lime. Should only be used as a spring application for rapid results. "Agricultural L ime" refers to this form but the use of this term is not recommended. It is the quicker acting form of agricultural lime. It will correct soil acidity quickly, but is usually two or more times as expensive. Excessive rates above 1100 kg/ha (450 kg/acre) may be quite caustic and "burn out" organic matter. 3. Ground limestone — Calcium carbonate: The most convenient form to handle. May be applied at any time of the year. It dissolves slowly and lasts longer in the soil. 4. Ground dolomite — Calcium-magnesium carbonate: May be substituted for ordinary limestone. Contains magnesium. 5. Marl — essentially calcium carbonate. It is generally comparable to limestone where available, but most marl supplies are wet marl, requiring special spreading units. 6. Other materials — chalk, sea-shells, carbide residue. NOTE — Fineness of grind is very important, the finer grinds (100 mesh and above) react in soil much quicker than the coarser grind (10-100 mesh). Very coarse lime-stone (less than 10 mesh) is not recommended. Some coarse material is desirable to facilitate handling of the lime. Excessively fine material will not flow readily and is subject to wind drift during spreading. M A N U R E Cow and poultry manures are commonly used. Each type differs in nutrient content. Manures supply plant food over a period of time, but in the year of application each tonne ol menure should supply approximately the amount shown in the following table. Recommended rates of N, P^Or,, and K^O can be adjusted downward, depending upon the manure applied. Some chemical fertilizer should be used, even with heavy manure applications. Cow manure may be applied up to 45 tonnes/ha (18 tonnes/acre) where crops are to be grown. Poultry manure may be applied up to half this rate. Kilograms per Tonne Type of Manure Quality 1 N ol Manure P.Os K,0 Cow Good 2.5 1.5 4.0 Manure Poor 1.0 0.5 < 3.0 Poultry Good 5.0 3.0 4.0 Manure Poor 3.0 2.0 3.0 ' G o o d quality refers lo properly stored manure, containing lillle liller. poor quality refers to exposed leached manure or manure wilh consider-able liller. 282 H i e following (able provides a guide for convening spreader capacities to tonnes and cubic metres (Ions and cubic yards) of manure. Spreader Capac i ty Tonnes of M : 1 of bus. (heaping) Manure Manure 75 2.7 2.6 100 3.6 3.6 125 4.4 4.5 150 5.2 5.3 ' 175 6.2 6.2 3 35 11.9 11.9 I n i t ' i i - i m M r e m.inure - nppioximnlc ly 26 liuslids I Uintu'in.imin' = approximately 1.3 cubic metres G R E E N M A N U R E O n land where vegetables are to be grown for a number of years, green manur ing is a c o m m o n practice which is highly recommended. In addit ion to organic matter mainte-nance (especially important on mineral soils), green manur-ing crops are valuable in preventing nutrient locses follow-ing harvest of the vegetable crop. In fields where vegetable crops are to be seeded the fol lowing year, oats are preferred to rye. because the latter when p loughed down in the spring causes problems with the seeding of these crops. O n muck ioils oats should be seeded before September 1, and then ploughed clown in the fall before winter rains prevent imple-liont traffic. O n fields where potatoes are to be planted or .•egptables transplanted, rye (variety Tetra, Petkus) may be ree led (not later than October 15) for ploughing down in he fall or spring. G reen manure crops p loughed down in ho spring should receive 100 kg/ha (40 kg/acre) of ammon -uni nitrate (34-0-0) just prior to ploughing. See Soi l Bul letin 9, Green Manure, also Agriculture C a n -ida Bulletin 868. Manures and Composts . 5 T A R T E R S O L U T I O N S High analysis, readily soluble or in-solution fertilizers that i r e used with seedlings or transplants to help them to start trowing quickly are referred to as starter solutions. During A ' a r m weather and under drying condit ions the addit ion of . o a l o r by itself wil l often be beneficial. Starter solutions are l o l p f u l , however, especially in coo l weather, since nutrients n t h e solution are immediately available. C o m m o n l y used b a r t e r solutions include phosphor ic acid (0-52-0) 20-20-20, 10-50-10. 10-52-17 and 21-53-0. A starter solution high in phosphorus such as 10-50-10 shou ld be used for tomatoes, p e p p e r , eggplants, melons and cucumbers. Dissolve 2 kg of 10-5'"1-10 in 180 to 2 50 L of water and give each plant 0.3L ii t h i s solution at transplanting time. For crops such as cabbage, cauliflower, broccol i and cel-j e r y . 20-20-20 can be used at the rate of 1.5 kg/500 L of v a t e r . Each plant should receive 0.2 to 0.3 L of this solution i t transplanting time. Pesticides should only be added to the U a r t e r solution if r e commended by the manufacturer or if ,'xperience has indicated they are compatible. FERTIL IZER R E A C T I O N S IN SO IL Fertilizers added to the soil may become more or less available, depend ing on the type of fertilizer, soil moisture and pH condit ions and temperature. S o m e nutrient ele-ments may be complete ly lost, others may be temporanly " t i ed up." N I T R O G E N The most c o m m o n forms of fertilizer nitrogen are nitrate (NO.-,), a m m o n i u m (NH„) and urea (CO(NH 2]2). A l l are highly water soluble. Urea is converted to the ammon ium form when it is acted upon by enzymes in the soil. A m m o n -ium nitrogen is absorbed by clay minerals and organic mat-ter and is, therefore, not easily lost from the soil. Nitrate nitrogen is not absorbed by the soil and can be lost by leaching with water. Leach ing losses of nitrate nitrogen are greatest in sandy soils and in areas with high rainfall. S o m e a m m o n i u m and urea nitrogen may be converted to ammon i a gas, wh ich escapes into the atmosphere. This usually occurs in dry soil with surface-appl ied fertilizer. A m m o n i a losses are reduced or el iminated by ensuring that the fertilizer is wel l covered with moist soil. Losses are min-imized by banding, immediate incorporat ion after broad-casting, irrigation fol lowing appl icat ion or broadcasting onto moist soil in coo l weather. S o m e nitrate nitrogen may be converted to gases, which escape into the atmosphere. This frequently occurs in wet soils during fall, winter and spring. P H O S P H O R U S A l l phosphorus fertilizers are phosphate salts. They are water soluble, but tend to form insoluble compounds when incorporated into the soil. Unl ike nitrogen and potassium, phosphorus does not readily move in the soil. It tends to remain where it has been placed. Therefore, it is important to place phosphorus fertilizer in the rooting zone of the crop before the crop is established, or to band it next to the roots in the established crop. Surface appl ied phosphorus without incorporat ion is the least efficient way of utilizing fertilizer phosphorus. S o m e phosphorus becomes " t i e d - u p " at low p H (below 6.0) and at high p H (above 7.5). Very little phosphorus leaches out of the soil. P O T A S S I U M Potassium fertilizers are all s imple potassium salts, such as potassium chlor ide, potass ium sulfate, potass ium-mag-nes ium sulfate, or potass ium nitrate. A l l are readily water soluble. Potass ium is absorbed to some extent by organic matter and clay minerals. However , it is subject to leaching, especially in sandy soils. S E C O N D A R Y NUTR I ENTS Magnes ium and sulfur may be deficient in the soil lor good crop growth. So i l and tissue testing are the only accu-rate way to determine if these are lacking. S ince calc ium is appl ied as lime it is rarely deficient in soils. M a n y c o m m o n fertilizers contain ca lc ium and sulfur and magnes ium fertiliz-ers are available. 283 MICRONUTRIENTS Iron (Fe), manganese (Mn), copper (Cu), zinc (Zn) and boron (B) are sometimes deficient in the soil for crop pro-duction. Soil and/or tissue testing are the only accurate ways to determine if these elements are lacking. If they are needed, micronutrients can be added to blended fertilizers and applied along with the routine fertilizer program. If nec-essary, micronutrients can be applied in irrigation water. Micronutrients are required only in very small amounts and it is important to ensure that micronutrient fertilizers are applied at the correct rate. High levels of micronutrients, especially boron and manganese, are toxic to plants. BORON Boron deficiency causes a wide variety of abnormalities in vegetable crops. When required, broadcast a boron compound evenly on the soil with a grass-seeder of the cyclone type. Fertilizers which include boron can be be obtained in most areas. CAUTION — Do not exceed the recommended amount of boron per hectare as it may cause plant injury. If boron-deficiency symptoms occur during growing sea-son, boron can be applied as a spray. Apply either Borospray, Solubor or Polybor at manufacturers' directions. See also under Nutrient Deficiences. page 7. In the Interior, do not plant beans or cucumbers the year following an application of boron to other crops. CONTROL In the Interior, boron should be applied in the toll. At the Coast, it should be applied in the spring where a need for it has been shown. Refer to B.C. Ministry of Agriculture and Fisheries publication Boron, Soil Series, from which the fol-lowing table is an extract, or to your nearest district agricul-tural office. KILOGRAMS OF BORON COMPOUNDS TO SUPPLY THE PER HECTARE REQUIREMENTS FOR EACH CROP GROUP A 1 B C D E Asparagus, Broad Beans, Carrots, Beets, Broccoli. Celery. Corn , Brussels Sprouts Cantaloupe. Melons, Eggplant, Lettuce. Cabbage , Caulif lower. Oats. Parsley. Peas, L ima Beans. Muslard. Ch inese Cabbage. Per Cent Beans. Potatoes, Pumpkin, Onions, Peppers, Kohlrabi. Boron Cucumbers Squash, Watermelon Radish. Sp inach Tomatoes Rutabaga. Turnips A c t u a l Boron A p p l i e d 0 1.1 2.2 3.5 4.5 Boron C o m p o u n d s Borax 1 1.4 0 10.0 20 30 40 Borate granular f/crtilrzer 14.0 0 7.8 16 24 31 •Borate High Grade. Tronabor, Borate 46) Boric Ac id 17.5 0 6.8 12 18 25 Borate powder (Borate 65, 20.5 0 5.5 11 17 22 Borate 68, Borospray, Polvbor. Solubor) ' A group is most sensitive. NUTRIENT DEFICIENCIES When parts of green plants that are not yet mature become off colour (yellow, purple, etc.) or show abnormal growth, stunting, distortion, cracking, brittleness or a combi-nation of these symptoms, a nutrient deficiency may be the prcblem. The nutrient required may be present in the soil but unavailable to the plants because of weather or soil conditions. It is common to see early planted corn turn purple during prolonged cold weather because the phos-phate which the crop needs is not available to the plant, even though the soil supply is more than enough. Some nutrients will slow down the uptake of other nutrients unless they are present in the correct proportion. All nutrients have a pH range at which they are most avail-able to the plants, providing other (actors are favourable. Too much of a nutrient may cause growth problems as well. Tomato and bean plants may produce stems and leaves but not fruit when too much nitrogen is added. Excess fertilizer may cause leaf "burn " or stunted growth. The following table is offered as a guide for suspected nutrient deficiencies. For additional help consult your district agricultural office. 284 M.ijor Nuiricnls Nutrient Deficiency Symploms Niiiuycn S Y M P T O M S OF FOLIAR NUTR IENT D E F I C I E N C I E S A N D C O R R E C T I V E FERTIL IZER T R E A T M E N T S Application Hale of Suggested Source Soil Poliar Materials kg/ha g/HX) t S low growth. I 'ale green to yellow colours. Phosphorus S low growth. Reddish, purple colors. Potassium S low growth. Mottling followed by bronzing and drying o l leaf tips and margins. Ca lc ium S low growth. Die back of terminal growth. Lea l margins chlorotic and scorched. Spotting o l leaves. B lossom-end rot of tomatoes. Magnesium Yellow green interveinal mottling beginning wilh the older or lower leaves. Scorching and briltleness. Boron Growing points may die. Stems and leaves may he distorted. Stems may be hollow and roughened. Brown curd of cauliflower. Co rky scars on broccoli stalks. i:on Strong yellowing and sometimes whitening of the young or new leaves al the lips of the plants. Brassicas show chlorotic marbling. Manganese Stunting of plants. Smaller than normal leaves wilh yellowing between the veins. Mo lybdenum Interveinal yellowing on older larger leaves. Pul led appearance of chlorotic areas and upward curling of the leaflet margins Zinc Green and yellow broadslriping at base o l corn leaves. C o r n silks can be delayed in appearance. Poor filling of corn ears can result due to lack o l pollination. Interveinal yellowing with marginal burning on beet leaves. 3'1-U-O. 21-0-0 or 16-0-0 Sec crops such as corn. Cucumbers , etc. for side-dressing rales. See Starter Solutions See Starter Solutions Calc ium Chlor ide Calc ium Nilrale Lime Dolomitic Limestone Sul-po mag Epsom Salts O R Magnesium Sulphate Solubor Borax Ferrous Sulphate Chelated Iron Manganese Sulphate Sod ium Molybdale Molybdic acid Zinc Sulphate Zinc Chelate 2.2 to 4.5 tonnes 250 to 300 250 lo .100 170 to 225 See table 10 l o 4 5 15 to 15 5 0 0 lo 1000 500 lo 1500 10(10 lo 1500 100 to 150 200 to 300 75 to 100 200 lo 400 25 to 50 50 200 lo 400 25 to 100 S T E R I L I Z A T I O N A N D F U M I G A T I O N O F S O I L A N D E Q U I P M E N T G R E E N H O U S E SOILS, S EED -BEDS , G A R D E N S Soil sterilization or fumigation has an important place in controlling damping-off and other diseases, insects, weeds and nematodes in seed-beds and greenhouses. The following rules must be followed to achieve satisfac-tory results: 1. The soil temperature at 15 cm depth must be 13°C or higher for successful treatment with most chemicals. 2. Soil must be in a loose condition so that penetration is complete. Sods, lumps and organic materials must be thoroughly broken up. ' 3. If organic materials (manure, compost, elc.) are to be used, they must be incorporated before treatment so that recontaminarion does not occur. 4. The soil must be moist, but not wet. When soil is sterilized with steam or fumigated with chem-icals, the number of soil micro-organisms is greatly reduced (or the first few days, then it rises and eventually exceeds that of untreated soil. The sterilizing or fumigating destroys a large part of the dense population of soil microbes, and the first organisms to return after treatment meet no severe competition. Thus, if plant pathogens are among the first to recolonize the soil, they develop rapidly and cause severe disease losses. It is therefore important to the grower that every effort is made to prevent disease organisms from gaining entrance to the soil. Pathogens can gain entrance to the soil by: 1. splashing of rain or watering; 2. cutting; 3. soil in water hose; 4. infested containers; 5. infested tools and equipment. 6. grower's hands and footwear; 7. placing containers on ground; 8. unsterilized covers; 9. infected plants or seeds. Ste am Controls all disease organisms, insects.'nematodes and most weed seeds. Steam sterilization is done by passing large quantities of steam into the soil until the temperature becomes sufficiently high at the desired depth. For this pur-pose a steam boiler is required. A greenhouse steam boiler or a portable unit of about 200 kW (20 horsepower) may be used. For successful sterilization, the coldest part of the soil mass must be held at 82°C for 30 minutes. 8 285 Crop Loss Questionnaire (Part B) Answer and Assessment Booklet 286 PART I: Relative S e n s i t i v i t y of Crops to Ozone Crop Green snap bean Potato Pea Corn Br o c c o l i Raspberry Forage (Orchard grass -perennial ryegrass mix) A number between 1 (most s e n s i t i v e ) and 5 (least s e n s i t i v e ) should be placed beside each crop based on the following score: 1 = extremely s e n s i t i v e to ozone 2 = very s e n s i t i v e to ozone 3 = moderately s e n s i t i v e to ozone 4 = somewhat s e n s i t i v e to ozone 5 = not s e n s i t i v e to ozone Relative S e n s i t i v i t y Notes, assumptions, conditions 287 PART I I : Crop Y i e l d Loss Estimates Crop Type ( C i r c l e one): Green bean Potato Pea Corn B r o c c o l i Raspberry Forage F r a c t i l e .05 .50 .95 Low Ozone Scenario Crop Loss Estimate V a l i d i t y Check .95 . 50 . 05 F r a c t i l e .05 .50 .95 Medium Ozone Scenario Crop Loss Estimate V a l i d i t y Check .95 .50 .05 F r a c t i l e High Ozone Scenario 2 Crop Loss Estimate V a l i d i t y Check .05 .50 .95 .95 .50 . 05 Pr o b a b i l i t y that true crop loss i s l e s s than or equal to crop loss estimate. 2 Given i n u n i t s , based on a maximum hypothetical y i e l d of 100 units. 3 P r o b a b i l i t y that true crop loss i s greater than crop loss estimate. Note: Please include notes, assumptions or conditions on reverse s i d e of t h i s sheet. 288 Notes, Assumptions, C o n d i t i o n s 289 PART I I I : Dose S t a t i s t i c s and Other Cues U t i l i z e d P l e a s e p r o v i d e a l i s t i n g of the dose s t a t i s t i c s and o t h e r i n f o r m a -t i o n p r i m a r i l y used i n your crop l o s s e s t i m a t e d e c i s i o n s , f o r each crop type. I f the same s t a t i s t i c s were u t i l i z e d f o r a l l c r o p t y p e s , you may i n d i c a t e so and p r o v i d e a l i s t i n g f o r one c r o p o n l y . Crop Primary I n f o r m a t i o n Used i n Crop Loss E s t i m a t e Green snap bean Potato Pea Corn B r o c c o l i Raspberry Forage 290 PART IV: Respondent Information Please provide the following information by checking (V) the appropriate space ( S ) , Highest degree obtained: BSc MSc PhD Number of years career experience since graduation: Less than 10 10 to 20 21 to 30 31 to 40 41 to 50 Type of career experience: Research Experiments Administration Teaching Consulting Employer: U n i v e r s i t y Government Consulting Industry Number of Pu b l i c a t i o n s i n l a s t 20 years: (dealing with a i r p o l l u t i o n e f f e c t s on crop) Less than 10 10 to 20 21 to 30 31 to 40 more than 40 With respect to c o n f i d e n t i a l i t y and anonymity: I wish to remain t o t a l l y anonymous regarding t h i s p r o j e c t My name can be used as a p a r t i c i p a t i n g expert, but my judgments regarding crop loss should remain anonymous (e.g. I should be i d e n t i f i e d as Expert "A" or s i m i l a r ) It i s not necessary to treat my i d e n t i t y or judgments as c o n f i d e n t i a l Signature of Respondent 291 APPENDIX C Descriptive Statistics Associated with Crop Sensitivity. Descriptive Statistics Associated with Unweighted Expert CDFs. Descriptive Statistics Associated with Weighted Expert CDFs. 292 Relative Sensitivity of Crops to Ozone VARIABLE LABEL N MEAN C1 Green Snap Bean 9 4 11111111 C2 Potato 8 3 25000000 C3 Pea 9 3 00000000 C4 Corn 8 1 87500000 C5 Brocco 1 i g 2 22222222 C6 Raspberry 7 2 00000000 C7 Forage 8 2 00000000 t-o w 9:20 THURSDAY, JANUARY 4, 1990 1 STANDARD DEVIATION 0.78173596 1.28173989 1.11803399 0.83452296 1.20185043 1.00000000 0.92582010 MINIMUM VALUE 3.00000000 2.00000000 1.00000000 1.00000000 1.00000000 1.00000000 1.00000000 MAXIMUM VALUE 5.00000000 5.00000000 4.00000000 3.00000000 4.00000000 3.00000000 3.00000000 Unweighted Means VARIABLE LABEL N A 1 1 Low Ozone 0.05 Green Bean 8 A12 Low Ozone 0.50 Green Bean 8 A13 Low Ozone 0.95 Green Bean 8 A21 Medium Ozone 0.05 Green Bean 8 A22 Medium Ozone 0.50 Green Bean 8 A23 Medium Ozone 0.95 Green Bean 8 A31 High Ozone 0.05 Green Bean 9 A32 High Ozone 0.50 Green Bean 9 A33 High Ozone 0.95 Green Bean 9 B11 Low Ozone 0.05 Potato 9 812 Low Ozone 0.50 Potato 9 B13 Low Ozone 0.95 Potato 9 B21 Medium Ozone 0.05 Potato 9 B22 Medium Ozone 0.50 Potato 9 B23 Medium Ozone 0.95 Potato 9 B31 High Ozone 0.05 Potato 9 B32 High Ozone 0.50 Potato 9 B33 High Ozone 0.95 Potato 9 C11 Low Ozone 0.05 Pea 8 C12 Low Ozone 0.50 Pea 8 C13 Low Ozone 0.95 Pea 8 C21 Medium Ozone 0.05 Pea 8 C22 Medium Ozone 0.50 Pea 8 C23 Medium Ozone 0.95 Pea 8 C31 High Ozone 0.05 Pea 9 C32 High Ozone 0.50 Pea 9 C33 High Ozone 0.95 Pea 9 D11 Low Ozone 0.05 Corn 8 D12 Low Ozone 0.50 Corn 8 D13 Low Ozone 0.95 Corn 8 D21 Medium Ozone 0.05 Corn 8 D22 Medium Ozone 0.50 Corn 8 023 Medium Ozone 0.95 Corn 8 D3 1 High Ozone 0.05 Corn 9 D32 High Ozone 0.50 Corn 9 D33 High Ozone 0.95 Corn 9 E1 1 Low Ozone 0.05 B r o c c o l i 8 E12 Low Ozone 0.50 B r o c c o l i 8 E13 Low Ozone 0.95 B r o c c o l i 8 E21 Medium Ozone 0.05 B r o c c o l i 8 E22 Medium Ozone 0.50 B r o c c o l i 8 E23 Medium Ozone 0.95 B r o c c o l i 8 E31 High Ozone 0.05 B r o c c o l i 8 E32 High Ozone 0.50 B r o c c o l i 8 E33 High Ozone 0.95 B r o c c o l i 8 F 11 Low Ozone 0.05 Raspberry 7 F12 Low Ozone 0.50 Raspberry 7 F13 Low Ozone 0.95 Raspberry 7 F21 Medium Ozone 0.05 Raspberry 7 F22 Medium Ozone 0.50 Raspberry 7 F23 Medium Ozone 0.95 Raspberry 7 F31 High Ozone 0.05 Raspberry 7 F32 High Ozone 0.50 Raspberry 7 F33 High Ozone 0.95 Raspberry 7 G1 1 Low Ozone 0.05 Forage 8 3:20 THURSDAY, JANUARY 4, 1990 2 MEAN STANDARD MINIMUM MAXIMUM DEVIATION VALUE VALUE 0. 62500000 1 18773494 0 00000000 3 00000000 2. 62500000 2 97 309 363 0 00000000 a 00000000 4 62500000 4 10356987 0 00000000 10 00000000 2 . 00000000 2 26778684 0 00000000 5 00000000 5 50000000 4 53557368 0 00000000 12 00000000 9 . 62500000 5 37022213 0 00000000 15 00000000 4 77777778 4 35252162 0 00000000 10 00000000 13 22222222 6 62906060 5 00000000 25 00000000 23. 00000000 1 1 22497216 10 00000000 40 00000000 0 11111111 0 33333333 0 00000000 1 00000000 1 11111111 1 45296631 0 00000000 4 00000000 2 22222222 2 68224616 0 00000000 8 00000000 0 88888889 1 05409255 0 00000000 3 00000000 3 22222222 2 81858909 0 00000000 9 00000000 7 33333333 4 55521679 0 00000000 16 00000000 3 11111111 3 62092683 0 00000000 10 00000000 12 11111111 15 20233900 0 00000000 50 00000000 21 88888889 27 37446093 0 00000000 90 00000000 0 12500000 0 35355339 0 00000000 1 00000000 0 87500000 1 72688820 0 00000000 5 00000000 3 00000000 3 66450153 0 00000000 10 00000000 0 37500000 0 74402381 0 00000000 2 00000000 2 75000000 2 60494036 0 00000000 7 00000000 5 12500000 4 22365786 0 00000000 10 00000000 1 33333333 1 50000000 0 00000000 4 00000000 7 11111111 5 90432986 0 00000000 20 00000000 14 22222222 12 32657472 0 00000000 40 00000000 0 00000000 0 00000000 0 00000000 0 00000000 0 37500000 0 51754917 0 00000000 ' 1 00000000 1 12500000 1 12599163 0 00000000 3 00000000 0 75000000 1 03509834 0 00000000 3 00000000 2 25000000 2 31455025 0 00000000 6 00000000 4 87500000 4 35685011 0 00000000 11 00000000 1 44444444 1 50923086 0 00000000 4 00000000 5 77777778 3 56292639 0 00000000 10 00000000 11 00000000 8 20060973 1 00000000 25 00000000 0 25000000 0 70710678 0 00000000 2 00000000 1 00000000 1 77281052 0 00000000 5 00000000 2 37500000 2 87538817 0 00000000 7 00000000 0 50000000 1 06904497 0 00000000 3 00000000 1 87500000 3 04431555 0 00000000 8 00000000 4 25000000 4 80327269 0 00000000 10 00000000 1 12500000 1 88509189 0 00000000 5 00000000 6 37500000 6 90626010 0 00000000 20 00000000 13 25000000 12 98075499 0 00000000 40 00000000 0 2857 1429 0 75592895 0 00000000 2 00000000 1 .28571429 2 21466971. 0 00000000 6 00000000 3 . 142857 14 3 71611676 0 00000000 10 00000000 0 .71428571 1 11269728 0 00000000 3 00000000 2 .85714286 3 28778403 0 00000000 8 00000000 5 . 7 142857 1 5 73626724 0 00000000 13 00000000 1 .42857143 1 98805959 0 00000000 5 00000000 4 .42857143 4 35343324 0 00000000 10 00000000 11 .71428571 12 27075501 0 00000000 35 00000000 0 .12500000 0 35355339 0 00000000 1 00000000 Unweighted Means VARIABLE LABEL N G12 Low Ozone 0.50 Forage 8 G13 Low Ozone 0.95 Forage 8 G21 Medium Ozone 0.05 Forage 8 G22 Medium Ozone 0.50 Forage 8 G23 Medium Ozone 0.95 Forage 8 G31 High Ozone 0.05 Forage 8 G32 High Ozone 0.50 Forage 8 G33 High Ozone 0.95 Forage 8 to CO 9:20 THURSDAY. JANUARY 4 . 1990 3 MEAN 0.50000000 1.25000000 0.62500000 2.12500000 4.37500000 1.12500000 4.75000000 10.62500000 STANDARD DEVIATION 1.06904497 1.75254916 1.06066017 1.88509189 2.82526863 1.64208056 2.43486579 4.77904653 MINIMUM VALUE 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 2.00000000 5.00000000 MAXIMUM VALUE 3.00000000 5.00000000 3.00000000 5.00000000 8.00000000 4.00000000 8.00000000 20.00000000 we lyd LtU MtfdllS VARIABLE LABEL N A 1 1 Low Ozone 0.05 Green Bean 8 A 12 Low Ozone 0.50 Green Bean 8 A 1 3 Low Ozone 0.95 Green Bean 8 A21 Medium Ozone 0.05 Green Bean 8 A22 Medium Ozone 0.50 Green Bean 8 A23 Medium Ozone 0.95 Green Bean 8 A31 High Ozone 0.05 Green Bean 9 A32 High Ozone 0.50 Green Bean 9 A33 High Ozone 0.95 Green Bean 9 B1 1 Low Ozone 0.05 Potato 9 B12 Low Ozone 0.50 Potato 9 B13 Low Ozone 0.95 Potato 9 B21 Medium Ozone 0.05 Potato 9 B22 Medium Ozone 0.50 Potato 9 B23 Medium Ozone 0.95 Potato 9 B31 High Ozone 0.05 Potato 9 B32 High Ozone 0.50 Potato 9 B33 High Ozone 0.95 Potato 9 C1 1 Low Ozone 0.05 Pea 8 C12 Low Ozone 0.50 Pea 8 C13 Low Ozone 0.95 Pea 8 C21 Medium Ozone 0.05 Pea 8 C22 Medium Ozone 0.50 Pea 8 C23 Medium Ozone 0.95 Pea 8 C31 High Ozone 0.05 Pea 9 C32 High Ozone 0.50 Pea 9 C33 High Ozone 0.95 Pea 9 D11 Low Ozone 0.05 Corn 8 D12 Low Ozone 0.50 Corn 8 D13 Low Ozone 0.95 Corn 8 D21 Medium Ozone 0.05 Corn 8 D22 Medium Ozone 0.50 Corn 8 023 Medium Ozone 0.95 Corn 8 D31 High Ozone 0.05 Corn 9 032 High Ozone 0.50 Corn 9 D33 High Ozone 0.95 Corn 9 E11 Low Ozone 0.05 B r o c c o l i 8 E12 Low Ozone 0.50 B r o c c o l i 8 E13 Low Ozone 0.95 B r o c c o l i 8 E21 Medium Ozone 0.05 B r o c c o l i 8 E22 Medium Ozone 0.50 B r o c c o l i 8 E23 Medium Ozone 0.95 Brocco l i 8 E31 High Ozone 0.05 B r o c c o l i 8 E32 High Ozone 0.50 B r o c c o l i 8 E33 High Ozone 0.95 B r o c c o l i 8 F11 Low Ozone 0.05 Raspberry 7 F12 Low Ozone 0.50 Raspberry 7 F13 Low Ozone 0.95 Raspberry 7 F21 Medium Ozone 0.05 Raspberry 7 F22 Medium Ozone 0.50 Raspberry 7 F23 Medium Ozone 0.95 Raspberry 7 F31 High Ozone 0.05 Raspberry 7 F32 High Ozone 0.50 Raspberry 7 F33 High Ozone 0.95 Raspberry 7 Gl 1 Low Ozone 0.05 Forage 8 : <^U IMUM^UAr, JAINUAH r 4 , I'jau MEAN 0.33090909 2.65545455 5.24363636 1.88727273 5.96000000 10.69818182 4.97454545 14.49090909 26. 11636364 0.06909091 1.71272727 3.54545455 1.25454545 4.98545455 10.11636364 4. 16727273 13.87272727 24 . 738 18 182 0.06909091 0.71272727 3.54181818 0.20000000 3 . 1 1636364 5.94545455 1.28727273 7.26181818 15.17454545 0.00000000 0.54181818 1.52000000 0.76727273 2.93818182 6.14545455 1 .60727273 6.33818162 12.43636364 0.13818182 1.01090909 2.76727273 0.50545455 2.18181818 5.25818182 1.30181818 5.91636364 12.53818182 0.13281250 1 .17968750 3.16406250 0.64062500 3.37109375 6.87890625 1 .33593750 5.25000000 13.22656250 0.06909091 STANDARD DEVIATION 0.32941292 0.97320513 1 .52122240 0.69786519 1 .49988051 1 .68968591 1.43318175 2.79079064 4.20477415 0.08966420 0.58864461 1 . 10003757 0. 34521523 1 .05009563 1.48695484 1 .20986578 4.06663505 7.12323819 0.09585507 0.47608002 1.25092570 0.20639138 0.94862207 1.56689432 0.45840903 1 .94435653 3 . 82442044 0.00000000 0. 18832011 0.46073910 0.34947490 0.94688802 1.73462550 0.55086108 1 .38162439 2.86750100 0. 19171013 0 . 53011504 0.98061880 0. 30810845 1 .02764618 1 .79619643 0.62996561 2. 19149197 4.00352761 0. 19615142 0.61285936 1 . 13807636 0. 31231055 1 .22690695 2.22209921 0.59038220 1 . 65840376 4 . 4 18889 16 0.09585507 MINIMUM VALUE 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 5.00000000 10.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 1.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 MAXIMUM VALUE 3.00000000 8.00000000 10.00000000 5.00000000 12.00000000 15.00000000 10.00000000 25.00000000 40.00000000 1.00000000 4.00000000 8.00000000 3.00000000 9.00000000 16.00000000 10 .00000000 50.00000000 90.00000000 1 .00000000 5.00000000 10.00000000 2.00000000 7.00000000 10.00000000 4.00000000 20.00000000 40.00000000 0.00000000 1 .00000000 3.00000000 3.00000000 6.00000000 11.00000000 4.00000000 10.00000000 25.00000000 2.00000000 5.00000000 7.00000000 3.00000000 8.00000000 10.00000000 5.00000000 20.00000000 40.00000000 2.00000000 6.00000000 10.00000000 3.00000000 8.00000000 13.00000000 5.00000000 10.00000000 35.00000000 1.00000000 weighted Means VARIABLE LABEL N G12 Low Ozone 0.50 Forage B G13 Low Ozone 0.95 Forage 8 G21 Medium Ozone 0.05 Forage 8 G22 Medium Ozone 0.50 Forage B G23 Medium Ozone 0.95 Forage 8 G31 High Ozone 0.05 Forage 8 G32 High Ozone 0.50 Forage 8 G33 High Ozone 0.95 Forage 8 to CO 9:20 THURSDAY, - JANUARY 4, 1990 5 MEAN 0.32000000 0.90181818 0.46909091 2.17090909 4.58545455 0.77454545 4.66181818 0.89454545 STANDARD DEVIATION 0.30051039 0.528B3066 0. 36359480 0.61621669 0.89927585 0.53937040 0.84655117 1.71985228 MINIMUM VALUE 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 2.00000000 5.00000000 MAXIMUM VALUE 3.00000000 5.00000000 3.00000000 5.00000000 8.00000000 4.00000000 8.00000000 20.00000000 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0098765/manifest

Comment

Related Items