'1 f0 iq i 3300 w. Michlgan Avenue F-24 Lanalng, 517-3_21-7358 12 January 1983 Member DNR Environmental Protection Policy Advisory Committee Dear PAC Committee Member: Enclosed find a copy of my written comments to be submitted at the scheduled 19 January 1983 PAC meeting regarding risk as- sessment/risk acceptability in the-regulation of gene-damaging toxic substances. In summary I do not believe that the proposed water quality standard derivation procedures recommended by the Rule 57 Advisory Committee can assure the protection20f the public health at the design acceptable risk with a degree of statistical confidence considered adequate. This is because the margins of safety incorporated into the derivation procedures are insufficient to offset the total'quantifiable uncertainty and the postulated unknown and unquantifiable uncertainty inherent to such procedures. Please take some time prior to the 19 January 1983 meeting to review the enclosed materials. Following.a brief statement at the meeting, I will be glad to answer any questions you may have. Thank you very much for the time you have taken in evaluating the material which I forwarded to you in a;letter dated 3 January 1983 and the materials enclosed herein. photocopies: Sincerely, 57% G. Guenther Larry ?2 Fink, M.S. - Directo RISK ASSESSMENT AND RISK ACCEPTABILITY IN ENVIRONMENTAL PROTECTION COMMENTS BEFORE: DNH Environmental ?Protection Policy Advisory Committee 19 January 1983 DATE Mr. Chairman, members of the committee: I appreciate this opportunity to make this presentation before the DNR Environmental Protection Policy Advisory Committee. My written remarKs detail my concerns. My name is Larry Fink. I hold B.S. Chan. degrees from the University of Michigan-Ann Arbor. The M.S. is in Environmental Health Sciences, with specialization in Environmental Chemistry. 'For the past four years I have been employed by the DNR's Office of Toxic Materials Control, first in the capacity of reviewing Clean Water Act NPDES permit applications for toxic substances hazards and, more recently, in the capacity of evaluating estimator techniques and chemical environmental fate mathematical models for use in exposure assessment. Although I am an employee of the DNR, the views and opinions expressed herein are my own, as a private citizen and Director of the Foresight Society, and should not be construed to be those of my employer. Page Two HISTORY AND BACKGROUND I have been involved in the development of proposed Rule 323.1057 (Rule 57) of the Administrative Rules of the Water Resources Commis- sion (WRC) since September 1979, first as a DNR representative working with the WRC-appointed Water Quality Standards Advisory Task Force (WQSTAF) and later, following fundamental and irreconcilable differences of opinion which arose between me and my supervisors in DNR, as a pri- vate citizen. One of the recommendations made to the WRC by the WQSATF at its June 1980 meeting concerned the need to uniformly and consistently define carcinogenesis, hereditary mutagenesis and teratogenesis; establish a method of calculating genotoxic risk and adopt an accept- able "de minimus" genotoxic risk level. This recommendation was strongly supported by representatives of the private sector, who felt the language of proposed Rule 57 was so broad that it could not stand alone without rules establishing how the general narrative standard was to be translated into site-specific numerical standards. In res? ponse DNR staff indicated a willingness to develop guidelines for such purposes. Private sector representatives rejected this approach, arguing that the Administrative Procedures Act prohibited the adoption of a guideline in lieu of a rule. The proposed Rule 57 was first public noticed in August 1980. I commented on this draft as a private citizen, opposing the decision to authorize the presence of human genotoxic substances in the waters of the state. Private sector public comment opposed minimum data re- quirements and reiterated concerns regarding the vague language of the standard. In response DNR rewrote the draft rule, making it even less detailed. The public comment on this version, public noticed in April 1981, was even more vehement. Following private sector protests at Page Three the June 1981 WRC meeting, DNR staff recommended to the WRC that it adopt a resolution creating a committee of scientific experts to advise the DNR on how to go about developing toxic substances water quality standards-based effluent limits. The resolution was adopted at the July 1981 meeting and recommendations to the WRC as to who should sit on the committee were made and adopted at the August 1981 meeting. At that meeting I appeared before the WRC as a private citizen and indicated my opposition to the composition and constitution of the committee. I pointed out that environmental interests were woefully underrepresented on the committee, that the areas of expertise repre? sented on the committee were not sufficient for purposes of carrying out its mandates, and that none of the appointees were experts in can? cer risk modeling. To assure that the advisory committee be able to fulfill the functions for which it was created, I recommended that the committee operate in the following fashion: 1. . . . announce, conduct and record the minutes of open meetings according to the protocols set forth in the Open Meetings Act (D.A. 267 of 1976, as amended); 2. . . . provide for a period of public comment at each meeting; 3. . . . solicit written and oral testimony from recognized national and international experts; 4. . . . clearly and explicitly identify the issues it intends to address and the value judgments it makes, indicating what weights were given to what factors in determining minimum data requirements, adeguate risk assessment methodologies, acceptable risk, reasonable assurance of protection from unacceptable risk, egg. A follow-up 14 September 1981 letter to Robert J. Courchaine, Executive Secretary, WRC, expanded on these topics. In fact no period of oral public comment was provided for; written Page Four comments received were not put on the agenda for discussion; aside from Dow Chemical USA and EPA scientists, no written or oral comments from nationally or internationally recognized sc ientists were soli? cited prior to public noticing of the final draft: and many value judgments were buried in the text of the draft procedures adopted by the committee. My 24 September 1982 comments on the draft procedures are Attachment.gi. A 22 November 1982 letter to Robert J. Courchaine, summarizing the issues and questions raised by the development of proposed Rule 57 and the Rule 57 Advisory Committee effluent limit derivation pro- -cedures, which have not been adequately addressed to date, has been supplied to you previously as an enclosure in my letter of 1/3/83. The Foresight Society has petitioned for a contested case hearing in the matter of Dow Michigan Division NPDES Permit M10000868, the first such permit to contain acceptable risk-based genotoxic substance effluent limits. Although our comments on the permit are quite exten- sive, a summary of our concerns is contained in a 20 November 1982 letter, a copy of which has already been provided to you. A 20 November 1982 letter to Bailus Walker, Jr., M.P.H., informed him of our discovery that the Midland County white female soft and connective tissue cancer rate for the period 1970?78 exceeds the Michigan and U.S. averages by four times, representing a 771% increase over the rate for the period 1950?59. A copy of this letter has already been provided to you. The Foresight Society believes that, if acceptable cancer risk? based effluent limits are to be adopted, the background risks for the population(s) at risk should be evaluated prior to adoption to determine: l) 2) that the existing cancer, birth defects and hereditary mutations risks do not already exceed the Michigan averages by a factor Greater than l/lO0,000; and that the design acceptable increased incidence of l/lO0,000 is not eventually exceeded as a result of the discharges authorized by the permit. Page Six RISK ACCEPTABILITY AND THE QUALITATIVE DIFFERENCES OF RISK Often the acceptability of a given risk is determined by com- paring its magnitude to those of other common risks with which we live in a normal social context. Although such an approach is ap- pealing, great caution should be exercised in comparing the magni- tudes of qualitatively different risks. I do not believe that the Rule 57 Advisory Committee took sufficient note of these qualitative differences in adopting the l/lO0,000 risk as the design "acceptable" risk. I believe the following qualitative distinctions must be con? sidered when evaluating risk acceptability: acts of God vs human acts readily avoidable circumstances vs not readily avoidable circumstances accurately quantifiable vs/ inaccurately quantifiable quantified from observation of outcomes vs/ quantified by estimation of outcome corresponding social benefits vs corresponding individual benefits The following examples are introduced to illustrate the distinct categories into which various common risks would fall: Risk of Being Struck By A Meteorite act of God not readily avoidable circumstances not accurately quantifiable' for any particular place at any particular time Page Seven overall probability quantified from observation . corresponding benefit of living above ground Risk of Being Struck By Lightning on a Golf Course . act of God readily avoidable circumstances . not accurately quantifiable for any particular golf course during any particular storm event . overall probability quantified from observation corresponding benefit of enjoyment of the sport Risk of Developing Cancer From Medical X?Ray . human act . readily avoidable circumstance . not accurately quantifiable for any particular person overall probability quantified by estimation from high dose response observations of victims of Hiroshima and Nagasaki blasts . corresponding benefit of information on condition of internal organs to assist in accurate diagnosis and identfication of correct treatment Risk of Developing Cancer, From Environmental Exposure to Anthropogenic Toxic Chemicals . human act . not readily avoidable circumstance not accurately quantifiable for any particular person or chemical Page"Eisht 1 probabilities quantified by estimation from single substance laboratory animal high dose - response data corresponding benefits of lower costs of goods and services dependent upon the manufacture, use and disposal of anthropogenic toxic chemical . Many have argued that the state has made life-cost trade-offs similar to those contemplated in the proposed water quality?based effluent limit derivation procedures in the past. Theyfcite as an example the decisions of engineers to design out certain highway safety features in order to cut highway construction costs. However, I would argue that such decisions differ qualitatively from-the toxic substances life?cost trade-off in six important ways: First, the risks of highway accidents under design representative worst?case conditions for a given highway design are quantified from observation of accident frequencies and not from estimates using un? validated models and laboratory data. Second, highway injuries and mortalities result from accidents. Under normal operating conditions no highway is designed to cause death or injury. The death of innocent people as a result of con- suming water and food contaminated with genotoxic substances lawfully discharged to the nation?s waters under a Federal Clean Water Act permit is no accident. Third, driving on any given segment of highway under any given conditions is not a life necessity. One can alter one's travel plans to.avoid driving during inclement weather, to avoid a particularly dangerous stretch of highway, to drive with greater caution or use alternate transportation. Breathing air, drinking water and eating food are life necessities. If one's air; water and food are contami? nated, to avoid exposure involves much greater dislocations and hard" Page Nine ships. In many cases there is up escape. Fourth, if a particular=seg%ent of highway proves too dangerous, it can be redesigned. On the ot%er hand, should we decide later that we have seriously underestimatedithe risks from a particular substance or mixture of substances, the coktamination of the nations waters with persistent toxic substances cannbt be "redesigned" away. Only extra? ordinary measures can remove some but not all of the contaminants. Many years must elapse before natural physical, chemical and biologi? cal processes transform them into innoccuous products. But they never disappear completely. . Fifth, resources freed by the highway engineer in designing out certain highway safety features can be reinVested by the traffic engineer in traffic control features that save more lives for the dollar. Resources freed by the permit development engineer in design? ing out toxic substances zero~discharge requirements tend to appear? in stockholder dividends rather than pollution abatement RHD and 'equipment purchases. Sixth, an individual injured in a traffic accident cannot pass on that injury to innocent future generations. Genetic damage resulting from toxic substances exposure can be passed on to future generations. Often the damage cannot be detected, preventing potential parents from taking appropriate precautions. ?Once the genetic_damage is passed on, it can be propagated throughout a population, eventually expressing itself, perhaps as a debilitating deformity or enzyme deficiency, af- fecting thousands of future innocent lives. Page Ten RISK ASSESSMENT: A REVIEW Risk assessment involves the following: prioritization for limited resource allocation dose - response analysis exposure analysis risk evaluation RISK ASSESSMENT Prioritization For Limited Resource.Allocation l. 2. 6. substance known or reasonably expected to be intrinsically hazardous are identified; the use and discharge budgets of these substances in a given jurisdictiOnal, geographical, airshed, watershed or aquifer location are evaluated; the number of people in each location is quantified; an initial, crude risk estimate is made for each compound; the substances are prioritized from highest to lowest risk; substances to be regulated are identified. RISK ASSESSMENT Dose - Response Analysis 1. estimation of mammalian low dose - toxic response relationship for each route of exposure from laboratory animal high dose toxic response study data, usually for one route of exposure;. conversion of laboratory animal dose, data from concentrations in carrier medium to per unit body weight of test animal; scaling up of the laboratory animal dose in proportion to the greater mass of the design. representative most sensitive human (the so?called mega-rat transformation); "Page Eleven scaling down of the mega-rat dose in proportion to the slower metabolism of the design representative most sensitive human (usually done with the ratio of body surface areas3; 5. quantifiable and uhquantifiable uncertain? ties and sources of error are identified; 6. quantification of quantifiable uncertain- ties via propagated and total error analysis. RISK ASSESSMENT Exposure Analysis 1. identification of the protected population most sensitive to the_toxic effect of the chemical by one or more routes of exposure; 2. establishment of a representative worst?case exposure scenario for the most sensitive pro- tected population; 3. estimation of a representative dose scenario for the most sensitive protected population under the conditions of the design exposure scenario; 4. mathematical modeling of the concentration of the toxic substance in air, surface and ground waters, soil and food as a function of its rate of emission to the air, waters and - soil; 5. quantifiable and unquantifiable uncertainties and sources of error are identified; 6. quantification of quantifiable uncertainties via propagated and total error analysis. RISK EVALUATION . l. A conditional probability distribution function is constructed from the probabilities that: a) a member of the population is a member of the representative most sensitive population; b) a member)of the representative most sensitive population will encounter . circumstances in which the-representative worst?case exposure will occur for some period of time; Page Twelve c) a given dose is delivered_as a result of the integrated exposures throughout a representative worst?case social, occupational and recreational cycle; d) a given toxic response will be elicited by a given dose. 2. For purposes of simplification it is assumed that: 1) all the members of the potentially exposed population belong to the representative most sensitive sub population; 2) each of the design population will encounter the design exposure conditions; 3) the exposure will last a lifetime; 4) all of the substance to which one is exposed enters the body as a toxicologically active dose. 3. A design statistical confidence level for each of the quantifiable uncertainties and sources of error is adopted; 4. A total_error analysis on the risk estimate is performed at the design.statistical confidence level; S. Unquantifiable uncertainties are identified and the existence of unknown sources of error postulated; 6. The emission - risk relationship is adjusted so as to compensate for the quantifiable uncertainties at the design statistical confidence level; 7. The quantifiable uncertainty - adjusted emission - risk relationship is adjusted again to compensate for identifiable but unquantifiable uncertainties and sources of error and for as yet unidentifiable uncertainties and sources of error. Up till now a design acceptable risk.has not yet been adopted,. yet important value judgments have already been made. Among the most significant are those regarding the design statistical confidence level to be adopted for purposes of quantifiable error analysis and age. Thirteen the adjustment factors to be used to compensate for unquantifiable and unknown uncertainties and sources of error. I expand bn the significance of these "hidden" value judgments in the next section. Page Fourteen RISK ACCEPTABILITY, RISK AND MARGINS OF SAFETY The acceptability of any given risk cannot be divorced from the accuracy of its quantification. When the risk of injury from any particular activity is known with an acCeptable degree of accuracy, intelligent choices about the trade?off between risks and benefits can be made. Conversely,-when the risk of injury from any particu- lar activity cannot be accurately quantified such trade-offs cannot be made intelligently- The fear of injury.brought about by the un? certainty of the nature and magnitude of the risk of injury must be factdred into the determination of acceptable risk. As indicated previously, ?uncertainties are introduced into the estimate of risk at each stage in the process. Thus, the re- sulting estimate of risk can only be-quantified in terms of probabilities: (This approach has been used in Attachment II by O'Neill at al. at Oak Ridge National.Laboratories in calculating toxic sub? stances concentration ecosystemic toxic effect probabilities.) There is?always-some probability that a given risk will be significantly underestimated using the approach adopted. That probability can be decreased by factoring margins of safety into each step of the risk assessment process. The degree to which the unquantifiable and in- accurately quantifiable uncertainties need to be offset via the introduction of safety factors is a value judgment. While there is no accepted methodology for making such value judgments, they must be made explicitly rather than implicitly, as was done by the Rule 57 Advisory Committee. Page"Fifteen I do not believe that the use of the design once-in?ten?year, 7?day drought flow dilution factor; laboratory measured or estimated fish bioconcentration factors; the'design 70 kg man; and the design 6 gm/day fish consumption scenario introduce sufficient margins of safety to offset the compounded total uncertainty in the calculated l/lO0,000 risk dose; Further, I do not believe that the Rule 57 Advisory Committee gave sufficient consideration to the margins of safety which ought to be incorporated into the genotoxic substance effluent limit derivation procedures to assure the protection of the public health _at the design "acceptableIL risk to a degree of statistical confidence considered adequate. This may in part be attributable to the fact that the Office-of Toxic Materials Control technical staff with the most knowledge of the sources and magnitudes of uncertainty in the-risk estimate pro? cesses and the greatest misgivings about the degree to which these uncertainties were to be offset by the.proposed procedures were not allowed to put their ideas in writing or make presentation to the Rule 57 Advisory?Committee.? Unlike this committee, the Rule 57 Advisory Committee refused to allow public presentations at its meetings. I believe the failure to adequately offset the uncertainties in the risks estimated according to the methodologies recommended for adoption by the Rule 57 Advisory Committee is a fatal flaw that must not go uncorrected. Larry E. Pink 6300 v. Michigan Ave. P-24 Lansing, Mi. 48917 Richard Powers Chairman. Rule 57 Advisory Committee Office of Toxic Materials Control Michigan Department of Natural Resources 9/24/82 P.0. Box 30028 Lansing, Mi. 48909 Dear Mr. Powers: What follows are my recommendations, criticisms and questions regarding the document: ?Proposed Surface Water Quality-Based Effluent Limitation Derivation Procedures for Chemical Substances" dated 6/23/82 --and public noticed for sixty days beginning 7/23/82. Teresa Kent has indicated that the comment period closes on 9/24/82. I trust my come ments will receive due consideration. - I. Recommendations A. General 1. The draft procedures should be divided into two sets of-procedures. Steps relating to the review of the NPDES Permit Application and the development of toxic substances monitoring requirements and effluent limits should be promulgated as rules per the requirements of the Administrative Procedures Act (APA) under the Part 21 Administrative Rules of the Water Resources Commission (WRC) governing Wastewater Discharge Permits. Steps relating to the derivation of a toxic substance Water Quality Criterion or Standard should be promulgated as rules per the requirements of the APA under the Part 4 Administrative Rules of the NRC governing the physi- cal, chemical and biological quality of Michigan surface waters. The language of STEPS XVI should describe the procedures contemplated in far greater detail. Phrases such as "reasonably expected to be present", ?adequate wastewater charac- terization data", "no significant potential for ex- posure," "adverse impact to aquatic,.terrestrial or human life" and nomencalture such as "pre-screening wastewater monitoring study? and fprocess characteri- zation study" should be better defined and clarified so that regulated industry and the public are pro- vided with a clearer understanding of the tasks in- volved in carrying out each of the steps. 3. Because of the far-reaching implications of these - draft.procedures for the public health, safety and Powers 9/24/82 P- 2 welfare and for the protection of the natural -resources from pollution, impairment and destruction; and becauSe the proposed.state'regulatOry action meets one or more of the criteria defining a "major state .action? in Part 4 A of the Guidelines For the Prepa- ration and Rev ew of Entironmental Im act Statements .under Executive Order 1974-4 11 2 5 :and 'ecause such an'activityfi??not731empted CommissiOn.Policy Im sot Statements? Preparation.and Review (1/1/77g. I fOrmally request that an Environmental Impact state? ment (sis) be developed and published by the DNR for the proposed procedures according to the guidelines.set forth in the above referenced documentse 4. Following publication of the EIS and appropriate revisions of the proposed procedures in response thereto, I recommend that a.series of public hearings be held at which to take public testimony on the preposed procedures; 5. The-Attorney General should rule as to.the constituw tionality and lawfulness.of'prOCedures for implemene tation of a Wastewater discharge regulatory program which authorize: the discharge of-substances which will signifidantly adverSely affect the public and environmental health with the protections afforded the people under the 5th and 14th Amendments'to the-U.S.-Constitution, the National ?Entironmental Protection Act; Article 4, Section 52 of the 1963 Michigan Constitution and contrary_to the national policy set forth in the Federal Clean Water Act, the state policy set forth in the Michigan Environmental Protection Act and Section 6(a) of the-Michigan as'e; amended ~Specifics 1. STEP I. a. establish criteria-an NPPES Permit Application must meet to be-considered complete; b. promulgate the Michigan Addendum to the Form 2c Consolidated Permit Application as a rule ac- cording.to the requirements of the.APA; promulgate the list of substances regulatable under the~authority of the Michigan Water Resources ?Commission Act as a rule according to the require- ments of'the POyers 9/24/82 Po 3 d. establish criteria of time, location. durationm replication,?frequency and type of sampling; sample preservation; sample storage; sample preparation and sample analysis for the deter- mination of wastewater characterization adequacy and promulgate the criteria as a rule according to the requirements of the? II as establish the criteria for requiring the per?. . formance of a process? characterization study and promulgate the criteria as a rule per the requirementslof the b, establish criteria for requiring a prenscreening wastewater monitoring study and promulgate the criteria as a.rule per the requirements of the . ca- establish criteria of quantity and concentration in consideration of toxicity, biOaccumulation potential and persistence which determine whether a water qualitywbased effluent limit must be considered.and.promulgate the cri.teria as a 'rule ?per.r the requirements of the APA. STEP a. identify the' most sensitive populations at risk for purposes: of protecting the _receiving water for each designated use. STEP . as establish phfsiCal, chemical, biological and toxicological property criteria a substance must meet to- -have no significant potential for ?expOsure in water resulting in adverse impact to aquatic, terrestrial or human life; b. identify site-specific physical, chemical, bio? logical and technological conditions of the re? ceiving watershed which must be taken into account in deleting a substance from further considera? . STEP VII a. develop a tier testing scheme and a set of screening criteria so that staff of the DNR are authorized to call fOr more testi. data if warranted on the basis of review of the minimum test data, and . -.-. . . .o s' - . .gu. Powers 9/24/82 13.4 promulgate those criteria as a rule per the requirements-of the ABA. 6. STEP 1x a a. adopt appropriate models to account fer potenw tial additivity and nergism in multiple substance exposures. see for'example, Cala- mari, D. and Alabaster (1980), Chemosphere 9: 533.- 538 carry out a statistical analysis of the data used to develop the adjustment factor so as to define the_99% confidence interval; define the acute/pseudo- chronic adjustment factor as chronic axis intercept4bf'a ray perpendicular to and intersecting a ray drawn perpendicular to and passing through the +99%-confidence' contour line at that pointe Adjust this, factor by.a factor of ten to-take into ac? count.the probability that some of the 28- day embro-larval higher than the corresponding life cycle prohibit the net discharge of human carcino- _gens, hereditary-mutagens, teratogens, imu munosuppressants and sterilogens to any sure face water which is or-may'become lically connected.w1th any surface water that is-or may become a drinking water supply, sport or commercial fishery, agricultural water supply or full or partial body contact recrea- tional area; compliance-with this prohibition shall be established by passing the influent and effluent through a sorptive.cclumn in? volumes_sufficient to concentrate the.substance to a degree equivalent.to the repreSentative worst?case bioconcentration factor. either measured or estimated, if measured values are unavailhble.- prohibit the discharge of unregulatablelgeno- toxic substances, defined to be those substances fer which the estimated risk - .concentration is so low that it is non-detectable in the edible portion of the most significant consumable bioconcentrating adpatic organism. constrain the sum-of all anthropogenic genotoxic effects not to exceed - Powers 9/24/prevail upon the legislature to adopt the 1/1,000. 000 for all other permit effluent limits for genotoxic substances, if and only if reasonable and prudent .alternatives, including closed system use and carbon pre- filtering for BAT, are not technologically feasible or economically achievable. STEP XIV See: Tung, YeounKoung and L. Hr Mays (1980) "Risk Analysis for Hydraulic Design" J. draulics Div. ASCE 8?3 - 9f3 and Warn9 and.J.S. Brew (1980) "Mass Balance", Water Research V14 1427 1434 STEP XVII i as -use preJoncentration sampling to extend the working'range of analysis sufficiently low to detedt regulatable concentrations.. STEP e. when;- thelambient concentratibn of a chemical substance from anthropogenic sources in the receivin'g water exceeds its WQC value, staff shall recOmmend that a point and nonpoin.t source mass balance wasteload allocation be developed for the impacted watershed; b. should the limit prove to be unattainable or economically unachievable, require such studies on the receiving water and the sub? stance :2 =r water to permit - the modeling_ of t_he assimilative capacity ,of the water. Powers 9/24/82 p. 6 II. Criticisms A. Procedural 1. Enter into the public record copies of-letters to Robert'Courchaine, Executive Secretary, NRC, dated 9/14/81 and 11/10/81 and 12/12/81 regarding deficiencie - in.the.composition, constitution and operations of the Rule 57 Advisory Committee. B. IntroductiOn 1. On Page-4 of the-Introduction under "Objectives and PhiloSophy of the Committee" it is stated: It is often diffiCult to completely separate . science=?rom value judgments. Wherever possible, the_basis (science or value-judgment) for de? cisions and recommendations is identified." Unfortunately, a.number of hidden value judgments twere-made by the Committee which are nowhere discussed, rationalized or justified. Such hidden value judg- ments include those embodied in the decisions not to take-into_account the statistical uncertainties in the data from which the acute/chronic adju?tment factor was calculated and from which the-11100.000 cancer dose-reaponse-relationship was estimated. ?Another hidden judgment involved the decision.not to considernthe-great potential for.additive and synergistic effects resulting from multiple_substance exposures. The proposed procedures allow DNR staff to take.these effects into account only when test data are-available to Justify such a-Concern. Yet it is;a mathematical impOSSibility to test even a few of the-pdssible combinations of the various substances on EPA's Priority Pollutants list; let alone the vast number of substances in commerde. The presumption must be that synergistic effects must exist? As such appropriate safety could have been factored in to the NOAEL and cancer'riskrbased.effluent limits to protect the public health and the natural resources with a reasonable degree-of statistical confidence. On page 5-of the Introduction under "Objectives.and Philosophy of the_Committee" it is stated that: The Committee tried to develop recommendations which.w0uld adequately protect the public health and the environment while placing a minimum cost burden on Michigan industry or municipal discharge: Powers 9/24/82 Po. --7 0n*Page.6 of that same section it is stated:; The Committee has not considered in detail the economic?impacts of implementing these recommendations The Committee was not charged with .the responsibility and did not- have the tee nical expertise to conduct an economic feasibility analysis. It should be noted that the Committee attempted to ?balance the costsland benefits of a Particular degree of public and environmental health protection even though it lacked the necessary expertise. Further, such cost-benefit{considerations are inconsistent with the Michigan environmental protection mandates. - Article 4. Section 52 of the 1963 Michigan Constitution, the Michigan Environmental Protection Act and the WRC Act. This point has been made time and time again to the NRC by staff cf the Department of the Attorney General acting as counseI to the WRC. In my August 20, 1981 statement to 'the. NRC, which. is part of the public rem cord, regarding irregularities in the method of appoint? ment. to and defici encies in the composition and cons-- -;titution of th.e Rule 57 Advisory Committee, I requested the eao'e supportifor my recommendation that staff of DAG explain to the Committee members the COnstitu- tiOnal and statutory mandates of the agency they wer.e to advise. At the time Stu Freeman indicated that he would be willing to talk to the Committee if asked. This request was repeated in my letter of 9/14/81 and reiterated in .my letter of 12/12/81. Questions'. 1. -I have attached alcopy of a letter sent to you in November of 1.981? containing questions regarding the . permit*program in?general and then draft STEPS I - IV. Since STEPS I IV have not changed significantly from the initial to final draft; and since I have been un- able to determine'if my questions were ever discuSsed, because the minute's of the meetings were incomplete and the agendas did; net set aside time for discussion of public comment, I am reiterating those questions here. I have also attached additional questions covering STEES XIX. Regarding Appendix it is net clear whetheriyou will regulate substances which are genotoxic by the inhala- tion route but inadequately tested as to their oral and dermali ;;.genotoxicity as oral and dermal.genotoxio substances. How do you intend to handle this situation? Powers 1 .9/24/82 p. 8 Questions 3. ?What is your rationale for choosing the design weight as the 70 kg man? Given that.the young are generally more sus- ceptible to genotoxic effects than the mature, and given the fact that half the population of this state is female, wouldn't it have-been more-prudent to use the weight.of a ten year old girl as the design weight for purposes of regulating genotoxic substances?_ What is the fish consumption rate distribution function for the Michigan population? What are the 95th., 99th- and 99.9tn percentile rates of fish consumption(Kilograms/day) for the Michigan population? How was the.6.5 gram/day fish consumption rate derievd? - Note: 6.5 should be -0065 Kg Thank you for this opportunity to comment. I appreciate your - consideration of my viewse. . photocopies: Sincerely, sf: - Larry Holcomb, TSCC - F, git/?? Larry . Fink, Citizen l. m" lb? Printed in the USA. l'ergurtim l'rw. Ltd. [IUII-Tlh?l?l?llillh? I Copyright SETAC Hazard Assessment ECOSYSTEM RISK ANALYSIS: A NEW METHODOLOGY R. V. . R. H. GARDNER. L. W. G. W. SUTER. S. G. AND C. W. GEHRS Environmental Sciences Division Oak Ridge National Laboratory Oak Ridge. TN 37830 {Received l0 October 1981: Accepted [5 December l98ll Abstract A method is presented for extrapolating laboratory toxicrty data to aquatic ecosystem effects such as decreased productivity or reduction in game fish biomass. The extrapolation requires - translating laboratory data into changes in the parameters ot an ecosystem model. the Standard W\terCOIumn The translation isefleeted through knovt ledge of toxicological modes 0! aetion, The uncertainties assocmted with both laboratory measurements and extrapola- tions are explicitly retained. and risk estimates are given in the form of probabilities that an effect could occm The approach is illusuated b\ scenarios in which effeets of toxic substances are distributed across different trophic levels. Each scenario population interactions in different waits and alters both the level and the nature of the risks to ecosystem processes. Particular attention is paid to analyzing the interaction between toxicitt and the uncertainties associated v. ith extrapolation Keywords Gamel?ish biomass At present. accurate predicnon of the effects of toxic subSIances 0n ecosystem properties is impossible. One begins with a limited set of laboratory data composed of toxicity tests on single species. These mm are poorly designed for extrapolation to the field because the mortality of organisms in a laboratory system may bear little resem- blance to their response in the natural ecosystem From these laboratory data. one wishes to draw conclusions about effects on socially relevant endpoints which 'To whom correspondence may be addressed. Presented at the First Annual Meeting. of Environmental Toxicology and Chemistry. Arlington. Virginia. November 24?25. 1980. Aquatic Risk anaIySts Organic involve complex interaCtions in the eco- sy5tem. which themselves are far from well understood. DeSpite the intrinsic diffi- culty of the task. there is a critical need to estimate pctential impaets. The need is par- ticularly critical for evaluating new techno- logies such as fuel processes. The apparent impossibility of developing an Ecosystem Risk Analysis is not an unfamiliar challenge to science or society. The need to predict weather and projeCI economic trends has long been realized. even though predictions cannot be made with great accuracy or reliability. [n such cases. an effort must be made and then judged on whether it represents the best ?State-of-the-art? approach available. One of the approaches available for mak~ ing uncertain predictions is risk assessment. Io? 163 R. V. O'NI-tt'l ct?ttl. One begins by defining the endpoints of interest. In ERA. the endpoints are socially relevant changes in ecosystem processes such as productivity or ability to support economically important species. Risk assessment then provides techniques for estimating the probability of that endpoint being reached. given all of the uncertainties involved. In this paper we will outline a concept of ERA designed to predict higher order eco- logical effects based on laboratory tests of toxicity. The approach incorporates a num- ber of innovative elements designed to apply the best ?state-of-the-art? techniques to the problem. The techniques include ecosystem models. propagation of uncer- tainties with Monte Carlo simulation. and the use of expert Opinion to fill in critical gaps. We describe each of the components and present a series of calibration studies to test the sensitivity of the method. Finally. we'apply ERA to examine the relative'toxi- city of two organic substances: phenol and quinoline. OF ECOSYSTEM RISK ANALYSIS In ERA (Fig. I) laboratory data must be- translated into changes in parameter values of a Standard WAter COlumn Model This translation involves projecting mortalities. or other standard :ls. ?at-mu Isl) ?59$ "t?sePalm-Jinn; 0? sauna} :S'thm 3' Fig. l. A schematic illusnation of the flow of informa- tion leading from laboratory data to estimation of risk expressed-as a frequency distribution of effects. responses. 'into changes in model parame- ters associated with processes such as graz- ing or reSpiration. The eitpected change in each parameter value is expressed by an element of the Effects Matrix. E. This step ?requires knowledge about toxicological modes of action. Extrapolation of labora- tory data involves considerable uncertainty- Estimates of the uncertainty associated with each extrapolation are also included in E. Using the elements of and their associ- ated uncertainties. SWACOM estimates effects on the aquatic ecosystem by Monte Carlo simulation At the beginning of each run. all parameter values are chosen from the statistical distributions described by E. For example. a grazing rate of.0.5 and a corresponding matrix entry of l;0 20 percent is assumed to describe a Gaussian distribution with a mean of 0.5 and a coeffi- cient of variation (100 standard devia- tionfmean) of 20 percent. A random number is chosen from this distribution and the process is repeated for each parameter. Then? a single, one-year simulation of the model is made. A new set of parameter Values is chosen and another simulation per- formed. The procedure is iterated a large number of times until sufficient informa- tion is obtained to describe the frequency distribution of results. In this way. effects on endpoints of interest can be stated as the probability or risk of the event occurring. The uncertainties of extrapolating labora- tory data to the field have become state- ments about the . uncertainty of an undesirable effect. THE STANDARD WATER COLUMN MODEL The biological and physical details of SWACOM are important in determining the frequency distribution of results. How- ever, it must be understood that Eco?sysrem Risk Analysis is not limited to any single model. Another model can replace SWACOM without altering the remaining elements of the technique. The present I Fig. 2. A: ard WAter light. ant SWACOM phytoplanl carruvorou model. 1 CLEAN been des SWAI the pelat consider. zooplanl top can trophic It matical eter? val populatit photosyt stant, ture Opti The a environr The tern soidal cu 4? in 1 follows reduced add nu Nutrient umn in phytopla cesses ar reminere fall turn from the the wate light, ten can mar-mun ?smears 0 sea i us.? I I 5? Hm HM c.5? cm- mm- mm 0 360 Fig. 2. A schematic illustration of SWACOM [Stand- ard WAtcr'COlumn Model]. Daily levels of nutrients. light. and temperature serve as model input. SWACOM considers the trophic relationships of IO phytoplankton. 5 zooplankton. 3 forage fish. and single carnivorous fish Species. - model, SWACOM, is a simplification of CLEAN Portions of the model have? been described previously SWACOM (Fig. 2) is designed to mimic the pelagic portions of a lake ecosystem. It considers 10 phytoplankton populations, 5 zooplankters. 3 planktivorous fish, and a top carnivore. The populations in each trephic level are described by similar mathe- matical functions but with different param- eter values. Thus, each- phytoplankton pepulation is characterized by its maximum rate, light saturation con- stant, Michaelis-Menten constant, tempera- ture Optimum, and susceptibility to grazing. The abiotic driving variables mimic. the environment of a northern dimictic lake. The temperature describesan annual sinu- soidal Curve with lake turnover occuring at 4? in the spring and fall. Radiant energy follows a similar curve, with light greatly reduced under ice cover. External sources add nutrients each day of the year. Nutrients are removed from the water col- umn in stoichio?metric relationship with phytoplankton growth. Decomposition pro- cesses are not considered in the model, and remineralization occurs only at Spring and fall turnover when-regenerated nutrients from the hypolimnion are added back into the water column. Phytoplankton growth in response to light, temperature, and available nutrient is Ecosystem risk analysis 169 Idescribed Ein detail in Self-shading effects are. accou:nted for by integrating photosyn-. thesis over the euphotic zone. Each phyto- plankton! pepulation has an Optimal temperature at which its rate is maximum. Total fixation of biomass is primarily limited by available nutrients . which are exhausted in periods of rapid growth. Grazing or predation at all trophic levels is described by a nonlinear interaction unc- tion This function considers both lim- ited food supply and competition with other grazers. The consumer populations are lim- ited by their individual metabolic and mor- tality rates and by predation from higher trophic levels. Both grazing and respiration rates are affected by temperaturewith each population characterized by an optimal temperature. A typical simulation run of SWACOM is shown in Fig. 3. To simplify presentation, biomass for.each. trophic level has been summed. iover the individual populations. Growth begins in the Spring after the ice cover breaks up at about'day 80. Turnover occurs after day 100, significantly increasing the concentration of available nutrients. As nutrients are depleted. phyto- plankton biomass decreases rapidly. Zoo- plankton biomass reaches a peak about one month after the phytoplankton, and forage fish follows about a month later. Game fish biomass-is quite small, relative to the scale of the figure, and is?not shown. Fall turnover, just. before day 320, increases available DIG u- no": tilII. srauoano IITEH cow-m uooEL ?ranOPLan fen I BIOMASS E: I I ?must: nun-Ems 1 I Bus" 1 I a a mess. -a 0 4O EIO IZO USO 200 240 280 320 360 cars or rat: run Fig. 3. A typical simulation of SWACOM. showing seasonal dynamics of phytoplankton. zooplankton. and forage ?sh. Values shown on the graph are summed over the component pOpuiations. 170 R. V. 1. et at. nutrients and causes a minor increase in phytoplankton. The figure shows that the model is capable of reasonably complex seasonal behavior that follows the general pattern of northern lakes. The basic structure of per- mits it to simulate a number of higher order effects. Toxic effects which differ across the phyt0plankton p0pulations. can cause com- petitive release. Increased mortality in one population results in increased nutrient availability for the remaining populations. Thus. overall productivity may be unaf- fected even though one or more individual populations are strOngly affected by the tox- icant. Similarly. toxic effects on forage fish decrease grazing on zooplankton. prevent- ing phytoplankton blooms. It is this type of system-level effect which will be of concern in Ecosystem Risk Analysis. 'Parameterization of SWAC OM has been greatly simplified since we are primarily interested. at present. in devel0ping the methodology. The model currently simu- lates the behavior of an arbitrarily defined assemblage of populations. rather than Spe- cific taxa. This approach will suffice for present purposes but would. of course. need. to be improved before the model is applied for actual assessment purposes. EFFECTS MATRIX The parameters of the model can be writ- ten? as a matrix with rows corresponding to each of the 18 pOpulations. and columns correSponding to parameters (6 for each phytoplankton population and 5 for each consumer]. The Effects Matrix, E, corre- sponds element by element to the parameter matrix and is used to express the effect of a toxicant on each parameter. During Monte Carlo simulations, each parameter will be multiplied by the corre- sponding element of E. If an element is 1.0 (no toxicity]. then ?the parameter is unaltered. If the element of is greater than 1.0, the toxicant will increase the rate repre- sented by the parameter. If the value is less than 1.0. the effect will be to reduce the ratef The advantages of using an Effects Matrix are that different arrays of toxic effects can be imposed without changing the model or its basic set of parameter values. The uncertainties caused by extrapolat- ing laboratory data to the field are expressed as error terms associated with each element of- E. Thus. the information: would expect a ll) percent increase in grazing rate. and I would have a confidence in that extrapolation of 20 percent" would tran- slate into a matrix element of i- 20 pt?rcent. This is sufficient to describe a Gaussian distribution for that element from which random values may be drawn: Extrapolating laboratory data now becomes a matter of choosing values and uncertainties for the Effects Matrix. At this point. information is required on the mode of action of the toxicant. suggesting a direct relationship between concentration and changes in physiological processes represented in the model. An example might be bioassays of effects on maximum svixnaomc It is possible to describe a General Stress that simplifies the task of extra- polating laboratory results to changes in model parameters. For instance, a general stress response by phytoplankton might involve lower maximum rate. increased reSpiration. lower light satu- ration point, and increased Michaelis- Men- ten constant. Applying the concept to consumers, whether ZOOpIankton or fish. involves similar assumptions. Grazing rate- may be decreased, respiration increased. temperature Optimum lowered and b0th mortality and susceptibility to predation increased. However, caution must be_ exerted in applying such an approach because a toxicant may have additional physiological effects which would have to be taken into account. For example, some toxicants are narc0tic and reduce reSpira- tion rate: the purpt assumed I TheG some inte Matrix. photosyn about hc decreasec' rate will ment of l. but nm This ir by use of If an elei standard. and a approxim would be bution t: x. lying p-lp-xl. would be a distribt above the Before attention ior and se particular tive contr effect of lation. Th tion of bo large valu ity that that is no tions but - show an here the involved environm about itSE cause is d; To exz cases was model we tion rates. rather than increase them. For the purposes of the present study we have assumed the stress de?ned above. The General Stress introduces some interesting constraints on the Effects Matrix. If a toxicant reduces maximum there will be uncertainty about how much will be decreased. However, it is certain that the rate will not be increased. Thus, that ele- ment of can take on values between 0 and l. but not above 1.0. This information is included ?folded normal distribution" If an element of is equal to 0.8 with a standard deviation of 10 percent li.e.. 0.08l. and a Gaussian distribution is assumed. approximately 0.6 percent of the values would be above 1. The? folded normal distri- burion transforms the random number. lying beyond an arbitrary point. p. by p-lp-xl. In our example, a value of 1.02 would be transformed to 0.98. The result is a distribution with one-half of its values above the mean. bUt no values above 1.0. TOXICITY AND UNCERTAINTY Before any new technique is applied. attention must be paid to its general behav- ior and sensitivity. In the case of ERA it is particularly important to compare the rela- tive contribution of direct toxicity with the effect of uncertainty resulting from extrapo- lation. The risk esnmated in ERA is a func- tion of both factors. For a toxic chemical. a large value for risk implies a high probabil- ity that the effect will occur. A chemical that is not toxic at the observed concentra- tions but has large uncertainties in may show an equally high value for risk. But here the probability expresses the risk involved in releasing the chemical into the environment. given that we know so little about its effects. The risk is still real. but the cause is different. To examine this problem. a set of test cases was run in which all parameters of the model were assumed to be affected equally. Ecosystem risk analysis l?l Assuming the General Stress described above. all parameters were changed by the same percentage. For exam- ple. a 5 percent effect was represented in by either 0.95 or 1.05. depending on whether the effect was to increase or decrease the parameter. For each level of toxicity, different uncertainty levels. rang- ing from 1 to 20 percent. were tested. Test results for several endpoints are summarized in Fig. 4. The uaxis of each graph shows the risk of reaching an end- point. such as a 20 percent reduction in 200plankton biomass. The axis expresses toxicity as the percent change in parame- ters, and the axis shows percent uncer- tainty. Figure 4a shows that increased toxicity li.e.. percent change in parameters of all populations! increases the risk of a 25 percent reduction in forage fish biomass summed over the year. Considering the 1 percent uncertainty case. there is little risk at 5 or 10 percent toxicity; however. at 15 and 20 percent toxicity. the risk of reducing forage fish biomass rapidly increases to As uncertainty increases. at any level of toxicity. the risk of reaching the endpoint also increases. For example, at 0 percent toxicity, the risk becomes 0.40 as uncer- tainty approaches 20 percent. lRemernber that 0 percent toxicity means that we do n0t think an effect will occur. but 20 percent uncertainty means that we are nor at all sure). Therefore. there is still some risk involved in releasing the subStance to the environment. The situation reverses at 20 percent toxicity where uncertainty reduces the risk from LG to 0.8. In Other words. we Fig. 4. Response of forage fish phytoplankton and zmplankton biomasses to the effeCt of a toxic substance (X axis]. uncertainty of effect axis} and frequencv of response axis]. 172 R. V. cl al. believe that the chemical 'will change model processes by 20 percent. but again. we are far from secure about this prediction. This uncertainty translates into a decreased risk. The results for forage fish biomass (Fig. 4a) follow the expected pattern: risk of an undesirable effect increases with toxicity and also with uncertainty; however, other patterns are possible. A 20 percent reduc- tion in phytoplankton biomass is insensitive to toxicity (Fig. 4bl. At any given level of uncertainty. the graph is relatively flat from left to right. showing that increased toxicity does not necessarily increase the risk of a reduction in biomass. Because of nutrient limitation. increased stress does little more than change species composition As dominant species are affected by the toxi- cant, other populations. previously kept down by competition for nutrients, are now able togrow more successfully. even under some toxic stress. The result is that total phytoplankton biomass. summed over all populations, changes very little. On the" other hand, increasing uncertainty. at any given level of toxicity has. the expected effect of increasing. risk. The complexity of interacrions in the model clearly affects the risk associated with a 20 percent reduction in zooplankton biomass (Fig. 4c). Increasing uncertainty has the eXpected result of increasing risk; however. the interaction of uncertainty and toxicity is more complex. At 5 percent uncertainty, the highest risk occurs with 5 percent toxicity; at 7? percent uncertainty, the peak shifts to 10 perCent toxicity; and at higher uncertainties, the peak shifts to 15 percent toxicity. The decreasing risk at higher toxicity results from nonlinear relationships in the model. As all populations in the system are affected, predation on 200plankton decreases. At some combinations of toxicity and uncertainty, the zooplankton seem to prosper, and its food supply, phytoplank- ton, remains relatively constant (Fig. 4b). At the same time, forage fish biomass is decreased (Fig. 4a). The result is lower risk of significant reductions in 200plankton' biomass. These tests demonstrate that uncertain- ties have an important. sometimes domi- nant. effect on risk. In some cases, expected toxicity is less important than our uncer-' tainty in applying results to the ?eld. This result emphasizes that the risk associated with releasing a toxicant may be dominated by our lack of understanding rather than the actual toxicity of the chemical. A second result is that toxicant effects are difficult or impossible to predict without the intervention of an ecosystem model. Significant impacts on phytoplankton may have little effect on ecosystem undtion in a nutrient limited system. On the other hand, small toxic effects may result in greater risk for 200plankton populations than for larger effects. ECOSYSTEM ENDPOINTS Assessment problems require that spe- cific undesirable effects be defined as end- points that are. relatively independent. Multiple endpoints are of little value if the analysis indicates a high risk for all of them. Different endpoints should show significant risks in response to different impacts. The sensitivity of ERA endpoints was tested by imposing an arbitrary 15 percent toxicity with a 1 percent uncertainty on different trophic levels or on various combi- nations of trOphic levels (Table 1). Each risk calculation is based on 500 Monte Carlo iterations. No two endpoints show Signifi- cant risk at exactly the same combination of cases. Additionally, there is a real difference in the sensitivity of the endpoints. End- points associated with fish (decrease in forage fish, game fish, and the ratio between them) respond to most of the cases. This is consistent with our understanding of aquatic ecosystems, since higher trophic levels often show greater sensitivity to impacts. On the other hand, a 5 percent decrease in total productivity never occurred under any of the cases testedzoo 20% for: 50% I gan 5% dt tot: i (01.2 10% 50% fora ?Trop risk, It The re' predicted phytoplar are affecr respiratiOI (ZFG case of reduce: ratio whe although decreased increased. Althou not be of warrant their incl nonlinear impact is plankton reduced lated on reducing: ing the blooms. decreased since this similar pr Ecosystem risk analysis 173 Table I. Risk of specific ecosystem effects when toxicant shows an effect effect. 1% uncertainty) on various trophie levelsa 'Endpoinls I Trophic Levellsl- 1 ZFG PZFG 5% Increase In maximum phytoplankton bloom 0.30 20% decrease in - phytoplankton biomass 20% decrease in zooplankton biOmass 0.12 20% decrease in forage fish biomass 50% decrease in game fish biomass 5% decrease in total productivity 10% increase in total reSpiration 10% decrease in production to respiration ratio leRl 0.04 50% decrease in game to forage ?sh ratio 1.00 0.93 0.04 1.00 0.87 1.00 0.90 1.00 1.00 1.00- 1.00 1.00 0.95 1.00 0.58 0.47 0.99 1.00 .00 1.00 aTrophic levels indicated are the only portions of the ecosysrem affected: phytOplankton. zooplankton. forage fish. game fish. A missing value indicates an insigni?cant risk. less than 0.00 l. The results in Table i could not easily be. predicted a priori. There is a small risk of phytOplankton blooms if game fish alone are affected..There is a risk of increased respiration if all consumer populations (ZFG case) are affected. There is a small risk of reduced PIR ratio when zooplankton alone are affected, although productivity is not significantly decreased nor respiration significantly increased. Although some of the endpoints would nor be of sufficient interest to society to warrant emphasis in assessment problems, their inclusion emphasizes the complex; nonlinear dynamics of the model. When impact is isolated on zooplankton, phyto- plankton biomass increases because of reduced grazing. When the impact is iso- lated on game fish, forage fish increases, reducing zooplankton biomass and increas- ing the probability of phytoplankton blooms. Phyt0plankton biomass is decreased when forage fish are reduced since this leads to increases in grazers. By a similar process. 200plankton biomass is decreasedieither by a reduction in its food supply (P case) or by a reduction in predator fish-which increases forage fish grazing pres- sure on zooplankton. Total respiration is increased in the F, G, ZF cases since all of these cases increase 200plankton biomass which is primarily responsible for total respiration. Table 1 indicates which endpoints to con- sider in further studies. The insensitivity of a 5 percent decrease in produetivity indi- - cates that this endpoint is unnecessary. The ratio of game to forage fish is redundant because it is dependent on the values of the two trophic levels. This endpoint will net be considered in subsequent analyses. Of course, both of these results should be veri- fied by field studies since they could be due to problems in the model. APPLICATION OF ERA TO TWO TOXIC CHEMICALS Having examined the sensitivity of the analysis to various test cases, we can now apply the method to a preliminary assess- 174 R. V. et al. ment. For this application we have chosen to compare two organic chemicals: phenol and quinoline. These chemicals are of par- ticular interest since they are among the toxic substances expected in the ef?uent of fuel plants. . Toxicity data (Table 2) on these chemi- cals are typical of what is available for assessment. The set of organisms tested is not completely relevant for application to our lake ecosystem assessment: however. we will attempt to use all of the information. Since we are attempting only a relative risk assessment. many assumptions can be . incorporated into the analysis as long as we do not clearly bias the results in favor of one of the chemicals. Table 2. Results of laboratory tests of acute toxicity for two organic chemicals?. Values are mg Organisms Phenol Quinoline Green Algae (Salaries-(rum) .25h Blue-green algae (Microcysn's) 54" Daphm?a magma 30 28.5 Chironomid i018 57.2 Snail 258.4 183.1 Fatheadminnow 23 1.5 Rainbow trout 0. l5 ll. ?The data are 48' h. LCm's lconceniration that kills 50% of organisms: Milleman. Pers. Comm. These values are EC 33?s. that is. concentrations that cause a 20% reduction in Giddings et al.. in press. We have assumed' that both chemicals are present in the environment in concen- trations that result in a 5 percent effect on rainbow trout. The relative toxicity or": - other organisms is calculated by dividing their LCgu?s (or ECwi into that of rainbow trout and multiplying by 5 percent. This maintains the same spectrum-ofsensitivities as indicated by the laboratory tests. it is necessary to identify the test organ- isms with populations represented in the model. The green alga. Scienastrum. tends to bloom in the late spring and corresponds to phytoplankton population 5 of SWACOM in many of its physiological reSponses. The bluegreen alga, is able to grow at low nutrient concentra- tions and may bloom in the summer. This general behavior identifies the with population 10 Populations 6-9 were assigned intermediate sensitivities to the toxicant. Therefore. the range between the percent effects on populations 5 and 10 was divided into; equal intervals and assigned to populations 6-9. We assumed that the measured values on populations 5 and [0 were known with 5 percent coeffi- cients of variation. For intermediate p0pu-- lations. we assumed that the uncertainty of assigning effects increased and therefore assigned uncertainties ranging from 10 per- cent to 20 percent. For" the remaining popu- lations, we assumed a. minimal effe?t (0.00011 but with a 30 percent uncertainty. Turning to the five zooplankton popula- tions. we assumed that Daphnia magnet cor- responded to population 2 with a 5 percent uncertainty. We used the data on the Chiro- nomid as an indicator of the sensitivity of another invertebrate consumer-in the sys- tem. widely different taxonomically from Daphm?a. That is. we assumed that the dif- ference in sensitivity between Daphnia and the Chironomid indicated the range that might be expected among invertebrate con- Sumers. Therefore. we assumed that popula- tion 5, the population most different from Daphnia in other physiological parameters, corresponded in sensitivity to the Chiro- nomid. This extrapolation was assigned a larger uncertainty of 10 percent. The remaining populations were assigned per- cent changes intermediate between Daph- nia and Chironomid with uncertainties of 20 percent. The fathead minnow was ?identified with the third forage fish population and assigned an uncertainty of 5 percent. We assumed that the snail represented the oppo- site end of the range of sensitivities for macroscopic consumers in the system and assigned its sensitivity to the first popula- tion with a CV of 10 percent. The second population was assigned an intermediate toxicity and a 20 percent uncertainty. l. i .i I I Finally. without effect?ar It is method however defensib ever. an for deve complex such as: equal ir. forage fi: been vet tainties; tiOn cou accuracy dependet ing differ might at isms. Ne the decis menting It only con the two- ested ins of either - no ERFECT case 510:: as?? 6 3 6 I I Fig. 5. Da line. Al. zo- line, BI. ant cases: no tc caIs are ass tions to pr biomass. [Ecosystem risk analysis Finally. the rainbow trout was identified with our game fish and assigned a 5 percent effect and given a 5 percent uncertainty. it is obvious that _such an arbitrary method of assigning values for the matrix. however internally consistent. would not be defensible in an actual assessment. How- ever. an arbitrary approach should suf?ce for devel0ping the methodology. Other less complex scenarios might be envisioned. such as assuming that all zooplankters were equal in sensitivity to Daphnia and all forage fish to the fathead minnow. We have been very conservative in assigning uncer- tainties; much larger coefficients of varia- tion could be justified based on the typical accuracy of ecological data. We have also depended exclusively on the General Stress defined earlier rather than seek- ing differences in the way the two chemicals might affect the physiology of the organ- isms. Nevertheless. the example illusnates the decisions which must be made in imple- menting the approach. It should be remembered that we are only concerned with the relative toxicity of the two chemicals. Thus. we are less inter- ested in an accurate prediction of the effects of either chemical than in comparing the Tut' IO- IID limit?: 1003003Fig. 5. Daily total biomass of phytoplankton tsolid line. At. zooplankton Idashed line, At. forage fish tsolid line. Bi. and carnivorous fish [dashed line. BI for three cases: no toxicants. phenol. and quinoline. Beth chemi- cals are aSSumed to be present in sufficient concentra- tions to produce a 5 percent decrease in game fish biomass. 175 chemicals within the context of the scenario we have developed. Such a comparative study provides the assessor with consider- able flexibility. Details about the assump- tions are less important than avoiding anything that biases the results in favor of one chemical or the other. The deterministic time behavior of the model li.e.. ignoring the effects of uncer- tainty! is shown in Fig. 5. Compared to the ?no effects? case. phenol causes an increase in phytoplankton. particularly in the late Slimmer. Forage fish biomass is increased. and game fish biomass is decreased. Quinoline causes a reducnon in the phyto- plankton bloom and a signi?cant increase in the zooplankton. particularly during the late Spring and early summer. The forage fish biomass is dramatically reduced. com pared to the other two cases. Game ?sh biomass is reduced and barely visible along the abscissa of the graph. Table 3. Risk of undesirable effects on an aquatic ecosystem resulting from two organic chemicals1 Endpoint Trophic Levellst Phenol Quinoline 5% increase in maximum phyto- 0.16 0.05 plankton biomassl< 26.] ll 20% decrease in phytoplankton 0.16 0.4] biomass 276.8] 20% decrease in zooplankton 0.07 0.002 biomass l< 731.0l 20% decrease in forage 0.2] 0.84 fish biomassl< 257.4! 50% decrease in carnivorous 0.3] 0.88 fish biomasst< . increase in toral 0.32 0.78 respiration t> l03.3) decrease in producrivity: 0.37 0.82 respiration ratiotPlR .l6l Values are based on 500 Monte Carlo iterations. The risk of exceeding each of the end- points. based on 500 Monte Carlo iterations are listed in Table 3. The results show that quinoline poses a greater risk of undesirable reductions in fish populations and ratio, caused mainly by increases in consumer respiration. Phenol poses a greater risk of phytoplankton blooms and reduction in zooplankton biomass. In general, quinoline [7'6 R. V. O?NtenJ. er :11. yields larger risks for effects on the aquatic ecosystem. Much of the increased risk appears to be traceable to the greater sensi- tivity of forage fish to quinoline (Table 2). The difference in sensitivity of this trophic level dominates the results. leading toa high probability that fish biomass will be reduced to an unacceptable level. As a result of reduced fish biomass, zooplankton biomass increases, leading to increased respiration and significant effects on the FIR ratio. Although results in the form of Table 3 represent typical output from the analysis, ERA produces a complete frequency distri- bution for each of the endpoints. The distri- butions of phytoplankton biomass (Fig. 6a) are very similar for the two chemicals. It is interesting that the deterministic solution for quinoline lies below the mode of the . distribution, whereas it lies above the mode in the distribution for phenol. The differ- ence between the deterministic solution and the mean of distribution represents the bias in predicting effects without including the effects of uncertainty The biases in Fig. 6 demonstrate the diffi- culties of predicting ecosystem effects with- out explicitly considering uncertainties. The frequency distributions for zoo- plankton biomass (Fig. 6b) show the curve for quinoline shifted to the right, predicting higher probabilities for large zooplankton biomass asindicated in Table 3. in this case the deterministic solution is below the mode for both chemicals. The distributions for PI ratio (Fig. 6c) show the curve for quino- line shifted? to the' left and. both deterministic solutions lying above the modes. In all cases in the figure there is significant bias. indicat- ing the problems involved in predicting ecosystem effects with a deterministic model. DISCUSSION This paper has presented a new meth- odology for extrapolating laboratory toxi- city data to risks of ecosystem effects. It is clear from Tables I and 3 that the ecosystem . it] 1' . 1.?hl: 0 Poor: woo soon 0 mo econ soon a I 2 a mum. an. ion mass :9th nu ma ?lm on: Fig. 6. Frequency distributions of effeCts of phenol (solid line} and quinoline (dashed linel on phytoplank- ton biomass (al, zooplankton biomass lb]. and PIR ratio lei. determined by 500 Monte Carlo iterations. The points on the lines indicate the deterministic solutions. effects calculated by ERA could not be pre- dicted without this form of analysis. The nonlinear interactions incorporated into models such as SWACOM are needed to predict the higher-order effects which result from toxic effects distributed over the vari- ous populations in the ecosystem. The limi- tations of the approach can be clearly identi?ed with the procedures employed in choosing the values of the effects matrix and their associated uncertainties. The test cases presented in this paper have illustrated that uncertainties of extra.- polation must be explicitly considered in any analysis of ecosystem risks. Figure 4 shows that uncertainties can dominate results of the analysis, producing significant risks even if no direct'toxicity is expected. On a pessimistic note, it must be added that only a fraction of the uncertainties involved in the prediction have been included in the analysis. There are uncertainties involved in the formulation of the mathematical model [1 l] and in the basic parameter set [12] used for the simulation. If all of these independ- ent sources of uncertainty were included in the analysis it is likely that our lack of knowledge about the dynamics of ecological systems would result in unacceptable risks, no matter what level of toxicity were being considered. This reinforces the argument offered 3] that we must be extremely con- servative about any societal operation which impinges on the ecosystem. It also emphasizes our inadequate scienti?c base for decision making in an ecosystem context. 'It is clear from the phenol and quinoline examp far sii attemt ing ass may bi and U1 single restrict paratit gained The shown predict tial errl effects tainty. predict advoca indicat of prob the chosen analysi the for necess: assessn Acknovi the Em agency Nationa gram with [ht Vii-7504 lication ORNL. example that comparative risk assessment is far simpler than an assessment which attempts to predict absolute risks. Simplify- ing assumptions that do not bias the results may be acceptable in a comparative analysis and unacceptable in the assessment of a single impacr. Thus. it may be advisable to restrict applications of the ERA to com- parative analyses until more experience is gainedwith the method. The Strong in?uence of uncertainties shown in Fig. 4 and the resulting biases in predictions shown in Fig. 6 show the poten- tial error in attempting to predict ecosystem effects with a method that ignores uncer- tainty. The use of deterministic models to predict ecosystem effects has often been advocated, but the analyses presented here indicate that calculation of risks in the form of probabilities is a more honest approach to the problem. Whatever specific method is chosen. the inclusions of uncertainty in the analysisand the statement of conclusions in the form of probabilities should become a necessary criterion for environmental assessments. Acknowiedgement? Research supported in part by the Environmental PrOtection Agenet under Inter- agency Agreement 40- 740- 78. and in part by the National Science Foundation?s Ecosysrem Studies Pro- gram under Intragency Agreement DEB-7725781 With the US. Department of Energy under centract W-7504-eng-26 with Union Carbide Corporation. Pub- lication No. 1994. Environmental Sciences Division. ORNL. I I . I Ecosystem risk analysis i I77 RFFERENCES . Lane. RA. and R. Levins. I977. The dynamics of aquatic 2. The effects of nutrient enrichment on model plankton communities. LiInnoi. Oceanogr. 22: 454 471. . 0' Neill R. V. and J. B. Waide. I981. Ecosystem theory arid the unexpected: Implications for envi- toxicology. In W. Cornaby. ed.. Toxic Substances In the Environment. Ann Arbor Science. Ann Arbor. Michigan. . Rubinstein. R. V. 1981. SimulationandtheMonte Curio Method. John Wiley and Sons. New York. . Park R. A. 1974. A generalized model for simulat- ing lake ecosystems. Simulation 23: 33- 50. . 0' Neill R. V. and]. M. Giddings. I979. Population interactions and ecosystem function. In G. S. Innis and R.V. O'Neill. eds. Systems Anab'sis of Ecosvisterns. International COOperative Publishing House. Fairland. Maryland. p. 103. . DeAngeiis. D. R. A. Gold5tein and R. V. 0? Neill 1975. A model for trophic interaction. Ecologt S6: 881 -892.- . Giddings J.M.. AJ. Stewart. R.V. O?Neill and R.H. Gardner. An ef?cient algal bioassay based on short-term response. In press. . Gardner. DR. 1975. Structure-Activity relation- ships of DDT analogs in non-target organisms. In 6D. Veith and DE. Konasewich. eds. Sympo- sium on Structure in Correlations in Studies of Toricit} and Bioaccumuiation Itith Aquanc Organisms Great Lakes Research Advisory Board. Windsor. Ontario. p. 77. . andS.S. Shapiro. 1968.5tatIZI'ticaiMod- eisin Engineering Wiley. New York. . GardnenR. H. D. D. Huff. R. V. O'Neill. J. B. Man- kin..l. H. [Carney andJ. Jones. 19802:. Application of error. analysis to a marsh hydrology model. Water Resour. Res. 16'. 659- 664. . Gardner R.H.. R.V. O'Neill. J.B. Mankin and D. Kumar. 1980. Comparative error analysis of six predator-prey models. Ecology 61 :323-332. . O?Neill R.V.. R.H. Gardner and J.B. Mankin. 1980. Analysis of parameter error in a nonlinear model. Ecoi. 8: 297- 31 I. . Holling C. S. 1973. Resilience and stability of eco- logicalsySIems. Annu. Ecoi. Stat 4:1-23 i o? II: Toxic Substances Regulation in Michigan's NPDES Wastewater Discharge Permit Program By Larry E. Fink, M.S. Director, The Foresight Society In response to the problem of mercury contamination of Lake St. Clair fish and the revelation that Michigan businesses could discharge vast quantities of mercury and other toxic substances to the state's waters without even having to report such discharges, The Michigan Critical Materials Register came into being in 1971 pursuant to The Truth in Pollution Act (1970 P.A. 200) which amended the Michigan Water Resources Commission Act Act), 1929 P.A. 245. Section 6b of the WRC Act provides for the creation of a committee of scientific experts to advise the Water Resources Commission (WRC) as to what substances to include on a register of critical materials. Michigan businesses dis- charging other than sanitary sewage directly or indirectly to the waters of the state must report the pounds of each critical material. used or produced and discharged each year on forms provided by the WRC. The first Critical Materials Register (CMR) was published in October of 1971 pursuant to Rule 323.1233 of the Administrative Rules of the NRC. The use/production and discharge of substances new to the CMR need not be reported until the following year, so the first Critical Materials Reports were not received until 1972. Responsibility for the CMR Program was transferred to the DNR by Executive Order along with other programs in 1974. The requirement to report the discharge of substances on the CMR is Michigan's early warning system when it comes to potential toxics problems. For example, in 1972 Dow Chemical Company's Michigan Division in Midland, Michigan reported the discharge of to the waters of the State. At that time it was known that was contaminated with that caused birth defects in mice and that was a potent human chloracnegen. By the law of conservation of mass was present in Dow's discharges. With the early warning given by Dow's 1972 Critical Materials Report, the dioxin problem could have been detected ten years ago. Act 293 of the Public Acts of 1972 amended the WRC Act to give Michigan the statutory authority equivalent to that established by the Federal Water Pollution Control Act Amendments of 1972 for the purpose of issuing NPDES permits. Section 7(1) of the WRC Act says, in part: After April 15, 1973, a person shall not discharge any waste or waste effluent into the waters of this state unless he is in possession of a valid permit therefor from the Commission. It also required all dischargers of Critical Materials to pay a surveillance fee in prOportion to the quantities discharged up to $8,000/year. The monies collected were to be spent on the monitoring of toxic substance contamination problems and wastewater discharges. (The Surveillance Fee requirement is being phased out by the legislature this year.) In 1973 Michigan WQS were revised pursuant to the requirements of the Water Pollution Control Act amendments (P.L. 92-500). It is on the basis of these Water Quality Standards (WQS) that water quality-based wastewater discharge permit effluent limits are developed. (Although the Clean Water Act requires the states to revise their WQS every three years, Michigan has yet to do so despite trying continually since 1976, because of organized industry opposition to tough toxics standards.) Pursuant to Section 402(b) of the Water Pollution Control Act Amendments (later the Clean Water Act) Michigan applied for and was granted authority to issue Michigan NPDES permits by the U.S. Environmental Protection Agency in 1974 under a state takeover agreement. Rules were adopted by the WRC governing application for; terms, conditions and duration of; drafting and public noticing of; hearings on; and modifica- tion, revocation and enforcement of Michigan NPDES permits. The first Michigan NPDES permits issued in 1974 were based on Refuse Act (Corps of Engineers) permit applications. The first round of 5-year NPDES permits issued by EPA or the states were designed to bring industrial dischargers into compliance with Best Practicable Treatment (BPT) effluent limits for conventional pollutants by July 1, 1977. (Conventional pollutants include acidity/ alkalinity dissolved solids [salts], suspended solids [particles], biochemical oxygen demanding substances and oil and grease.) Publicly Owned Treatment Works (POTWs) were to achieve secondary treatment capability by that same date. (Primary treatment involves gross filtra? tion and then removal of settleable solids in clarifiers; secondary treatment follows primary treatment with accelerated bacterial digestion in aerated tanks, then clarification; tertiary treatment removes toxic chemicals via carbon filtration or the equivalent.) Only a few industrial dischargers DiMichigan missed the July 1, 1977 BPT deadline (BASF Wyandotte Corporation?Wyandotte, for example). A number of Michigan municipalities failed to achieve secondary treatment capability by that date (chief among them, the Detroit POTW). Although nothing prevented Michigan from restricting the toxic substances content of wastewaters discharged under NPDES permit, in lieu of EPA regulations, with limited DNR staff inexperienced in reviewing NPDES permit applications and only a short time in which to issue thousands of permits, few of the earliest first round NPDES permits required monitoring for toxic substances and fewer still set toxics discharge limits. (For example, Dow Midland Division's NPDES Permit MI0000868 limited only phenolic compounds on the basis of their ability to cause taste and odor problems in fish. There were no dioxin limits or monitoring requirements.) with time DNR staff began to set discharge limits for heavy metals cadmium, chromium, copper, lead, mercury, nickel, silver and zinc), ammonia and cyanide. In 1976 Michigan adopted Act 60, which regulated the use and disposal of PCBs?containing equipment and wastes. Rules promulgated under Act 60 prohibited the discharge of PCBs to the waters of the state. 1. Although PCBs discharge limits were not written into NPDES permits, users of PCBs?containing equipment had to report the quantities of PCBs used and disposed of annually. Monitoring paid for with Surveillance Fees revealed extensive PCBs contamination in Michigan's streams and significant concentrations in some industrial discharges, especially paper mills and foundries. Also in 1976 the Natural Resources Defense Council sued EPA for failure to implementt?uatoxics control provisions of the WPCA in a timely fashion, forcing EPA to develop Water Quality Criteria for 129 toxic pollutants and to identify likely navigable waters toxic contaminant "hot spots." (The Saginaw River of itsconfluencewdth the Tittabawassee River was named a toxic "hot spot" by EPA in July 1981.) Following passage of the Federal Clean Water Act Amendments in 1977 (P.L. 95-217), with their new emphasis on toxics regulation, federal funds were made available to takeover agreement states to beef up the NPDES permit application review and effluent limit setting process. Created in that same year, DNR's Office of Toxic Materials Control began to address toxics problems and by the fall of 1978 had two full time staffers reviewing municipal and industrial permit applications for toxics problems. With access to information from Critical Materials reports, Michigan was able to identify potential toxic substances problems well ahead of other states. An NPDES permit can limit toxics discharges in two ways: (1) forcing a facility in a particular industrial category to install Best Available (Treatment) Technology Economically Achievable by the industrial category as a whole; and (2) calculating quantities in the discharge which will not cause the concentrations in the water to eXCeed respective Water Quality Standards, or in lieu of duly promulgated WQS, water quality criteria derived from available toxicity and bioconcentra- tion information. EPA missed its July 1981 deadline to establish BATEA guidelines, thus forcing states like Michigan to issue NPDES permits with water quality?based effluent limits in the interim. EPA offers guidance on how to set WQS, but it's up to the states to adopt WQS. Rule 57, Michigan's Toxic Substances Water Quality Standards, is presently undergoing revision. Procedures have been proposed for calcu- lating carcinogen discharge limits on the basis of an "acceptable" increased human cancer risk of 1/100,000, as well as calculating toxics levels safe for fish. These procedures have never been scientifically validated and have yet to be approved by the WRC. Dow?s Midland facility is contesting an interim NPDES permit issued to it containing acceptable cancer risk~based effluent limits, arguing that the permit limits are the product of ad hog rulemaking and overly restrictive. The Foresight Society is also contesting the permit, contending that it is too limited and too lenient, because it authorized Dow to increase its discharge of four carcinogens by seventeen times and did not make the discharge of up to twenty times existing dioxin levels a violation of the permit. The contested case hearing is scheduled to begin in early October. 1. Rather than detail the routes and schedules of permit application review, permit recommendations development and timetables for EPA comment, I would rather focus on how you as a concerned citizen can become actively involved in protecting the waters of the state from pollution, impairment and destruction. The following strategy is recommended for effective participation in the NPDES permit issuance process: Write William McCracken, Chief, Permit Section, Surface Water Quality Division, Michigan Department of Natural Resources, P.0. Box 30028, Lansing, MI, 48909 and ask: a. that your name be added to the list ofpersonsautomatically receiving all public notices of NPDES permits to be issued in your geographical region; for a cepy of the Michigan Water Resources Commission Act; for a copy of the Part 21 Administrative Rules of the Water Resources Commission governing the issuance of NPDES permits; for a list of major industrial and municipal dischargers to the river(s) of interest and their respective dates of expiration; for the procedures followed in reviewing NPDES permit applications and developing NPDES permit conditions. Write Rich Powers, Chief, Toxic Chemical Evaluation Section, Michigan DNR, P.0. Box 30028, Lansing, MI, 48909 and ask: a. for a copy of the policies, guidelines and procedures governing the calculation of water quality-based toxic substance effluent limits; that you be given a copy of the most recent publication explaining the Critical Materials Register (last published as Critical Materials Register 1980). One hundred eighty days prior to expiration of the NPDES permit of interest, under the authority of the Michigan Freedom of Information Act request from William McCracken: a. a copy of the facility's Form 2c Consolidated Permit Application and Michigan Supplement; if the facility is an industry, a computer printout of its Critical Materials Report for the previous year; if the facility is a Publicly Owned Treatment Works (POTW), a computer printout of the statistical summary of Critical Materials discharges to the POTW for the previous year; 10. Vt c. a?c of the report of the most recent wastewater survey co- cted at the facility. For those substances reported discharged in some concentrations in the permit application or the wastewater survey or in some quantities in the Critical Materials report, write William McCracken and ask him to indicate: a. concentrations of each above which an effluent limit will be developed; b. concentrations of each above which the discharge would violate state law. Review the Critical Materials Register book to identify cancer? causing (carcinogenic), birth defects-causing (teratogenic) and mutation?causing (mutagenic) substances being discharged. Write Mr. McCracken and ask him if there are any fish contaminant advisories for any toxic substances down stream of the facility. When the draft NPDES permit is public noticed: a. ask the local radio and TV stations to announce that the permit is on public notice as a public service; b. if the facility is a major discharger, review the Fact Sheet to determine if all toxic substances being discharged by the facility are listed. If not, Fact Sheet violates requirements of Rule Call Mr. McCracken and ask him to schedule a time within five working days of your request when you can review the NPDES Permit File and the Surface Water Quality Division File on the facility. When reviewing the files, look for Notices of Noncompli- ance, Notices of Violation, reported fish kills, 322' If the draft NPDES permit doesn't regulate toxic substances to your satisfaction, write William McCracken with carbon copy to the Executive Secretary, Water Resources Commission identifying areas of deficiency and proposed modifications that would make it acceptable to you. Be sure to mention the facility's history of non?compliance, notices of violation, foul odors, fish kills, in your letter. Encourage other concerned citizens to write Mr. McCracken, as well, indicating their dissatisfaction with the draft permit and requesting a public hearing pursuant to Rule 323.2130. Write the Executive Secretary of the Water Resources Commission and ask that you be scheduled for a presentation regarding "draft NPDES Permit MI_jpe?mit_?) for (facility name) public noticed on (date) Make a concise, direct 11. 12. pres to the Commission, identifying the significant defic 's in the permit and the changes you propose to correct them. A chronology of events indicating dates of Notices of Non?compliance and Violation, fish kills, is effective. Photographs of oil slicks, discoloration of the water at the point of discharge and fish kills made into slides have tremendous impact. Bring your friends and encourage them to participate. If the permit is issued over your objections and you feel strongly that the permit as is will not protect the public and the environ? ment from toxic substances contamination, prepare an administrative complaint and petition for a contested case hearing. The complaint must be signed, notarized and two copies filed within fifteen days of the date of issuance of the permit with the Executive Secretary of the Water Resources Commission. The Commission will probably consider DNR's recommendation to accept or reject the petition at the next regularly scheduled meeting. If the petition is accepted, the Commission will direct the Hearings Examiner, Mr. William Fulkerson, to schedule a contested case hearing. Following a delay determined by the number of contested cases ahead of yours, the contested case hearings will commence. Get a lawyer and prepare for a long, protracted battle. Solicit the support of environmental, public interest groups and law firms interested in doing BEE bono legal work. TOXIC MATERIALS CONTROL STRATEGIES FOR WATER QUALITY MANAGEMENT IN MICHIGAN BY: Larry E. Fink, Chemist .Office of Toxic Materials Control Michigan Department of Natural Resources If all toxic materials were manufactured, stored, shipped, formulated, packaged, used, and diSposed of properly, we would have a manageable toxic materials problem. Unfortunately, until relatively recently, the potential for injury to these and future generations of protected populations resulting from the release of toxic materials to the environment was often not considered in making toxic materials waste management decisions. Diluting a dangerous contaminant from a facility's process wastewater in the nearest river was considered by many to be an acceptable means of diaposal. Burying toxic wastes in fifty-five gallon drums in a nearby landfill was believed to be a safe way to get rid of untreatable process wastes. Burning municipal refuse in the local incinerator was a good way to reduce the tremendous volume of material to a manageable size. . For the public the wastes were out of sight and therefore out of mind. Only recently have peOple come to realize that there are limits to the quantities of pesticides, chemicals, trash, and smoke that can be released to the environment without consequence. Today we live with many of those consequences. SCOPE OF THE TOXIC SUBSTANCES 0f the sixty to seventy thousand substances commercially available in the U.S., hundreds have been identified in Great Lakes water at the parts per quadrillion and parts per trillion levels. Of particular concern are the heavily chlorinated benzenes, biphenyls, naphthalenes, phenols, styrenes, and terphenyls showing up in the tissues of fish and eggs of Great Lakes gull populations at parts per million concentrations. These materials persist for long periods in water and sediment because they are resistant to chemical and biochemical processes of degradation. They also bioaccumulate in the fatty tissues of aquatic and terrestrial organisms. Since peeple eat Great Lakes fish, these materials are also bioaccumulating in people. What is perhaps of greater concern are recent findings identifying chlorinated benzenes, biphenyls, dioxins, phenols and styrenes in samples of human semen. What effect these contaminants have on the ability of the human race to procreate is open to Speculation, but it brings the toxic materials problem close to home. DEFINITIONS Before we go any further, it is necessary to define terms: Toxic Substance: a substance capable of inducing harmful changes in the structure or function of dosed organisms or their off5pring. Hazardous substance: in the context of toxic materials management a Substance is hazardous if it possesses physical, chemical, and/or biological properties which: 1. cause it to persist and accumulate in the environment to levels which increase the likelihood of exposure; 2. cause it to contaminate living systems as a result of exposure; 3. cause it to increase the incidence of harmful effect in a pepulation of organisms which it contaminates. Environmental hazard assessment/hazard evaluation: a methodology for quantifying the potential for harm resulting from the physical, chemical, and toxicological prOperties of a substance released to the environment. :4 Risk: the possibility of realization of harm resulting from exposure to a hazardous substance or circumstance. . Risk assessment: a procedure for identifying potentially exposed populations and the probable routes of exposure to a potentially harmful substance or circumstance and for estimating the incidence of harm realized in a population as a result of the integrated incidences of exposure via each route. Safe: a substance or circumstance is judged to be safe if its risks are acceptable. Acceptable risk: A risk to which society is willing to be exposed in return for social, aesthetic, or economic benefits. ACCEPTABLE RISK There is no absolute safety. In everything there is the potential for harm. There is no risk free environment. Something is judged or determined to present an acceptable level of risk. Acceptable risk is a subjective concept. What is an acceptable risk for one may be unacceptable to another. Risk, Hazard, and the Slippery Road Driving on an icy highway is illustrative of applied hazard evaluation and risk assessment. Given icy driving conditions, individual drivers perceive the hazards of driving differently. Some drivers slow down to'a crawl, while others zip by at normal or near normal Speeds. Each driver weighs the risks of Spinning out and the benefits of making good time in the context of his or her perception of driving skills, performance history of the car on icy roads, traffic density, etc. A wide variety of drivers in a wide variety of autos will "solve"?fhe risk/benefit equation differently, resulting in a multitude of Speeds chosen, all considered safe, given the drivers' perceptions of the hazards of the icy road. ?Page 3 watching the movie "The Deeruunter" gave me an idea for an analogy useful in explaining risk/benefit assessment. Suppose you play a game of Russian Roulette using a revolver containing one million chambers. A computer assigns or fails to assign a bullet to each chamber as a decision process algorithm which accesses a random number generator subroutine. The gun is loaded according to the assignment of bullets to chambers in the cylinder of the revolver. The player can choose the following play strategies: Risk and Russian Roulette bullet in the gun: .01 per bullet 10 bullet in the gun: .1 per bullet 100 bullet in the gun: 1. per bullet 1000 bullet in the gun: 10. irper bullet 10,000 bullet in the gun: $100. per bullet If the player blows her/his brains out, she/he loses everything, including the pot. A thousand bullets at $10 per bullet looks like a pretty ggod gisk/benefit investment strategy--that's odds of blowing yourself away thousand, with a payment of $10,000. But what do we do when the value of a bullet is unknown or the number of bullets in the revolver are unknown? Do we proceed ahead on the assumption that things can't be that bad, that the benefits will always outweigh the risks,_or do we refuse to play until we more clearly perceive the risks and benefits associated with the game? Acceptable Risk and Toxic Materials Control Strategies in the Home Now many of you are parents, and you are concerned about the welfare of your children. Since accidental poisoning is one of the leading causes of child injury and death in the home, you take care to keep your caustics and cleaners and bug Sprays out of the reach of children. From sad experience poison control centers know that the greatest hazard exists where the cleaners, pesticides, and caustics have been transferred to an unmarked or wrongly marked container. So, to minimize the risk, you make sure containers are properly marked and you make sure to tell your child not to drink anything out of any container until he or she checks with mommy or daddy. But the good judgment you excercise in protecting your child from poisoning in the home could be extended to protecting yourself and your family from environmental exposure to poisons. Just as we don't drink from unmarked bottles containing undefined liquids, so we do not release to the waters of the State uncharacterized effluents containing undefined materials. Unless we know what it is that is being released to the environment and the physical, chemical, and toxicological properties which define its environmental hazard potential, we are playing Russian Roulette with the lives of these and future generations of papulations potentially exposed, without knowing how many bullets are in the gun. i Page 4 Risk Assessment and the Acceptable Risk Decision Process One can perform a risk assessment without determining what an acceptable degree of risk is. What degree of risk is acceptable is not a question to be answered by processes of scientific reasoning and research. What degree of risk is acceptable is a question to be answered by those put at risk, usually through their elected representatives or through the courts. Hazard Evaluation What properties of a substance or mixture should we look at to evaluate its environmental and public health hazards? Acute Toxicity Carcinogenicity Hereditary Mutagenicity Teratogenicity Persistence Bioaccumulation Aesthetics Chronic Adverse Effects Acute toxicity results from short-term exposure to relatively high concentrations of a toxic substance or mixture, usually resulting in death. Chronic Adverse Effects result from long-term exposure to low levels of a toxic substance. Such effects include organ degeneration or malfunction, changes in behavior, inability to propagate due to sterility, and increased susceptibility to diseases and parasites. Carcinogenicity is a property of toxic effect exhibited by substances capable of inducing changes in somatic cells resulting in the develoPment of cancer. cratogenicity is a property of toxic effect exhibited by substances capable of causing birth defects or birth deformities. Persistence is a property of a substance with a tendency to resist environmental degradation, measured in terms of the time required for it to be completely transformed to inert or innoccuous forms by natural physical, chemical, and biological processes. Bioaccumulation is the process of preferential absorbtion of a toxic substance by living organisms, directly, by contact with a contaminated environmental compartment, or indirectly, by consumption of contaminated food, or both. Aesthetics considers the effects of a substance on the taste, odor, color, or appearance of a contaminated environmental compartment. Hereditary Mutagenicity is a preperty exhibited by a substance which causes changes in the genome of an organism which is inherited by successive generation of the organism. TOXIC MATERIALS CONTROL IN MICHIGAN: GOAL STATEMENT To protect public health, wild and domesticated plant and animal life, and the physical environment from the hazards resulting from the use, discharge, transport, disposal or spillage of materials or substances which have the capacity, through physical, chemical, or biological properties, to pose a substantial risk of injury. Page 5 TOXIC MATERLALS CONTROL IN MICHIGAN: CONTROL OBJECTIVES To limit the loss of toxic and hazardous materials from the economy to the environment to quantities which can be assimilated by the environment without resulting in an unacceptable risk of injury to human health or the physical or biological components of the environment. To eliminate the loss from the economy to the environment of persistent, bioaccumulative, and toxic materials which cannot be assimilated by the environment without resulting in unacceptable risk of injury to human health or the physical or biological components of the environment. The goals, objectives, and activities of toxic and haZardous material control in Michigan are outlined in the booklet: Michigan Environmental Protection Bureau Toxic and Hazardous Material Management Program. But the focus of our discussion is on-the water resource and the control of toxic substances for water quality management in Michigan. TOXIC MATERIALS CONTROL STRATEGIES The strategies of toxic materials control Open to the State of Michigan in managing the quality of its water resources are constrained and limited by the following: 1. 2. 5. 6. 7. Enabling Legislation goals and objectives Rules goals, objectives, and standards policies, procedures Outlined explicitly or implicitly in the rules Permits regulations and guidelines milestones and performance timetables Programs structure personnel plans Treaties and Agreements IJC Great Lakes Basin Commission Judicial Decisions Funding kinds quantities Although much could be said about the need for expanded enabling legislation to fill gaps in the State's administrative authorities and the woeful lack of funds allocated for the implementation of toxic materials . control programs, our discussion will focus on the protection of water quality from adverse effects of toxic substances-under existing and recently proposed programs impacted by Proposed Rule 57 - Toxic Substances Water Quality Standards. Page 6 PERMIT PROGRAM FOR POLLUTION CONTROL Although Federal and State laws were evolved to address environmental contamination by individual environmental compartments, clearly the toxic materials problem spans all environmental compartments and attempts at categorizing and compartmentalizing the toxic materials problem is artificial and ultimately self-defeating. Recently EPA has issued proposed Consolidated Permit Regulations which incorporate air, water, and land waste diaposal control programs into one program. A Consolidated Facility Permit application form has been deve10ped which should substantially reduce the redundancy and duplication of effort resulting from the existing compartmentalized approach. The State of Michigan is presently developing a Consolidated Permit Program Facility Evaluation protocol which will meet the needs and requirements of the Consolidated Permit Program. :4 NPDES PERMITS a Until the Consolidated Permit Program is implemented, the State of Michigan must rely on the National Pollution Discharge Elimination System (NPDES) Permit Program created under P.L. 92-500 - the Water Pollution Control Act Amendments of 1972 in the water pollution control efforts. Facilities desiring to discharge wastewater to the waters of the State must obtain an NPDES Permit. The permit sets forth the conditions of flow and composition which each allowable discharge must meet. When compliance with permit conditions cannot be achieved immediately upon issuance of the permit, a schedule of performance is included in the permit conditions designed to bring a facility into compliance in a timely and reasonable fashion. Facilities which fail to meet permit conditions according to their schedules of compliance are subject to civil and criminal penalties. P.L. 95-217 - the Clean Water Act of 1977 - placed greater emphasis on the control of the discharge of toxic substances to the nation's waters. In order to carry out the mandate of the act, the Office of Toxic Materials Control reviews and evaluates information contained in the NPDES permit application, Water Quality Division files, and Wastewater Report files, identifying Michigan Critical Materials and other toxics used or produced and discharged by a facility in order to develop a list of toxic materials known to be or reasonably expected to be present in wastewater effluent at concentrations of concern. Initially, only effluent monitoring is required, except in those cases where a toxic materials problem has already been identified in the permit application or via routine or Special surveys. The results of monitoring are reviewed and final permit conditions are develoPed, limiting the discharge of substances which can be adequately assimilated by the receiving water and eliminating the discharge of substances which cannot be adequately assimilated by the receiving water. Effluent limits are designed to permit the receiving surface or ground water to meet all Water Quality Standards apprOpriate to its designated uses. The implementation and enforcement of toxic substances discharge limitations in NPDES permits is considered one of the State's most important tools in protecting the quality of its Surface and ground water resources. Page 7 WATER QUALITY AND WATER QUALITY STANDARDS For purposes of our discussion we will define water quality as a state of the physical, chemical, and biological condition of the water. Water quality must be evaluated in the context of the designated uses of water resources: -drinking water supplies -agricultural water supplies -full body contact recreation -partial body contact recreation ?warm and cold water fisheries ~commercial and other For flowing or standing surface waters designated warm or cold water fisheries, the health of the fishery can be evaluated according to the following criteria: 1. Diverse aquatic pOpulations survive and prOpagate in natural balanced proportions. 2. Diverse terrestrial wildlife papulations which feed on aquatic life or which are fed upon by aquatic life survive and propagate in natural balanced prOportions. 3. The incidence of cancer, genetic mutation, birth defects, and sterility in aquatic and terrestrial papulations does not result in an unacceptable risk of papulation extinction. 4. The risks of cancer, genetic mutation, birth defects, and sterility in human papulations exposed via sport or commercial fishing or hunting are acceptable. 5. The water is free of aesthetically unacceptable sensible proPerties such as taste, odor, or color. 6. Protected game species are free of aesthetically unacceptable - sensible proPerties such as taste, odor, color or form. To evaluate water quality for acceptability and healthfulness according to these criteria, important physical, chemical, and biological parameters must be identified and ranges and ratios of the parameters indicative of acceptable and healthful water quality determined. The boundry between the acceptable and unacceptable for each indicator parameter becomes the standard of water quality for that parameter. But we are not only obliged to protect water quality to a level deemed acceptable, we are obliged to improve the quality of the water resources where feasible. A Water Quality Standard is an indicator of a minimum acceptable level of water quality. EXISTING STANDARD: RULE 57. TOXIC SUBSTANCES Toxicity of undefined toxic substances not Specifically included in subrules (2) and (3) shall be determined by development of 96 hour TLm's or other appropriate effect end points obtained by continuous- flow or in situ bioassays using suitable test organisms. Concentrations of undefined toxic substances in the waters of the state shall not exceed Page 8 safe concentrations as determined by applying an application factor, . -based on knowledge of the behavior of the toxic substances and the organisms to be protected in the environment, to the or other appr0priate effect and point. (2) For all waters of the state, unless on the basis of recent information a more restrictive limitation is required to protect a designated use, concentrations of defined toxic substances, including heaVy metals, shall be limited by application of the toxic substance recommendations contained in the chapter of Freshwater Organisms, "Report of the National Technical Advisory Committee to the Secretary of the Interior, Water Quality Criteria, 1968", or by application of any toxic effluent standard, limitation or prohibition promulgated by the administrator of the United States Environmental Protection Agency pursuant to section 307(a) of the United States Public Law 92- 500. (3) In addition to the standards prescribed in Subrules and (2), waters of the state used for public water supply shall, at the point of water intake, not exceed the permissible inorganic and organic chemicals criteria for raw public water suppiy in: "Report of the National Technical Advisory Committee to the Secretary of the Interior, Water Quality Criteria, 1968", except that chlorides shall be limited to the same extent as perscribed by Rule 1051(2). NEEDS FOR REVISION OF RULE 57 Rule 57 - Toxic Substances Standards was last revised by the Water Resources Commission six years ago. The needs for revision now include: 1. existing language employs outdated terminology and incoroprates by reference outdated Federal Regulations. 2. existing language is unclear 3. existing language unnecessarily limits the water Resources Commission's scape of authority in carrying out the toxic and injurious substances control requirements set forth in Section 6(a) of the WRC Act of 1929 and the NPDES Regulations promulgated pursuant to the Water Pollution Control Act Ammendments of 1972 and the Clean Water Act of 1977. Preposed Rule 57 was develoPed to be consistent with the goals and objectives of toxic materials control for water quality management in Michigan. OVERVIEW OF PROPOSED RULE 57 Any material can be toxic to some form of life when present in the environment in sufficient concentrations, quantities or combinations. Thus, our concerns cannot be narrowed to any lists of priority pollutants or register of critical materials. The environmental health risks posed by a particular material cannot be estimated unless potentially exposed populations are identified; the duration and extent of exposure is evaluated; and the physical, chemical, biological, and ecological preperties of concern are measured or approximated. PHILOSOPHY OF RULE 57 The uncontaminated natural environment is the status quo. Presumption rests with the uncontaminated natural environment. There will.be no contamination of the environment without due consideration for the risks incurred by pepulations exposed to the contaminating material. The burden of proof of acceptable risk rests with those who intend to contaminate or who have contaminated. PROPOSED STANDARD: RULE 57. TOXIC SUBSTANCES "As a goal, the waters of the state shall not contain substances in concentrations, quantities, or combinations which have adverse effects on aquatic, terrestrial, avian, or human pepulations. The presence of carcinogens, mutagens or teratogens, in quantities which are determined by the Commission to represent an unacceptable level of risk, is prohibited in the waters of the State. (2) The presence in the waters of the State of a substance or mixture of Substances of undefined physical, chemical or toxicological prOperties which results in detectable contamination of the waters of the State, sediments underlying those waters, organisms inhabiting those waters, organisms which drink those waters or feed upon organisms inhabiting those waters, is prohibited. The Commission shall determine the physical, chemical, biological and ecological tests to be performed, and who should perform these tests on such substances or mixtures, to generate the information required by the Commission, for the determination of concentrations or quantities estimated to be safe for aquatic, terrestrial, avian, and human populations and for the determination of the capacity of the receiving waters to assimilate such substances. (3) For all waters of the State, outside of established mixing zones, concentrations of defined toxic substances, including heavy metals, shall be limited by application of the more restrictive of the following toxic substances recommendations: By the application of any toxic effluent standard, limitation or prohibition promulgated as of the effective date of these rules by the administrator of the U.S. Environmental Protection Agency pursuant to F.R. 2588 (January 12, 1977) and 42 F.R. 6532 33.323. (February 2, 1977). By the application of the toxic substances recommendations contained in the chapter on Freshwater Organisms, "Report of the National Technical Advisory Committee to the Secretary of the Interior, water Quality Criteria, 1968". By the determination of limiting concentrations of chemically defined toxic substances based on the available scientific data base, including, but not limited to toxic substance recommendations contained in chapters entitled, "Freshwater Aquatic Life and Wildlife" Section and "Agricultural Uses of water" (Section V) of "Water Quality Criteria, 1972" and criteria published as of the effective date of these standards pursuant to Section 304(a) of P.L. 92-500, as amended by P.L. 95-217. i??age 10 (4) Waters of the State protected for public water Supply source _shall not exceed the maximum contaminant levels for inorganic and organic chemicals Specified in National Interim Primary Drinking Water Regulations, 40 CFR, Section 141.11 and 141.12,promulgated prusuant to Section 1412 of the Public Health Service Act, as amended by the Safe Drinking Water Act, P.L. 93?523." EXPLANATION OF PROPOSED RULE 57 Rule 57(1): 'As originally worded Rule 57(1) prohibited the presence of carcinogens, mutagens, and teratogens in the waters of the State. The Water Quality Standards Task Force was uncomfortable with this wording, recognizing the economic, socia1,:and political impracticalities involved in total prohibition, and, after a series of revisions, evolved the wording: presence . . . in quantities which are considered by the Commission to represent an unacceptable level of risk is prohibited Incorporation by reference of a specific risk assessment methodology was judged by the Water Quality Standards Task Force to unnecessarily restrict the state in carrying out the intent of Subrule (1). Since Subrule (1) requires a formal determination of a concentration or quantity at and below which the risk of injury to exposed papulations is acceptable, the risk assessment methodology or methodologies employed and the data base accessed could be reviewed, evaluated, and challenged during the course of the hearings before the Commission. Rule 57(2): Contamination *Contaminate - to make impure or corrupt by contact or mixture. Contamination - the state of being contaminated. Contaminant - that which contaminates. In the context of water quality management, a contaminant is any substance which is present in unnatural forms, quantities, or concentrations. Those materials which are present in natural forms at natural or background concentrations are not contaminants. As an example, mercury in the sediments is not a contaminant until it is measured to be present in concentrations above those considered _to be natural or background in a Specific segment of the receiving water. Further, the presence of suSpended solids in the water column would not be considered contamination unless the concentrations exceeded natural background levels for those Specific flow conditions enc0untered in the receiving water during a particular season or period. *American Heritage Dictionary of the English Language, Houghton Mifflin, Boston, 1976. Page 11 Detectable Contamination There can be no potential for injury to aquatic life, if the substance is not present in the aquatic ecosystem. For those substances found to be present in detectable concentrations, contamination cannot be established until natural or background concentrations of the substances are established. It is only that concentration or quantity in excess of the natural or background concentration which is contamination. For those substances which are not naturally present in the receiving water, the detection of any concentration or quantity shall be considered contamination. The detection of contamination is not limited to a method or instrument of chemical analysis. Biomonitoring methodology can also be employed. Intent of Rule 5762): Where information sufficient to define the physical, chemical and toxicological prOperties of a substance or mixture is not available to the Water Resources Commission, and where detectable contamination of any environmental compartment results from the presence of the substance in the waters of the State, that presence is prohibited. Existing scientific information and methodologies are insufficient for the deveIOpment of a model which would permit the prediction of the physical, chemical, and toxicological preperties of environmental concern from the chemical structure of a pure substance with acceptable precision and accuracy. However, any substance is or may become injurious to exposed protected populations when present in sufficient concentrations, quantities, or in particular combinations. Since the degree of risk of injury to protected populations when exposed to some concentration or quantity of the substance cannot be estimated without minimal information on the physical, chemical, and toxicological proPerties of the substance, the presence of the substance in one or more compartments of the environment above natural or background concentrations places exposed protected papulations at some increased but inestimable risk of injury. In order to determine if the increased risk of injury to exposed protected papulations is acceptable, the risk must be estimable. Where the acceptability of risk cannot be evaluated, the presence in the waters of the State of a substance which may become injurious is,prohibited. The language of proposed Rule 57(2) replaces the language of existing Rule 57(1), which limited testing to be required by the Commission to a narrow definition of bioassay only. The new language of Rule 57(2) allows the Commission to require physical, chemical, biological, and ecological testing of substances or mixtures of substances, including wastewaters, to reveal those properties of the substance or mixture which define its potential for injury when present in the waters of the State. - Page 12 Assimilative Capacity The wording - And for the determination of the capacity of the receiving water-to assimilate such substances was added to Rule 57(2) so that the need for information on the transport, partitioning, storage, release, and fate of toxic substances in the various compartments of the watershed would be considered explicitly. Once a long-term safe concentration is estimated, it is necessary to calculate the mass rate of loading of the substance to the receiving watershed which results in a steady state concentration less than or equal to the long-term safe concentration in every segment of each receiving water. Each discharger can then be allocated its fair share of the watershed assimilative capacity for each substance it discharges according to a formula consistent with the riparian doctrine. Some who reviewed the proposed Rule 57(2) expressed concern over the use of the term "assimilative capacity" in connection with toxic substances, arguing that materials cannot be assimilated like natural materials for which the river has evolved physical, chemical, and biological assimilative processes. While this may be true for some substances, it is not true for all. It is the intent of Rule 57(2) to require a substance to be tested so that the physical, chemical, biological, and ecological prOperties which determine the capacity of Michigan waters to assimilate the substance are revealed. When information sufficient to define these properties for a particular substance is unavailable to the Commission, the presumption is that the receiving water assimilative capacity for that_substance is zero. This is consistent with the concept that materials foreign to natural water systems are not readily assimilated by them until proven otherwise. Under such a preSumption, conservation of toxicant mass and dilution with receiving water flow as a function of location on the receiving water can be modeled in a straight forward fashion. - When information sufficient for calculating the rate of toxicant assimilation by important routes is available, the mass of toxicant available to the water column will change with time and location of the receiving water. The modeling of toxicant concentration as a function of location on the receiving water then requires much more saphisticated modeling technologies. CONCLUSION PrOposed Rule 57(1) recognizes the need to limit carcinogenic, mutagenic, and teratogenic substances in the waters of the State to those quantities which present an acceptable risk of increased incidence of toxic effect in protected papulations. Proposed Rule 57(2) provides a mechanism for evaluating the potential for injury resulting from detectable contamination of the environment with inadequately tested substances or mixtures after the fact and prohibits the continued presence of such substances or mixtures until adequate testing permits estimation of a quantity or concentration resulting in an acceptable risk of injury to exposed pepulations. Page 13 . But, it is not enough to prohibit the presence of a substance or nmixture of undefined properties whose release to the environment causes detectable contamination, because the detectable contamination may result in irreversible harm to exposed populations. Ultimately the release of a substance or mixture of undefined properties to the waters of the state must be prohibited. Such a prohibition is seen to be a necessary complement to preposed Rule 57(2), if the objectives of toxic materials management in Michigan are to be achieved in a meaningful way. When the burden of proof is shifted from the State to the discharger and transformed from proof of harm after the fact to proof of acceptable risk before the fact, there can be no discharge of uncharacterizable, uncharacterized, or inadequately characterized substances or mixtures, including wastewaters. Without such a shift and transformation of the burden of proof, the State will continue to manage its toxic materials by crisis, resulting in unnecessary and unacceptable contaminated of its natural resources with substances of undefined properties which may become injurious to exposed sensitive populations. The single biggest task facing the agencies of State government appointed to administer the State's environmental quality laws is the deve10pment of a mechanism for evaluating the risks which result from the release to the environment of materials which are or may become injurious to life. The development of such a risk assessment methodology should be given the highest priority. The Toxic Materials Lesson If we have learned nothing else from the Love Canal and PBB disasters, we have learned one thing: we can no longer afford to wait until we have epidemiologically validated proof of harm after the fact before we take action to control the release of a particular toxic material or mixture of toxic material to the environment. The risk that in making no decision and in taking no action we condemn these and future generations of protected populations to unacceptable increases in the incidence of genetic mutation, birth defects, cancer, and sterility is real, if unquantifiable. Can we afford to take the risk of assuming that that presently unquantifiable risk is acceptably small? I think not. The National Pollutant Discharge Elimination System (WPDES) Wastewater Discharge Permit Program 1? By Larry E. Fink, Director The Foresight Society With the Clean Water Act (CWA) before Congress for reauthorization this year, attention is now focused on its adequacy. While minor revisions of CWA have been prOposed, including those dealing with nonpoint sources of pollution and the clean-up of pollution "hot spots," the language of CWA is considered by many environmentalists to be generally sound. Unfortunately, the same cannot be said for the United States Environmental Protection Agency's implementation enui enforcement of the water pollution control programs brought into being by that language. One of the most important of these programs is the National Pollutant Discharge Elimination System (NPDES) under which wastewater discharge permits are issued. Only by effectively participating in the development of NPDES permits can we be assured that the physical, chemical and biological integrity of our surface waters will be protected. Such participation requires an understand- ing of the NPDES Permit Program and the procedures used by the State of Michigan to issue NPDES permits. In Part I, following a review of the history, philosophy, goals and policy of the CWA, I will summarize the sections of CWA which define the NPDES Permit Program. In Part II I will describe Michigan's NPDES permit issuance process in some detail so that the reader will be able to have a direct and meaningful impact in reducing and eventually eliminating water pollution. Part I: Toxic Pollutant Discharge Regulation History and Statutory Mandates Prior to passage of the Federal Water Pollution Control Act Amend- ments of 1972 (P.L. 92-500), which created the NPDES Permit Program, the regulation of pollutant discharges was governed by one of two statutes. To prevent interference with navigation, Section 13 of The Rivers and Harbors Appropriation Act of 1899, commonly referred to as The Refuse Act, prohibited the discharge from ships or on-shore facilities of "any refuse matter of any kind of description whatever other than that flowing from streets and sewers and passing therefrom in a liquid state." The Army Corps of Engineers was given the authority to administer and enforce the Act. Such authority included the issuance of discharge permits. Unfortunately, the Corps rarely exercised such authority. For purposes of protecting a body of water for drinking water supply, swimming, fishing, agricultural irrigation and/or industrial water supply uses, the Federal Water Pollution Control Act (FWPCA) of 1948 allowed courts to require reductions in pollutant discharges only after ?giving due consideration to the practicability and to the physical and economic feasibility of securing abatement of any pollution proved." In 1965 Congress amended FWPCA to require the states or, in lieu of state action, the Secretary of the Interior, to establish water quality standards for interstate waters. While FWPCA required the development of water quality standards and allowed the government to sue to abate pollutant discharges where those standards were being violated, when more than one facility was discharging to a polluted stretch of water, it was often impossible to accurately quantify the degree to which each such discharger contributed to the pollution. This hampered government efforts to correct pollution problems. Further, facilities discharging to water with correspondingly large capacities to assimilate sewage and toxic wastes enjoyed a competitive 2 42:; over facilities discharging to small bodies of water. Following mbunting pressure to get tough with polluters, on De--u--r 23, 1970 President Richard M. Nixon issued Executive Order 11574 directing the Army Corps of Engineers to adopt a discharge permit program under the authority of the 1899 Refuse Act. Applications for these permits began to be regeived by the Corps in the fall of 1971. However, growing awareness that existing water pollution control laws, with their emphasis on water quality standards to regulate water pollution, were unworkable, forced Congress to pass The Federal Water Pollution Control Act amendments of 1972, marking the birth of the modern era of pollutant discharge regulation. The Act's objective is to "restore and maintain the chemical, physical and biological integrity of the Nation's waters." Toward achieving this objective the Act adopts as a national goal that "the discharge of pollutants into the navigable waters be eliminated by 1985; and establishes as a national policy that "the discharge of toxic pollutants in toxic amounts be prohibited.? Section 402 establishes the National Pollutant Discharge Elimination System (NPDES) point-source wastewater discharge permit program. NPDES permits issued to facilities discharging wastewater to surface waters must contain conditions which assure that the discharges meet all applicable requirements of Sections 301, 302, 306, 307 and 308. The Clean Water Act (CWA) amendments of 1977 (P.L. 95-217) made significant changes to Sections 301 and 307, placing a new emphasis on the regulation of toxic pollutants. Section 301(a) of CWA states: "Except as in compliance with this section and sections 302, 306, 307, 318, 402 and 404 of this Act, the discharge of any pollutant by any person shall be unlawful." This is a clear rejection of water quality standards-based pollution regulation. This section requires that point sources other than Publicly Owned Treatment Works (POTWs) implement Best Practicable Control Technology Currently Achievable (BPT) by July 1, 1977; that by that same date point sources achieve any more stringent limita- tion required to meet applicable water quality standards; that by July 1, 1984 point sources implement: 1) Best Available Technology Economically Achievable (BAT) for the toxic pollutants listed in Section 307(a)(1) and for any additional toxic pollutants, within three years of the establishment of effluent limitations; 2) Best Conventional Pollutant Control Technology (BCT) for conventional pollutants, biochemical oxygen demand (BOD), total suspended solids (TSS), 3) BAT for nonconventional pollutants-"grey area" pollutants not subject to toxic-BAT or conventional-BCT treatment. "Notwithstanding any other provisions of this Act it shall be unlawful to discharge any radiological, chemical or biological warfare agent or high-level radioactive waste into navigable waters." For POTWs in existence on July 1, 1977, effluent limitations based upon secondary treatment must be achieved. Per the authority of Section 302, if, in the judgment of EPA's Administrator, the discharge of pollutants from a point source or group of point sources in compliance with Section 301 effluent limitations "would interfere with the attainment or maintenance of that water quality in a specific portion of the navigable waters which shall assure protection of . . . [its uses] and the protection and propagation of a balanced population of shellfish, fish and wildlife . . . effluent limitations . . . shall be established which can reasonably be expected to contribute to the attainment or maintenance of such water quality." 3 tion 306 requires EPA's administrator to set forth new source .?for the control of the discharge of pollutants which reflects the a, effluent reduction . . . achievable through application of the ,{able demonstrated control technology, processes, operating methods, rlternatives, including, where practicable, a standard permitting ~1e of pollutant." A minimum of twenty-seven categories of sources . to which the above applies. Provisions for revisions to the ?1 standards and categories are also established. The Section 307(a)(1) list of substances to be regulated as toxic pollutants and for which effluent standards are to be established pursuant to Section 301 and 304 is the list developed by the Natural Resources Defense Council (NRDC) in the consent decree stemming from its successful 1976 suit to compel EPA to regulate toxic pollutants (NRDC vs Train). The so-called EPA Priority Pollutants include 65 chemical categories involving 129 individual substances. Each effluent standard is to be based on the physical, chemical, biological, ecological and toxicological properties of each of the substances on the list and each is to provide for an ample margin of safety. Pursuant to Section 304, in October 1980 EPA published water quality criteria for all but one of the 129 toxic (TODD). Water quality criteria, unlike water quality standards, are not enforceable. Rather they offer scientific guidance to the states in promul- gating water quality standards pursuant to Section 303. Section 307(b)(1) provides for "regulations establishing pretreatment standards for introduction of pollutants into . . . for those pollutants which are determined not to be susceptible to treatment by such . . . or which would interfere with the operation of such . . . Section 308 provides EPA's Administrator with the authority to: 1) require the owner or operator of any point source to keep relevant records, monitor effluents and "provide such other information as he may reasonably require;" 2) enter any premises in which an effluent source or records file is located; 3) inspect and c0py relevant records and 4) inspect monitor? ing equipment and sample such effluents as the permittee is required to sample. Section 402(b) provides that a state with equivalent statutory authority may assume primary responsibility for issuing and enforcing NPDES permits under a NPDES Permit Program takeover agreement. Michigan assumed responsibility for issuing NPDES wastewater discharge permits in 1974 under the equivalent statutory authority of the Michigan Water Resources Commission Act (Public Act 245 of 1929, as amended). The Part 21 Administrative Rules of the Water Resources Commission (Rules) govern the Michigan NPDES permit issuance process. The Part 4 Rules establish the state's water quality standards as required by CWA Section 303. Using the toxic substances stand- ards promulgated pursuant to Rule 323.1057, the state establishes water quality-based effluent limits in NPDES permits. Presently Rule 323.105? is undergoing extensive revision, with far?reaching consequences for the protection of the public health and the public trust via the state's NPDES permit program. The regulation of the discharge of toxic pollutants will be the focus of the discussion of the NPDES permit issuance process in Part II of this article. RISK ASSESSMENT AND RISK ACCEPTABILITY IN ENVIRONMENTAL PROTECTION COMMENTS BEFORE: DNR Environmental irotection Policy Advisory Committee 19 January 1983 DATE Mr. Chairman, members of the committee: I appreciate this opportunity to make this presentation before the DNR Environmental Protection Policy Advisory Committee. My written remarks detail my concerns. My name is Larry Fink. I hold B.S. Chem. degrees from the University of Michigan?Ann Arbor. The M.S. is in Environmental Health Sciences, with Specialization in Environmental Chemistry. For the past four years I have been employed by the DNR's Office of Toxic Materials Control, first in the capacity of reviewing Clean Water Act NPDES permit applications for toxic substances hazards and, more recently, in the capacity of evaluating estimator techniques and chemical environmental fate mathematical models for use in exposure assessment. Although I am an employee of the DNR, the views and opinions expressed herein are my own, as a private citizen and Director of the Foresight Society, and should not be construed to be those of my employer. Page Two HISTORY AND BACKGROUND I have been involved in the development of propose (Rule 57) of the Administrative Rules of the Water Resou sion (WRC) since September 1979, first as a DNR represen with the WRC?appointed Water Quality Standards Advisory Task Force (WQSTAF) and later, following fundamental and irreconcilable differenCes of opinion which arose between me and my supervisors in DNR, as a pri- vate citizen. One of the recommendations made to the WRC by the WQSATF at its June 1980 meeting concerned the need to uniformly and consistently define carcinogenesis, hereditary mutagenesis and teratogenesis; establish a method of calculating genotoxic risk and adopt an accept- able "de minimus" genotoxic risk level. This recommendation was strongly supported by representatives of the private sector, who felt the language of proposed Rule 57 was so broad that it could not stand alone without rules establishing how the general narrative standard was to be translated into site-specific numerical standards. In res- ponse DNR staff indicated a willingness to develop guidelines for such purposes. Private sector representatives rejected this approach, arguing that the Administrative Procedures Act prohibited the adoption of a guideline in lieu of a rule. The proposed Rule 57 was first public noticed in August 1980. I commented on this draft as a private citizen, opposing the decision to authorize the presence of human genotoxic substances in the waters of the state. Private sector public comment opposed minimum data re? quirements and reiterated concerns regarding the vague language of the standard. In response DNR rewrote the draft rule, making it even less detailed. The public comment on this version, public noticed in April 1981, was even more vehement. Following private sector protests at 1981 WRC meeting, DNR staff recommended to the WRC that it 'esolution creating a committee of scientific experts to advise the 1N .on how to go about developing toxic substances water quality stan ards-based effluent limits. The resolution was adopted at the July 1981 meeting and recommendations to the WRC as to who should sit on the committee were made and adopted at the August 1981 meeting. At that meeting I appeared before the WRC as a private citizen and indicated my opposition to the composition and constitution of the committee. I pointed out that environmental interests were woefully underrepresented on the committee, that the areas of expertise repre? sented on the committee were not sufficient for purposes of carrying out its mandates, and that none of the appointees were experts in can? cer risk modeling. To assure that the advisory committee be able to fulfill the functions for which it was created, I recommended that the committee operate in the following fashion: 1. . . . announce, conduct and record the minutes of open meetings according to the protocols set forth in the Open Meetings Act (D.A. 267 of 1976, as amended); 2. . . . provide for a period of public comment at each meeting; 3. . . . solicit written and oral testimony from recognized national and international experts; 4. . . . clearly and explicitly identify the issues it intends to address and the value judgments it makes, indicating what weights were given to what factors in determining minimum data requirements, adeguate risk assessment methodologies, acceptable risk, reasonable assurance of protection from unacceptable risk, etc. A follow-up 14 September 1981 letter to Robert J. Courchaine, Executive Secretary, WRC, expanded on these topics. In fact no period of oral public comment was provided for; written Page Four comments received were not put on the agenda for discuss from Dow Chemical USA and EPA scientists, no written or mments from nationally or internationally recognized sc ientists . soli- cited prior to public noticing of the final draft; and many?value judgments were buried in the text of the draft procedures adopted by the committee. My 24 September 1982 comments on the draft procedures are Attachment Ei- A 22 November 1982 letter to Robert J. Courchaine, summarizing the issues and questions raised by the development of proposed Rule 57 and the Rule 57 Advisory Committee effluent limit derivation pro- cedures, which have not been adequately addressed to date, has been supplied to you previously as an enclosure in my letter of 1/3/83. The Foresight Society has petitioned for a contested case hearing in the matter of Dow Michigan Division NPDES Permit M10000868, the first such permit to contain acceptable risk-based genotoxic substance effluent limits. Although our comments on the permit are quite exten- sive, a summary of our concerns is contained in a 20 November 1982 letter, a copy of which has already been provided to you. A 20 November 1982 letter to Bailus Walker, Jr., M.P.H., informed him of our discovery that the Midland County white female soft and connective tissue cancer rate for the period 1970-78 exceeds the Michigan and U.S. averages by four times, representing a 771% increase over the rate for the period 1950?59. A copy of this letter has already been provided to you. The Foresight Society believes that. if acceptable cancer risk- based effluent limits are to be adopted, the background risks for the population(s) at risk should be evaluated prior to adoption to determine: ive 1) 2) that the existing cancer, birth defects and. hereditary mutations risks do not already exceed the Michigan averages by a factor greater than l/lO0,000; and that the design acceptable increased incidence of l/lO0,000 is not eventually exceeded as a result of the discharges authorized by the permit. Page Six RISK ACCEPTABILITY AND THE QUALITATIVE DIFFERENCES OF RISK ad Often the acceptability of a given risk is determined by com? paring its magnitude to those of other common risks with which we live in a normal social context. Although such an approach is ap? pealing, great caution should be exercised in comparing the magni- tudes of qualitatively different risks. I do not believe that the Rule 57 Advisory Committee took sufficient note of these qualitative differences in adopting the 1/100,000 risk as the design "acceptable" risk. I believe the following qualitative distinctions must be con- sidered when evaluating risk acceptability: acts of God vs human acts readily avoidable circumstances vs not readily avoidable circumstances accurately quantifiable vs/ inaccurately quantifiable quantified from observation of outcomes vs/ quantified by estimation of outcome corresponding social benefits vs corresponding individual benefits The following examples are introduced to illustrate the distinct categories into which various common risks would fall: Risk of Being Struck By A Meteorite act of God not readily avoidable circumstances . not accurately quantifiable for any particular place at any particular time ag Seven . overall probability quantified from observation corresponding benefit of living above ground Risk of Being Struck By Lightning on a Golf Course act of God readily avoidable circumstances not accurately quantifiable for any particular golf course during any particular storm event overall probability quantified from observation corresponding benefit of enjoyment of the sport Risk of Developing Cancer From Medical X?Ray human act readily avoidable circumstance not accurately quantifiable for any particular person overall probability quantified by estimation from high dose - response observations of victims of Hiroshima and Nagasaki blasts . corresponding benefit of information on condition of internal organs to assist in accurate diagnosis and identfication of correct treatment Risk of Developing Cancer From Environmental Exposure to Anthropogenic Toxic Chemicals . human act not readily avoidable circumstance not accurately quantifiable for any particular person or chemical Page Eisht probabilities quantified by estimation from single substance laboratory animal high dose - response data corresponding benefits of lower costs of goods and services dependent upon the manufacture, use and disposal of anthropogenic toxic chemical Many have argued that the state has made life-cost trade-offs similar to those contemplated in the proposed water quality-based effluent limit derivation procedures in the past. They cite as an example the decisions of engineers to design out certain highway safety features in order to cut highway construction costs. However, I would argue that such decisions differ qualitatively from the toxic substances life?cost trade-off in six important ways: First, the risks of highway accidents under design representative worst-case conditions for a given highway design are quantified from observation of accident frequencies and not from estimates using un- validated models and laboratory data. Second, highway injuries and mortalities result from accidents. Under normal operating conditions no highway is designed to cause death or injury. The death of innocent people as a result of con? suming water and food contaminated with genotoxic substances lawfully discharged to the nation's waters under a Federal Clean Water Act permit is no accident. Third, driving on any given segment of highway under any given conditions is not a life necessity. One can alter one's travel plans to avoid driving during inclement weather, - to avoid a particularly dangerous stretch of highway, to drive with greater caution or use alternate transportation. Breathing air, drinking water and eating food are life necessities. If one's air; water and food are contami? nated, to avoid exposure involves much greater dislocations and hard- agqune ships. In many cases there is no escape. Fourth, if a particular segment of highway proves too dangerous, it can be redesigned. On the other hand, should we decide later that we have seriously underestimated the risks from a particular substance or mixture of substances, the contamination of the nations waters with persistent toxic substances cannot be "redesigned" away. Only extra- ordinary measures can remove some but not all of the contaminants. Many years must elapse before natural physical, chemical and biologi- cal processes transform them into innoccuous products. But they never disappear completely. Fifth, resources freed by the highway engineer in designing out certain highway safety features can be reinvested by the traffic engineer in traffic control features that save more lives for the dollar. Resources freed by the permit development engineer in design? ing out toxic substances zero discharge requirements tend to appear in stockholder dividends rather than pollution abatement R.D and equipment purchases. Sixth, an individual injured in a traffic accident cannot pass on that injury to innocent future generations. Genetic damage resulting from toxic substances~exposure can be passed on to future generations. Often the damage cannot be detected, preventing potential parents from taking appropriate precautions. Once the genetic damage is passed on, it can be propagated throughout a population, eventually expressing itself, perhaps as a debilitating deformity or enzyme deficiency, af- fecting thousands of future innocent lives. Page Ten RISK ASSESSMENT: A REVIEW Risk assessment involves the following: prioritization for limited resource allocation dose response analysis exposure analysis risk evaluation RISK ASSESSMENT Prioritization For Limited Resource Allocation l. substance known or reasonably expected to be intrinsically hazardous are identified; the use and discharge budgets of these substances in a given jurisdictional, geographical, airshed, watershed or aquifer location are evaluated; the number of people in each location is quantified; an initial, crude risk estimate is made for each compound; the substances are prioritized from highest to lowest risk; substances to be regulated are identified. RISK ASSESSMENT Dose Response Analysis 1. estimation of mammalian low dose toxic response relationship for each route of exposure from laboratory animal high dose - toxic response study data, usually for one route of exposure; conversion of laboratory animal dose, data from concentrations in carrier medium to per unit body weight of test animal; scaling up of the laboratory animal dose in proportion to the greater mass of the design representative most sensitive human (the so-called mega-rat transformation); age ?El even scaling down of the mega-rat dose in proportion to the slower metabolism of the design representative most sensitive human (usually done with the ratio of body surface areas); quantifiable and unquantifiable uncertain- ties and sources of error are identified; quantification of quantifiable uncertain- ties via propagated and total error analysis. RISK ASSESSMENT Exposure Analysis 1. identification of the protected population most sensitive to the toxic effect of the chemical by one or more routes of exposure; establishment of a representative worst-case exposure scenario for the most sensitive pro? tected population; estimation of a representative worst-case dose scenario for the most sensitive protected population under the conditions of the design exposure scenario; mathematical modeling of the concentration of the toxic substance in air, surface and ground waters, soil and food as a function of its rate of emission to the air, waters and soil; quantifiable and unquantifiable uncertainties and sources of error are identified; quantification of quantifiable uncertainties via propagated and total error analysis. RISK EVALUATION l. A conditional probability distribution function is constructed from the probabilities that: a) a member of the population is a member of the representative most sensitive population; b) a member of the representative most sensitive population will encounter circumstances in which the representative worst-case exposure will occur for some period of time; Page Twelve c} a given dose is delivered as a result of the integrated exposures throughout a representative worst?case social, occupational and recreational cycle; d) a given toxic reSponse will be elicited by a given dose. 2. For purposes of simplification it is assumed that: 1) all the members of the potentially exposed population belong to the representative most sensitive sub population; 2) each of the design population will encounter the design exposure conditions; 3) the exposure will last a lifetime; 4) all of the substance to which one is exposed enters the body as a toxicologically active dose. 3. A design statistical confidence level for each of the quantifiable uncertainties and sources of error is adopted; 4. A total error analysis on the risk estimate is performed at the design statistical confidence level; 5. Unquantifiable uncertainties are identified and the existence of unknown sources of error postulated; 6. The emission risk relationship is adjusted so as to compensate for the quantifiable uncertainties at the design statistical confidence level; 7. The quantifiable uncertainty - adjusted emission - risk relationship is adjusted again to compensate for identifiable but unquantifiable uncertainties and sources of error and for as yet unidentifiable uncertainties and sources of error. Up till now a design acceptable risk has not yet been adopted, yet important value judgments have already been made. Among the most significant are those regarding the design statistical confidence level to be adopted for purposes of quantifiable error analysis and age Thirteen the adjustment factors to be used to compensate for unquantifiable and unknown uncertainties and sources of error. I expand on the significance of these "hidden" value judgments in the next section. PaFe Fourteen RISK ACCEPTABILITY, RISK UNCERTAINTY AND MARGINS OF SAFETY The acceptability of any given risk cannot be divorced from the accuracy of its quantification. When the risk of injury from any particular activity is known with an acceptable degree of accuracy, intelligent choices about the trade?off between risks and benefits can be made. Conversely, when the risk of injury from any particu- lar activity cannot be accurately quantified such trade?offs cannot be made intelligently. The fear of injury brought about by the un? certainty of the nature and magnitude of the risk of injury must be factored into the determination of acceptable risk. As indicated previously, uncertainties are introduced into the estimate of risk at each stage in the process. Thus, the re- sulting estimate of risk can only be quantified in terms of probabilities: (This approach has been used in Attachment II by O'Neill at al. at Oak Ridge National Laboratories in calculating toxic sub~ stances concentration ecosystemic toxic effect probabilities.) There is always some probability that a given risk will be significantly underestimated using the approach adopted. That probability can be decreased by factoring margins of safety into each step of the risk assessment process. The degree to which the unquantifiable and in- accurately quantifiable uncertainties need to be offset via the introduction of safety factors is a value judgment. While there is no accepted methodology for making such value judgments, they must be made eXplicitly rather than implicitly, as was done by the Rule 57 Advisory Committee. age'Fifteen I do not believe that the use of the design once-in-ten-year, 7?day drought flow dilution factor; laboratory measured or estimated fish bioconcentration factors; the design 70 kg man; and the design 6 gm/day fish consumption scenario introduce sufficient margins of safety to offset the compounded total uncertainty in the calculated l/lO0,000 risk dose. Further, I do not believe that the Rule 57 Advisory Committee gave sufficient consideration to the margins of safety which ought to be incorporated into the genotoxic substance effluent limit derivation procedures to assure the protection of the public health at the design "acceptable" risk to a degree of statistical confidence considered adequate. This may in part be attributable to the fact that the Office of Toxic Materials Control technical staff with the most knowledge of the sources and magnitudes of uncertainty in the risk estimate pro- cesses and the greatest misgivings about the degree to which these uncertainties were to be offset by the proposed procedures were not allowed to put their ideas in writing or make presentation to the Rule 57 Advisory Committee. Unlike this committee, the Rule 57 Advisory Committee refused to allow public presentations at its meetings. I believe the failure to adequately offset the uncertainties in the risks estimated according to the methodologies recommended for adoption by the Rule 57 Advisory Committee is a fatal flaw that must not go uncorrected. Larry E. Fink 6300 W. Michigan Ave. F-24 Lansing, Mi. 48917 Richard Powers Chairman. Rule 57 Advisory Committee Office of Toxic Materials Control Michigan Department of Natural Resources 9/24/82 P.0. Box 30028 Lansing, Mi. 48909 Dear Mr. Powers: What follows are my recommendations, criticisms and questions regarding the document: "Proposed Surface Water Quality-Based Effluent Limitation Derivation Procedures for Chemical Substances" dated 6/23/82 --and public noticed for sixty days beginning 7/23/82. Teresa Kent has indicated that the comment period closes on 9/24/82. I trust my com? ments will receive due consideration. I. Recommendations A. General 1. The draft procedures should be divided into two sets of-procedures. Steps relating to the review of the NPDES Permit Application and the development of toxic substances monitoring requirements and effluent limits should be promulgated as rules per the requirements of the Administrative Procedures Act (APA) under the Part 21 Administrative Rules of the Water Resources Commission (NRC) governing Hastewater Discharge Permits. Steps relating to the derivation of a toxic substance Hater Quality Criterion or Standard should be promulgated as rules per the requirements of the APA under the Part 4 Administrative Rules of the WHO governing the physi- cal, chemical and biological quality of Michigan surface waters. The language of STEPS XVI should describe the procedures contemplated in far greater detail. Phrases such as "reasonably expected to be present", ?adequate wastewater charac- terization data", ?no significant potential for ex- posure," "adverse impact to aquatic,.terrestrial or human life" and nomencalture such as "pre~screening wastewater monitoring study" and fprocess characteri- zation study" should be better defined and clarified so that regulated industry and the public are pro- vided with a clearer understanding of the tasks in- volved in carrying out each of the steps. 3. Because of the far-reaching implications of these draft.procedures for the public health, safety and Powers 9/24/82 p. 2 welfare and for the protection of the natural resources from pollution, impairment and destructidn; and because the proposed state regulatory action meets one or more of the criteria defining a ?major state action? in Part 4 A of the Guidelines For the Prepa- ration and Regiew of Environmental Im act Statements under Executive Order 1974-4 (11725775);and Because sucn an activity?i??not exempted under?t?e - Commission Policy {'1036: Environmental Im act Statements- Preparation and Review (1/1/77g. I formally request that an Environmental Impact State- ment (BIS) be developed and published by the DNR for the proposed procedures according to the guidelines set forth in the above referenced documents. Following publication of the EIS and appropriate revisions of the proposed procedures in response thereto, I recommend that a series of public hearings be held at which to take public testimony on the proposed procedures. The Attorney General should rule as to the constitu- tionality and lawfulness of procedures for impl?mens tation of a wastewater discharge regulatory program which authorize: the discharge of substances which will significantly adversely affect the public and environmental health and thus inconsistent with the protections afforded the people under the 5th and 14th Amendments to the U.S. Constitution, the National Environmental Protection Act; Article 4, Section 52 of the 1963 Michigan Constitution and contrary to the national policy set forth in the Federal Clean Water Act, the state policy set forth in the Michigan Environmental Protection Act and Section 6(a) of the Michigan Water Resources Commission Act, as i amended. B. Specifics 1. STEP I a. establish criteria an NPDES Permit Application must meet to be considered complete; b. promulgate the Michigan Addendum to the Form 2c Consolidated Permit Application as a rule ac- cording to the requirements of the c. promulgate the list of substances regulatable under the authority of the Michigan Water Resources Commission Act as a rule according to the require- ments of the overs 9/24/82 P- 3 establish criteria of time, location, duration, replication, frequency and type of sampling; sample preservation; sample storage; sample preparation and sample analysis for the deter- mination of wastewater characterization adequacy and promulgate the criteria as a rule according to the requirements of the STEP II a. b. establish the criteria for requiring the per- formance of a process characterization study and promulgate the criteria as a rule per the requirements of the establish criteria for requiring a pre-screening wastewater monitoring study and promulgate the criteria as a rule per the requirements of the establish criteria of quantity and concentration in consideration of toxicity, bioaccumulation potential and persistence which determine whether a water quality-based effluent limit must be considered and promulgate the criteria as'a rule fper thetreQuir?ments?of the APA. STEP a. identify the most sensitive populations at risk for purposes of protecting the_receiving water for each designated use. STEP a. establish physical, chemical, biological and toxicological property}criteria a substance must meet to have no significant potential for exposure in water resulting in adverse impact to aquatic, terrestrial or human life; identify site-specific physical, chemical, bio- logical and technological conditions of the re- ceiving watershed which must be taken into account in deleting a substance from further considera- tion. STEP VII 8.. develop a tier testing scheme and a set of screening criteria so that staff of the DNR are authorized to call for more testidata if warranted on the basis of review of the minimum test data; and Powers 9/24/82 p. 4 promulgate those criteria as a rule per the requirements of the APA. STEP IX - a. adopt appropriate models to account for poten- tial additivity and nergism in multiple substance exposures. {see for example Cala- mari, D. and J.S. Alabaster (1980), Chemosphere 9: 533 - 538 carry out a statistical analysis of the data used to develop the acute/pseudo-chronic adjustment factor so as to define the 99% confidence interval; define the acute/pseudo? chronic adjustment factor as the acute/pseudo- chronic axis rntercept*bf a ray perpendicular to and intersecting a ray drawn perpendicular to and passing through the +99% confidence contour line at that point. Adjust this factor by a factor of ten to take into ac? .count the probability that some of the 28? day embro-larval NOAELs are higher than the corresponding life cycle prohibit the net discharge of human carcino- gens, hereditary mutagens, teratogens, im? munosuppressants and sterilogens to any sur- face water which is or may become or?is hydrau- lically connected with any surface water that is or may become a drinking water supply, sport or commercial fishery, agricultural water supply or full or Partial body contact recrea- tional area; compliance with this prohibition shall be established by passing the influent and effluent through a sorptive column in volumes sufficient to concentrate the substance to a degree equivalent to the representative worst-case bioconcentration factor, either measured or estimated, if measured values are unavailhble. prohibit the discharge of unregulatable geno- toxic substances, defined to be those substances for which the estimated risk concentration is so low that it is non?detectable in the edible portion of the most significant consumable bioconcentrating adgatic organism. constrain the sum of all anthropogenic genotoxic effects not to exceed Powers 9/24/prevail upon the legislature to adopt the for all other permit effluent limits for genotoxic substances, if and only if reasonable'and prudent alternatives, including closed system use and carbon pre-filtering for BAT, are not technologically feasible or economically achievable. STEP XIV See: Tung, Yeou?Kbung and L. U.?Mays (1980) "Risk Analysis for Hydraulic Design" J. Hydraulics Div. - ASCE 893 - 913 and Warn, A.E. and J.S. Brew (1980) "Mass Balance", Water Research V14 1427 - 1434 See #6 a. STEP XVII a. use preconcentration sampling to extend the working range of analysis sufficiently low to detect regulatable concentrations. STEP XIX a. when the ambient concentration of a chemical substance from anthropogenic sources in the receiving water exceeds its WQC value, staff shall recommend that a point and nonpoint source mass balance wasteload allocation be developed for the impacted watershed; b. should the limit prove to be unattainable or economically unachieyable, require such studies on the receiving water and the sub- stance 2 3* - water to permit - the modeling of the assimilative capacity .of the water. Powers 9/24/82 13. 6 II. Criticisms A. Procedural 1. Enter into the public record copies of letters to Robert Courchaine, Executive Secretary, NRC, dated 9/14/81 and 11/10/81 and 12/12/81 regarding deficiencies in the composition, constitution and operations of the Rule 57 Advisory Committee. B. Introduction 1. On page 4 of the Introduction under "Objectives and Philosophy of the Committee" it is stated: It is often difficult to completely separate science from value judgments. Wherever possible, the basis (science or value judgment) for de- cisions and recommendations is identified.? Unfortunately, a number of hidden value judgments were made by the Committee which are nowhere discussed, rationalized or justified. Such hidden value judg- ments include those embodied in the decisions not to take into account the statistical uncertainties in the data from which the acute/chronic adjustment factor was calculated and from which the cancer dose-response relationship was estimated. Another hidden judgment involved the decision not to considerrthe great potential for additive and synergistic effects resulting from multiple substance exposures. The proposed procedures allow DNR staff to take these effects into account only when test data are available to justify such a concern. Yet it is a mathematical impossibility to test even a few of the possible combinations of the various substances on EPA's Priority Pollutants list, let alone the vast number of substances in commerce. The presumption must be that synergistic effects must exist. As such appropriate marginsIOf?safety could have been factbred in to the NOAEL and 1/100,000 cancer risk-based effluent limits to protect the public health and the natural resources with a reasonable degree of statistical confidence. On page 5 of the Introduction under "Objectives and Philosophy of the Committee" it is stated that: The Committee tried to develop recommendations which would adequately protect the public health and the environment while placing a minimum cost burden on Michigan industry or municipal dischargers Powers 9/24/82 P. 7 On page 6 of that same section it is stated: The Committee has not considered in detail the economic impacts of implementing these recommendations The Committee was not charged with the responSibility and did not have the technical expertise to conduct an economic feasibility analysis. It should be noted that the Committee attempted to balance the costs and benefits of a particular degree of public and environmental health protection even though it lacked the necessary expertise. Further, such cost-benefit considerations are inconsistent with the Michigan environmental protection mandates - Article 4, Section 52 of the 1963 Michigan Constitution, the Michigan Environmental Protection Act and the WHO Act. This point has been made time and time again to the WHO by staff of the Department of the Attorney General acting as counsel to the HBO. In my August 20, 1981 statement to the NRC, which is part of the public re- cord, regarding irregularities in the method of appoint- ment to and deficiencies in the composition and cons- titution of the Rule 57 Advisory Committee, I requested the wsc'e support for my recommendation that staff of DAG explain to the Committee members the Constitu- tional and statutory mandates of the agency they were to advise. At the time Stu Freeman indicated that he would be willing to talk to the Committee if asked. This request was repeated in my letter of 9/14/81. and reiterated in my letter of 12/12/81. Questions. 1. -I have attached a copy of a letter sent to November of 1981 containing questions regarding the permit program in?general and then draft STEPS I - IV. Since STEPS I - IV have not changed significantly from the initial to final draft; and since I have been un- able to determine if my questions were ever discussed, because the minutes of the meetings were incomplete and the agendas did not set aside time for discussion of public comment, I am reiterating those questions here. I have also attached additional questions covering STEPS - XIX. Regarding Appendix E, it is not clear whether you will regulate,substances which are genotoxic by the inhala- tion route but inadequately tested as to their oral and dermal a igenotoxicity as oral and dermal.genotoxic substances. How do you intend to handle this situation? Powers 9/24/82 p. 8 Questions 3. What is your rationale for choosing the design weight as the 70 kg man? Given that the young are generally more sus~ ceptible to genotoxic effects than the mature, and given the fact that half the population of this state is female, wouldn't it have been more prudent to-use the weight of a ten year old girl as the design weight for purposes of regulating genotoxic substances?_ What is the fish consumption rate distribution function for the Michigan population? What are the 95th.. 99th. and 99.9th percentile rates of fish consumption(Kilograms/day) for the Michigan population? How was the 6.5 gram/day fish consumption rate derievd? Note: 6.5 should be .0065 Kg Thank you for this opportunity to comment. I appreciate your consideration of my views.. photocopies: Sincerely, Larry Holcomb, TSCC ajv ii Larry 22 Pink, Citizen rnlnulutft imiJ?Un-mntn \?ul 1. pp In? lull) 1"Intcd til the I're? I It] llHU I lSll.? WK) Si Hazard A ssessment ECOSYSTEM RISK ANALYSIS: A NEW METHODOLOGY R. V. O?Nt-tti R. H. L. W. BARNTHUL SE. C. W. St TER. S. G. HILDEBRAND. ?so C. W. [Em Sciences Du istott Oak Ridge National Laboratory Oal. Ridge. TN 373% lRt?t?e?tt?e?d IU October 1981. Accepted IS December l9li ll Abstract method is presented for estrapolanng laboratory data to aquatic ecosystem effects such as decreased or reduction in game fish biomass The reuunes translating laboratory. data into changes in the parameters ol' an ecosystem model. the Standard (Martin Model The translation tset'leCted through know ledge of modes of action. The uncertainties associated with both laboratory measurements and extrapola- tions are retained. and risk estimates are given in the I?orm ot' probabilities that an effect Could occur The approach is illustrated h\ scenartOs in which effects of tone substances are distributed trophtc letels Each scenario att?eets population interactions in different stats and a.ters built the lctel and the nature ol the risks to ecosystem prmesses Particular attention is paid to anals/tng the interaction hetsseen and the uncertatnttes associated ith extrapolation. tianieltsh biomass SWAL At present. accurate prediction of the effects of [0le substances on ecosystem properties is impossible. One begins with a limited set of laboratory data composed of toxicity tests on single species. These tests are poorly designed for extrapolation to the field because the mortality ol'organisms in a laboratory system may bear little resem- blance to their response in the natural ecosystem [1-21. From ?these laboratory data. one wishes to draw conclusions about effects on socially relevant endpoints which ecosgstents 'To whom correspondence may. he addressed. Presented at the First Annual Meeting. Society of Em tronmental and Chemistry. Virginia. Nmeniher 2-1-25. 1950. Risk analysis Organic int-'olie complex interacnons in the eco- system. which themselves are far from well understood. Despite the intrinsic diffi- culty ot~ the task- there is a critical need to estimate potential impacts. The need is par- ticularly critical for evaluating new techno- logies such as l'uel processes. The apparent impossibility of developing an Ecosystem Risk Analysis is net an unfamiliar challenge to science or society. The need to predict weather and project economic trends has long been realized. even though predictions cannot be made with great accuracy or reliability. [n such cases. an effort must be made and then judged on whether it represents the best ?state-ol-the-art? approach available. One of the approaches available for mak- ing uncertain predictions is risk assessment. lb? I?x UNI ct til One begins by defining the endpoints of interest. in ERA. the endpoints are socially relevant changes in ecosystem processes such as productivity or ability to support economically important species. Risk assessment then provides techniques for estimating the probability of that endpoint being reached. given all of the uncertainties involved. In this paper we will outline a concept of ERA designed to predict higher order eco- logical effects based on laboratory tests of toxicity. The approach incorporates a num- ber of innovative elements designed to apply the best ?state-of?the-art" techniques to the problem. The techniques include ecosystem models. prOpagation of uncer- tainties with Monte Carlo simulation. and the use of eXpert opinion to fill in critical gaps. We describe each of the components and present a series of calibration studies to test the sensitivity of the method. Finally. we apply ERA to examine the relative toxi- city of two organic substances: phenol and quinoline. DESCRIPTION OF RISK SIS in ERA (Fig. ll laboratory data must be translated into changes in parameter values of a Standard WAter COlumn Model This translation involves projecting mortalities. or other standard - a. Luau-av?. '1 4/ -. <2 . to.- Fig. A schematic illustration of the 00wofrnforma- tron leading from laboratory data to estimation of risk expressed as a frequency distribution of effects. responses. into changes in model parame- ters associated with processes such as gra/.~ ing or respiration. The expected change in each parameter value is expressed by an element of the Effects Matrix. E. This step requires knowledge about toxicological modes of action. Extrapolation of labora- tory data involves considerable uncertainty. Estimates of the uncertainty associated with each extrapolation are also included in E. Using the elements of and their associ- ated uncertainties. SWACOM estimates effects on the aquatic ecosystem by Monte Carlo simulation At the beginning of each run. all parameter values are chosen from the statistical distributions described by E. For example. a grazing rate of 0.5 and a corresponding matrix entry of 1.0 i 20 percent is assumed to describe a Gaussian distribution with a mean of 0.5 and a coeffi cient of variation t100 standard deviav tion/meanl of 20 percent. A random number is chosen from this distribution and the process is repeated for each parameter. Then a single. oneyear simulation of the model is made. A new set of parameter values is chosen and another simulation per formed. The procedure is iterated a large number of times until sufficient informa- tion is obtained to describe the frequency disrribution of results. In this way. effecrs on endpoints of interest can be stated as the probability or risk of the event occurring. The uncertainties of extrapolating labora- tory data to the field have become state- ments about the uncertainty of an undesirable effect. THE STANDARD COLUMN The biological and physical details of SWACOM are important in determining the frequency distribution of results. How- ever. it must be understood that Ecosystem Risk Analysis is not limited to any single model. Another model can replace SWACOM without altering the remaining elements of the technique. The present 5" Fig. I. ard WAtcr (I light. and SWACOM ct phytoplankto carnivorous 1' model. SV CLEAN been descri the pelagic considers zooplankte [Op carniv trOphic levt matical fur eter value p0pulation stant. Micl ture optimt The abiu Environme The tempe soidal cury 4? in the follows a 5 reduced ur add nutri Nutrients umn in s1 phytOplanl cesses are reminerali; fall turnox from the the water Phyt0pl light. temp r1 - n-h- ml i mu: u? n-nq 0 Ha ha 4??,le I 3.. quanta-ls? I imd cum 1. Eh: Fig. 3. A schematic illustration tifS\\'s\COl\1 iStand- ard WAter ('Olumn Moduli. Daily of nutrients. light. and temperature serve as model input. SW-KCOM considers the trophtc relationships of H) phytoplankton. 5 aooplankton. 3 forage fish. and single carnivorous fish species model. SWACOM. is a simplification of CLEAN Portions of the model have been described previously SWACOM iFig. 2) is designed to mimic the pelagic portions of a lake ecosystem. it considers lO phytoplankton populations. 5 7.00plankters. 3 planktivorous fish. and a top carnivore. The populations in each trophic level are described by similar mathe- matical functions but with different param- eter values. Thus. each phytOplankton population is charaCterized by its maximum rate. light saturation con- SIant. Michaelis-Menten constant. tempera- ture optimum. and susceptibility to grazing. The abiotic driving variables mimic the environment of a northern dimictic lake. The temperature describes an annual sinu- soidal curve with lake turnover occuring at 4? in the spring and fall. Radiant energy follows a similar curve. with light greatly reduced under ice cover. External sources add nutrients each day of the year. Nutrients are removed from the water col- umn in stoichiometric relationship with phytoplankton growth. Decomposition pro? cesses are not considered in the model. and remineralization occurs only at spring and fall turnover when regenerated nutrients from the hypolimnion are added back into the water column. Phyt0plankt0n growth in response to light. temperature. and available nutrient is Ecosystem risk analysis I69 described in detail in Self-shading effects are accounted for by integrating photosyn- thesis over the euphotic zone. Each phyto- plankton population has an optimal temperature at which its rate is maximum. fixation of biomass is primarily limited by available nutrients which are exhausted in periods of rapid growth. Grazing or predation at all trophic levels is described by a nonlinear interaction funco tion This function considers both lim- ited food supply and competition with other grazers. The consumer populations are lim- ited by their individual metabolic and mor- tality rates and by predation from higher trophic levels. Both grazing and respiration rates are affected by temperature with each population characterized by an optimal temperature. A typical simulation run is shown in Fig. 3. To simplify presentation. biomass for each trophic level has been summed over the individual populations. Growth begins in the spring after the ice cover breaks up at about day 80. Turnover occurs after day 100. significantly increasing the concentration of available nutrients. As nutrients are depleted. phyto~ plankton biomass decreases rapidly. Zoo. plankton biomass reaches a peak about one month after the phytoplankton. and forage fish follows about a month later. Game fish biomass is quite small. relative to the scale of the figure. and is not shown. Fall turnover. just before day 320. increases available 195! l1: rum-.90 we: (Emma 00:! - ?Bn' "?Elgun I rue-Lamh?E wean Fig. 3. A typical simulation of SWACOM. showing seasonal dynamics of phytoplankton. zooplankton. and forage fish. Values shown on the graph are summed over the component populations. I70 L'l .il. nutrients and causes a minor increase in phytOplankton. The figure shows that the model is capable of reasonably complex seasonal behavior that follows the general pattern of northern lakes. The basic structure of SWACOM per mits it to simulate a number of higher order effects. Toxic effects which differ across the phytoplankton populations. can cause com petitive release. Increased mortality in one population results in increased nutrient availability for the remaining populations. Thus. overall productivity may be unaf' fected even though one or more individual populations are strongly affected by the tox- icant. Similarly. toxic effects on forage fish decrease grazing on zooplankton. prevent- ing phytoplankton blooms. It is this type of system-level effect which will be of concern in Ecosystem Risk Analysis. Parameterization has been greatly simplified since we are primarily interested. at present. in developing the methodology. The model currently simu- lates the behavior of an arbitrarily defined assemblage of populations. rather than spe- cific taxa. This approach will suffice for present purposes but would. of course. need to be improved before the model is applied for aetual assessment purposes. EFFECTS MATRIX The parameters of the model can be writ- ten as a matrix with rows corresponding to each of the 18 populations and columns corresponding to parameters [6 for each phytOplankton pepulation and 5 for each consumer}. The Effects Matrix, E. corre? sponds element by element to the parameter matrix and is used to express the effect of a toxicant on each parameter. During Monte Carlo simulations. each parameter will be multiplied by the corre- sponding element of E. If an element is ID mo toxicityi. then the parameter is unaltered. 1f the element of is greater than 1.0. the toxicant will increase the rate repre- sented by the parameter. If the value is less than l.0.the effect willbe to reducethe rate. The advantages of using an Effects Matrix are that different arrays of toxic effects can be imposed without changing the model or its basic set of parameter values. The uncertainties caused by extrapolat- ing laboratory data to the field are expressed as error terms associated with each element of E. Thus. the information: would expect a 10 percent increase in grazing rate. and I would have a confidence in that extrapolation of 20 percent" would tran- slate into a matrix element of 20 percent. This is sufficient to describe a Gaussian distribution for that element from which random values may be drawn. Extrapolating laboratory data now becomes a matter of choosing values and uncertainties for the Effects Matrix. At this point. information is required on the mode of aetion of the toxicant. suggesnng a direct relationship between concentration and changes in physiological processes represented in the model. An example might be bioassays of effects on maximum GENERAL STRESS It is possible to describe a General Stress that simplifies the task of extra- polating laboratory results to changes in model parameters. For instance. a general Stress response by phytoplankton might involve lower maximum rate. increased reSpiration. lower light satu- ration point. and increased Michaelis- .Vlen- ten constant. Applying the concept to consumers. whether zooplankton or fish. involves similar assumptions. Grazing rate may be decreased. respiration increased. temperature optimum lowered and both mortality and susceptibility to predation increased. However. caution mu5t be exerted in applying such an approach because a toxicant may have additional physiological effects which would have to be taken into account. For example. some toxicants are narcotic and reduce respira- tion rates. the purpose assumed [ht The Gen some intere Matrix. lf about how decreased. rate will nc ment of l. but not a This info by use of a If an eleme standard de and a Gau approximati would be at button trar x. lying be) p-tp-xt. In would be tr; 3 distributit above the rr Before a attention mt ior and sens particularly tive contribi effect ofunc lation. The tion of both large value ity that the that is net [i tions but ha show an eq here the t: involved in environmen about its eff cause is diffi To exam cases was model were tion rates. rather than increase them. For the purposes of the present study we have assumed the stress defined above. The General Stress introduces some interesting constraints on the Effects Matrix. If a toxicant reduces maximum there will be uncertainty about how much will be decreased. However. it is certain that the rate will not be increased. Thus. that ele- ment of can take on values between Oand l. but n0t above 1.0. This information is included ?folded normal distribution" [91. If an element of is equal to 0.8 with a standard deviation of IO percent li.e.. 0.08). and a Gaussian distribution is assumed. approximately 0.6 percent of the values would be above 1. The folded normal distri- bution transforms the random number. lying beyond an arbitrary point. p. by p-tp-xl. In our example. a value of 1.02 would be transformed to 0.98. The result is a distribution with one-half of its values above the mean. but no values above 1.0. TOXICITY AND LNCERTAINTY Before any new technique is applied. attention must be paid to its general behav- ior and sensitivity. In the case of ERA it is particularly important to compare the rela- tive contribution of direCt toxicity with the effect of uncertainty resulting from extrapo- lation. The risk esrimated in ERA is a funco tion of both factors. For a toxic chemical. a large value for risk implies a high probabil- ity that the effect will occur. A chemical that is not toxic at the observed concentra- tions but has large uncertainties in may show an equally high value for risk. But here the probability expresses the risk involved in releasing the chemical into the environment. given that we know so little about its effects. The risk is Still real. but the cause is different. To examine this problem. a set of test cases was run in which all parameters of the model were assumed to be affected equally. Ecosystem risk uttulyus Assuming the General Stress described above. all parameters were changed by the same percentage. For exam- ple. a 5 percent effect was represented in by either 0.95 or 1.05. depending on whether the effect was to increase or decrease the parameter. For each level of toxicity. different uncertainty levels. rang- ing from I to 20 percent. were tested. Test results for several endpoints are summarized in Fig. 4. The _v axis of each graph shows the risk of reaching an end- point. such as a 20 percent reduction in 200plankton biomass. The axis expresses toxicity as the percent change in parame- ters. and the axis shows percent uncer- tainty. Figure 4a shows that increased toxicity li.e.. percent change in parameters of all populationst increases the risk of a 25 percent reduction in forage fish biomass summed over the year. Considering the 1 percent uncertainty case. there is little risk at 5 or 10 percent toxicity; however. at 15 and 20 percent toxicity. the risk of reducing forage fish biomass rapidly increases to 1.0. As uncertainty increases. at any level of toxicity. the risk of reaching the endpoint also increases. For example. at 0 percent toxicity. the risk becomes 0.40 as uncer- tainty approaches 20 percent. (Remember that 0 percent toxicity means that we do not think an effecr will occur. but 20 percent uncertainty means that we are not at all sure}. Therefore. there is still some risk involved in releasing the subsrance to the environment. The situation reverses at 20 percent toxicity where uncertainty reduces the risk from l.0 to 0.8. in other words. we a ?x .l 1 Fig. 4. Responseof forage fishlAl.phytoplanktontBI. and zooplankton hiomasscs to the effect of a toxic substance 1X axist. uncertainty of effect 42 EMSI and frequency of response I i" axisi. V. (J'Nt-Iit et ul. believe that the chemical will change model processes by 20 percent. but again. we are far from secure about this prediction. This uncertainty translates intoa decreased risk. The results for forage fish biomass tFig. 4a) follow the expected pattern: risk of an undesirable effect increases with toxicity and also with uncertainty: however. other patterns are possible. A 20 percent reduc- tion in phytoplankton biomass is insensitive to toxicity (Fig. 4hr. At any given level of uncertainty. the graph is relatively ?at from left to right. showing that increased toxicity does n0t necessarily increase the risk of a reduction in biomass. Because of nurrient limitation. increased stress does little more than change species composition As dominant species are affected by the toxi- cant. other populations, previously kept down by competition for nutrients. are n0w able to grow more successfully. even under some toxic stress. The result is that total phytoplankton biomass. summed over all populations. changes very little. On the Other hand. increasing uncertainty. at any given level of toxicity has the expected effect of increasing risk. The complexity of interactions in the model clearly affeCts the risk associated with a 20 percent reduction in zooplankton biomass [Fig 4c}. lncreasing uncertainty has the expected result of increasing risk; however. the interaction of uncertainty and toxicity is more complex. At 5 percent uncertainty, the highest risk occurs with 5 percent toxicity; at 7 percent uncertainty. the peak shifts to 10 percent toxicity: and at higher uncertainties. the peak shifts to 15 percent toxicity. The decreasing risk at higher toxicity results from nonlinear relationships in the model. As all populations in the system are affected. predation on zooplankton decreases. At some combinations of toxicity and uncertainty. the zooplankton seem to prosper. and its food supply, phytoplank- ton. remains relatively constant (Fig. 4b). At the same time. forage fish biomass is decreased lFig. 4al. The result is lower risk of significant reductions in zooplankton biomass. These tests demonstrate that uncertain- ties have an important. sometimes domi- nant. effect on risk. In some cases. expected toxicity is less important than our uncer- tainty in applying results to the field. This result emphasizes that the risk associated with releasing a toxicant may be dominated by our lack of understanding rather than the actual toxicity of the chemical. A second result is that toxicant effects are difficult or impossible to media without the intervention of an ecosystem model. Significant impaCLs on phytoplankton may have little effect on ecosystem funcrion in a nutrient limited system. On the other hand. small toxic effects may result in greater risk for zooplankton populations than for larger effects. ENDPOINTS Assessment problems require that spe. cific undesirable effects be defined as end- points that are relatively independent. Multiple endpoints are of little value if the analysis indicates a high risk for all of them. Different endpoints should show significant risks in response to different impacts. The sensitivity of ERA endpoints was tested by imposing an arbitrary 15 percent toxicity with a percent uncertainty on different trophic levels or on various combi- nations of trophic levels lTable ll. Each risk calculation is based on 500 Monte Carlo iterations. No two endpoints show signifi- cant risk at exactly the same combination of cases. Additionally. there is a real difference in the sensitivity of the endpoints. End- points associated with fish tdecrease in forage fish. game fish. and the ratio between them) respond to most of the cases. This is consistent with our understanding of aquatic ecosystems. since higher trophic levels often show greater sensitivity to impaCts. On the other hand, a 5 percent decrease in toral productivity never occurred under any of the cases teSted. T?hli 5% inei 20% dc 30% dc 30% de foragi 50% de- game 5% deer total l0% inc total I 10% dot to res; 5.0% dot forage ?Trophii risk. less The resu predicted a phytOplank are affectet respiration easel of reduced ratio when ahhough decreased increased. Althougl not be of warrant em their inclus nonlinear impact is is plankton reduced gr: lated on ga reducing zo ing the blooms. I decreased since this le; similar pro Ecosystem risk 173 Table 1. Risk of specific ecosystem effects when toxicant shows an effect (15% effect. 1% uncertainty-i on various trophic let'els? Trophic Leyeltsi Endpoints ZFG PZFG 5% increase in maximum phytoplankton bloom 20% decrease in phy-tOplankton biomass 3.0% decrease in zmplankton biomass decrease in forage fish biomass 50% decrease in game fish biomass 5% decrease in total productivity 10% increase in total respiration 10% decrease in producuon to respiration ratio leR1 50% ?Jenna: in game to forage fish ratio 0.12 1.00 0.30 0.93 0.0-1 0 0-1 1.00 0.87 00 [.00 0.90 1.011 1 00 1.00 1.00 0.95 1.00 0.58 0.4? 0.99 1.00 1.00 1.00 ?TrOphic leyels indicated are the only portions of the ecosystem affected; phytoplankton. zooplankton. forage fish. C3 game fish. A missmg value indicates an risk. less than 0.001. The results in Table 1 could n0t easily be predicted a priori. There is a small risk of phytoplankton blooms if game fish alone are affected. There is a risk of increased respiration if all consumer populations case] are affected. There isa small risk of reduced ratio when 200plankton alone are affecred. although productivity is not significantly decreased nor respiration significantly increased. Although some of the endpoints would not be of sufficient interest to society to warrant emphasis in assessment problems. their inclusion emphasizes the complex. nonlinear dynamics of the model. When impaCt is isolated on 200p1ankton. phyIO? plankton biomass increases because of reduced grazing. When the impacr is iso- lated on game fish. forage fish increases, reducing zooplankton biomass and increas- ing the probability of phytoplankton blooms. Phytoplankton biomass is decreased when forage fish are reduced since this leads to increases in grazers. By a similar process. zooplankton biomass is decreased either by a reduction in its food supply (P case! or by a reduction in predator fish which increases forage fish grazing pres- sure on zooplankton. Total respiration is increased in the F. G. ZFG cases since all of these cases increase zooplankton biomass which is primarily responsible for total respiration. Table 1 indicates which endpoints to con- sider in further Studies. The insensitivity of a 5 percent decrease in productivity indi- cates that this endpoint is unnecessary. The ratio of game to forage fish is redundant because it is dependent on the values of the two trophic levels. This endpoint will not be considered in subsequent analyses. Of course. both of these results sh0uld be veri- fied by field studies since they could be due to problems in the model. APPLICATION OF ERA TO TWO TOXIC CHEMICALS Having examined the sensitivity of the analysis to various test cases. we can now apply the method to a preliminary assess- [74 R. V. cl al ment. For this application we have chosen is able to grow at low nutrient concentra- to compare two organic chemicals: phenol . . lions and may bloom in the summer. This and qumoltne. These chemicals are of par- general behavior identifies the ticular interest since they are among the with population l0 Populations 6-9 toxic substances expected in the effluent of were assigned intermediate sensitivities to fuel plants. the toxicant. Therefore. the range between Toxicity data tTable 2) on these chemi- the percent effects on populations 5 and 10 cals are typical of what is available for was divided into equal intervals and assessment. The set of organisms tested is assigned to populations 6-9. We assumed n0t completely relevant for application to that the measured values on populations 5 our lake ecosystem assessment: however. and l0 were known with 5 percent coeffi~ we willattempt to use allofthe information. cients of variation. For intermediate popu- Since we are attempting only a relative risk lations. we assumed that the uncertainty of assessment. many assumptions can be assigning effects increased and therefore incorporated into the analysis as long as we assigned uncertainties ranging from l0 per- do net clearly bias the results in favor ofone cent to 20 percent. For the remaining popu- of the chemicals. lations. we assumed a minimal effect {0.0001} but with a 30 percent uncertainty. Table 2. Results oflaboratory tests of acute toxicity Turning to the five zooplankton for two organic chemicals?. Values are mg Organisms Phenol Quinoline Green Algae (Seienaitrum; 25h Blue-green algae 54'" 1 l7" Daphnia magnet 30 28.5 Chironomtd l03.8 5.7.2 Snail 258.4 l83.l Fathead minnow 23 l.5 Rainbow trout 0.15 ll. ?The data are ~18 h. tconcentration that kills 50% of organismst Milleman. Pers. Comm. .These values are ECl?s. that 15. concentrations that cause a 30% reduction in Giddings et al.. in press. We have assumed that borh chemicals are present in the environment in concen- trations that result in a 5 percent effect on rainbow trout. The relative toxicity on other organisms is calculated by dividing their LCm's tor into that of rainbow trout and multiplying by 5 percent. This maintains the same spectrum of sensitivities as indicated by the laboratory tests. It is necessary to identify the test organ- isms with populations represented in the model. The green alga. Selenasrrum, tends to bloom in the late spring and corresponds to phytoplankton population 5 of in many of its physiological responses. The bluegreen alga. Microcysris. tions. we assumed that Daphm'a magna cor- responded to population 2 with a 5 percent uncertainty. We used the data on the Chiro- nomid as an indicator of the sensitivity of anOther invertebrate consumer in the sys- tem. widely different taxonomically from Daphnia. That is. we assumed that the dif- ference in sensitivity between Daphnia and the Chironomid indicated the range that might be expected among invertebrate con- sumers. Therefore. we assumed that popula- tion 5. the population most different from Daphnia in Other physiological parameters. corresponded in sensitivity to the Chiro- nomid. This extrapolation was assigned a larger uncertainty of l0 percent. The remaining populations were assigned per- cent changes intermediate between Daph- m'a and Chironomid with uncertainties of 20 percent. The fathead minnow was identified with the third forage fish population and assigned an uncertainty of 5 percent. We assumed that the snail represented the oppo- site end of the range of sensitivities for macroscopic consumers in the system and assigned its sensitivity to the first popula- tion with a CV of IO percent. The second population was assigned an intermediate toxicity and a 20 percent uncertainty. Finally. th with our ga effect and It is ot method of; however in defensible ever. an ar for develop complex st such as asst equal in St forage fish been very tainties: mt tion could accuracy of depended e) ingdifferent might affec isms. Never the decision menting the It shoulc only concert the two che ested in an a of either ch Fig. 5. Daily line. At. zoopla line. B). and ca cases: no toxica cals are assume tions to produ biomass. 0 Finally. the?ainbow trout was identified with our game fish and assigned a 5 percent effect and given a 5 percent uncertainty. It is obvious that such an arbitrary method of assigning values for the matrix. however internally consistent. would n0t be defensible in an actual assessment. How- ever. an arbitrary approach should suffice for developing the methodology. Other less complex scenarios might be envisioned. such as assuming that all zooplanltters were equal in sensitivity to Daphnia and all forage fish to the fathead minnow. We have been very conservative in assigning uncer- tainties; much larger coefficients of varia- tion could be jusnfied based on the typical accuracy of ecological data. We have also depended exclusively on the General Stress defined earlier rather than seek? ing differences in the way the two chemicals might affecr the physiology of the organ isms. Nevertheless. the example illusrrates the decisions which must be made in imple- menting the approach. It should be remembered that we are only concerned with the relative toxicity of the two chemicals. Thus. we are less inter- ested in an accurate prediction of the effects of either chemical than in comparing the ?In. 0' ?0h" unmet-5 I 360 '5 50 2?0 300 360 5?50 :va II h-tl qua lac. Fig. 5. Daily t0tal biomass of phytOplankton [solid line. At. zooplankton tdashed line. AI. forage fish tsolid line. Bl. and carnivorous fish tdashed line. Bl for three cases: no toxicants. phenol. and quinoline. Both chemi- cals are assumed to be present in sufficrent concentra- tions to produce a 5 percent decrease in game fish biomass. liens} stem risk until}. 175 chemicals within the context of the scenario we have deveIOped. Such a comparative study provides the assessor with consider. able ?exibility. Details about the assump- tions are less important than avoiding anything that biases the results in favor of one chemical or the other. The deterministic time behavior of the model li.e.. ignoring the effects of uncer- tainty] is shown in Fig. 5. Compared to the "no effects" case. phenol causes an increase in phytoplankton. particularly in the late summer. Forage fish biomass is increased. and game fish biomass is decreased. Quinoline causes a reduction in the phyto- plankton bloom and a significant increase in the zooplankton. particularly during the late spring and early summer. The forage fish biomass is dramatically reduced. com- pared to the other two cases. Game fish biomass is reduced and barely visible along the abscissa of the graph. Table 3. Risk of undesrrable effecrs on an aquatic ecosystem resulting from two organic chemicals? Endposnt Trophic Leveltsi Phenol Quinoline 5% increase in maximum phyto- 0. [6 0.05 plankton biomass I 26.! ll 20% decrease in phytOplankton 0.l6 0.4] biomass 276.31 20% decrease in zooplankton 0.07 0.002 biomass l< 73l.0i 20% decrease in forage 0.2l 0.8-1 fish biomass t< 257.41 50% decrease in 0.3l 0.88 fish biomasst< 3 .4l l0% increase in toral 0.32 0.78 respiration t> l03.3 l0% decrease in productivity: 0.37 0.32 respiration l. 6t Values are based on 500 Monte Carlo iterations. The risk of exceeding each of the end- points. based on 500 Monte Carlo iterations are listed in Table 3. The results show that quinoline poses a greater risk of undesirable reductions in fish populations and HR ratio. caused mainly by increases in consumer reSpiration. Phenol poses a greater risk of phytoplankton blooms and reduction in zooplankton biomass. In general. quinoline l76 R. V. et al. yields larger risks for effects on the aquatic ecosystem. Much of the increased risk appears to be traceable to the greater sensi- tivity of forage fish to quinoline lTable 2). The difference in sensitivity of this trophic level dominates the results. leading to a high probability that fish biomass will be reduced to an unacceptable level. As a result of reduced fish biomass, zmplankton biomass increases. leading to increased respiration and significant effects on the ratio. Although results in the form of Table 3 represent typical output from the analysis. ERA produces a complete frequency distri- bution for each of the endpoints. The distri- butions of phytoplankton biomass [Fig 63} are very similar for the two chemicals. It is interesting that the deterministic solution for quinoline lies below the mode of the distribuu?on. whereas it lies above the mode in the distribution for phenol. The differ- ence between the deterministic solution and the mean of distribution represents the bias involved in predicting effects without including the effects of uncertainty The biases in Fig. 6 demonstrate the diffi- culties of predicting ecosystem effects with- out explicitly considering uncertainties. The frequency distributions for 200- plankton biomass (Fig. 6b} show the curve for quinoline shifted to the right. predicting higher probabilities for large 200plankton biomass as indicated in Table 3. In this case the deterministic solution is below the mode for both chemicals. The distributions for HR ratio (Fig. 6c! show the curve for quino- line shifted to the left and both deterministic solutions lying above the modes. In all cases in the figure there is significant bias. indicat- ing the problems involved in predicung ecosystem effeCts with a deterministic model. This paper has presented a new meth? odology for exuapolating laboratory toxi- city data to risks of ecosystem effects. It is clear from Tables I and 3 that the ecosystem rml-?Fig. 6. Frequency distributions of effects of phenol [solid liner and qumoline Idashcd ltnei on phytoplank- ton biomass Iat. zooplankton biomass tbt. and PR ratio tcl. determined by 500 Monte Carlo iterations. The points on the lines indicate the deterministic solutions. effects calculated by ERA could n0t be pre. dicted without this form of analysis. The nonlinear interactions incorporated into models such as SWACOM are needed to predict the higher-order effects which result from toxic effects distributed over the vari- ous populations in the ecosystem. The limi- tations of the approach can be clearly identified with the procedures employed in choosing the values of the effeCLs matrix and their associated uncertainties. The ten cases presented in this paper have illustrated that uncertainties of extra- polation must be explicitly considered in any analysis of ecosystem risks. Figure 4 shows that uncertainties can dominate results of the analysis. producing significant risks even if no direct toxicity is expected. On a pessimistic note. it must be added that only a fracuon of the uncertainties involved in the prediction have been included in the analysis. There are uncertainties involved in the formulation of the mathematical model lland in the basic parameter set [l 2] used for the simulation. if all of these independ- ent sources of uncertainty were included in the analysis it. is likely that our lack of knowledge about the dynamics of ecological syStems would result in unacceptable risks. no matter what level of toxicity were being considered. This reinforces the argument offered [13] that we must be extremely con- servative about any societal operation which impinges on the ecosystem. It also emphasizes our inadequate scientific base for decision making in an ecosystem context. It is clear from the phenol and quinoline example tl far simpl attempts 1 ing assum may be ac and unac single imp restrict at parative a gained wit The st shown in prediction tial error i. effects wi tainty. Th prediCt ec advocated indiCate th of probabi the proble chosen. th analysis at the form necessary assessment at no Mede- the Environ agency Agri NationalScn gram under \?ullh the licatton No ORNL. example th rative risk assessment is far simpl it an assessment which attempts to predict absolute risks. Simplify- ing assumptions that do not bias the results may be acceptable in a comparative analysis and unacceptable in the assessment of a single impact. Thus. it may be advisable to restrict applications of the ERA to com- parative analyses until more experience is gained with the method. The strong in?uence of uncertainties shown in Fig. 4 and the resulting biases in predictions shown in Fig. 6 show the poten- tial error in attempting to predict ecosystem effeCIs with a method that ignores uncer- tainty. The use of deterministic models to predict ecosy'Stem effects has often been advocated. but the analyses presented here indicate that calculation of risks in the form of probabilities is a more honest approach to the problem. Whatever specific method is chosen. the inclusions of uncertainty in the analysis and the statement of conclusions in the form of probabilities should become a necessary criterion for environmental assessments. Research supported in part by the Environmental ProreCtion Agency under Inter- agency Agreement 4034078. and in part by the National Science Foundation's Ecosystem Studies Pro- gram under liitragency Agreement itith the Department of Energy under contract uith Lnion Carbide Corporation. Puh- lication No. [994. Entironmental Sciences Ditision. ORNL. risk analysis l'l? REFERENCES . Lane. PA. and R. Levins. I977. The dynamics of aquatic ecosystems. 2. The effects of amount enrichment on model plankton communities. Limnol. Ot't'ariogr. 2145-1-47 1. . O?Neill and J.B. Waide. I98l. Ecosy?Siem theory and the unexpected: implications for envi- ronmental toxicology. In Cornaby?. ed.. Toxic Substances in the Environment. Ann Arbor Science. Ann Arbor. Michigan. . Rubinstein. R.V. [98 . Simulation and the Monte Curio Method. John Wiley and Sons. New York. Park RA. 1974. A generalized model for Simulat- ing lake ecosystems. Simulation 23:33-50. . O'Neill andJM. Giddings. [979. Population interactions and function. In G.S.1nl'llS arid R.V. O'Neill. eds. Systems 0} Ecosystems. International Cooperative Publishing House. Fairland. Maryland. p. [03. . DeAngelis D. RA. Goldstein and RV. O'Neill l975. A model for trophic interaction. Ecology 56:88l-892 . Giddings AJ. Stewart. R.V. O?Neill and RH. Gardner. An efficient algal bioassay based on short-term response. In press. . Gardner. DR. 1975 relation- ships of DDT analogs in non-target organisms. In 0.0. Veith and DE. Konasewich. eds. Sympo- sium on Correlations t'ri Studies of Timmy and Bioaccumulotion with Aquatic Organisms. Great Lakes Research Advisory Board. Windsor. Ontario. p. . Hahn Gland SS. Shapiro. I968. .1~fod~ els in Engineering. Wiley. New York. . Gardner R.H.. D. Huff, RN. O'Neill. J.B. Mane kin. .I.H Carney and Jones. 1980a. Application of error analysis to a marsh hydrology model. ll?arer Res. 16:659-664. . Gardner R.H.. RN. O'Neill. J.B. Mankin and D. Kumar. 1980. Comparative error of six predator-prey models. Ecology 61:333-332. O'Neill RH. Gardner and J.B. Mankin 1980. Analysis of parameter error in a ncmiirtear model. Et'ni. Modelling 8:297-3l I, Holling CS. I973. Resilience and stability of eco- logical systems. Annit. Rey. Ecol. Si st. 4:1-33. VF '1 f0 jq I1 1' SOC 6300 w. Michigan Avenue F-24 Ml 43917 517-321-7358 I i William D. Marks Acting Executive Secretary Water Resources Commission 17 January 1983 P.0. Box 30028 Lansing, Michigan 48909 Dear Mr. Marks: Given both the holiday season and Bureau reorganization with which you had to contend of late, the delay in your response is quite understandable. Unfortunately, while many of the issues I have raised have been previously expressed in far greater detail, I believe that to date most have not been adequately addressed. Your assurances that the requirements-of the Administrative Pro- cedures Act will be met by the Water Resources Commission (WRC) in the development and public noticing of proposed Rule 57 and attendant procedures are appreciated. Will the WHO also meet the requirements of Executive Order 1974-4 -- developing and publishing an Environ- mental Impact Statement (EIS) for each injurious substance whose discharge to and presence in the waters of the state the NRC intends to make lawful via the adaption of proposed Rule 57 and attendant procedures? I ask that this letter and the following enclosures be entered into the record of discussion and public comment on proposed Rule 57 and the final report of the Rule 57 Advisory Committee at your 20 January 1983 meeting, which I am, unfortunately, unable to attend: 1. letters dated 3 January 1983 with enclosures and 12 January 1983 to the DNR Environmental Protection Policy Advisory Committee (EPPAC) members;(Attachment 2. my comments on the draft of the final report of the Rule 57 Advisory Committee dated 24 September 1982; 3. my 22 November 1982 letter to Robert J. Courchaine summarizing my grievances with the HRC regarding its responses to my comments on proposed Rule 57 dated 27 August 1980, 24 April 1981, 17 February 1982; 'those referenced letters;and related-communications; 4. the 5 May 1982 petition for contested case hearing in the matter of Dow Michigan Division NPDES Permit ?10000868, filed on my own behalf and that'of the Foresight Society'(Attachment 5. the comments read to and enclosures handed out to the members of the DNR EPPAC members at its 19 January 1983 meeting (Attachment A . William D. Marks 17 January 1983 p. 2 6. letters to Mr. Courchaine dated 9/14/81, undated circa 11/10/81 (Attachment IV) and 12/12/81 (Attachment I urge rejection of the final recommendations of the Rule 57 Advisory Committee on the following grounds: 1. Environmental interests and environmental values were woefully underrepresented on the Committee; 2. The areas of scientific expertise represented on the committee were not adequate to fulfill the tasks assigned to the Committee by the 3. With but a few notable exceptions, the persons appointed to the Committee were not recognized scientific experts; 4. Although the public was allowed to attend the meetings, no period of oral public comment was allowed and no time was set aside in the meeting agenda to discuss written public comment received, contrary to the re uirements of the Open Meetings Act (1976 P.A. 267 5. With the exception of Dow Chemical USA and EPA scientists, written and oral presentations from nationally and internationally.reCQSRiZed 65P?rt8 in the fields of environmental economics, chemical environmental fate, exposure assessment, risk assessment, oncology, teratology, moiecular genetics, population genetics and statistical analysis of error and uncertainty were not solicited by the Committee during its deliberations; 6. Office of Toxic Materials Control technical staff with the most knowledge of the sources and magnitudes of uncertainty in the risk estimate processes and the greatest misgivings about the degree to which these uncertainties were to be offset by the proposed pro- cedures were not allowed to put their ideas in writing or to make oral presentations to the Committee; 7. Although the Federal Clean Water Act, the Michigan NRC Act and the 1963 Michigan Constitution make no provisions for using the projected costs of a proposed standard as a test of its reasonableness, the costs to the prixate sector were taken into account in adopt? ing the design acceptable risk of 1/100,000; William D. Marks 17 January 1983 P- 3 8. The substances to be regulated as toxic or other- wise injurious and for which standards are to be developed were nowhere identified or listed; 9. In developing the procedures for estimating the long term aquatic and terrestrial organism no effect levels from corresponding short term toxicity data, only 80% of the aquatic and terrestrial species populations are intended to be protected. Thus, 20% of the species populations may experience some, perhaps significant adverse toxicological effects; 10. When compounded error, methodological uncertainty and as yet unidentified and unquantified uncertainties are taken into account, the proposed procedures cannot be demonstrated to assure the protection of the public health or the public trust at the design risk to a degree of statistical confidence considered adequate. I ask that the following portion of this letter be read to the NRC on 20 January 1983 at the beginning of the period of public comment on the Rule 57 Advisory Committee Report: 10 date I am not satisfied with HBO responses to the fundamental issues of morality, constitutionality, legality, jurisdiction, authority, procedure and adequacy of administrative scope and technical detail raised in my comments on proposed Rule 57 and attendant procedures recommended by the Rule 57 Advisory Committee, But that is not to say that I do not appreciate the efforts of Robert J. Courchaine and the then Water Quality Division staff, especially John Robinson, in defending a very technically and politically complex and increasingly contrOversial position. A few years ago Manhattan College krofessor Donald O?Connor, an author, teacher and researcher whose work is widely re? spected in the field of water quality modeling, asked me to say ?Hello!" to Bob and convey his warm regards when the moment was opportune. Don O'Connor remembers Bob as one of the foundin fathers of modern Nitrogenous Biochemi- cal Oxygen Demand BOD-N) modeling, his work one'oflthe pillars upon which the Gdifice of modern environmental fate modeling rests. Like Don O'Connor I would like to express my appreciation and thanks. 'It has been a pleasure and a privilege to work with Bob in his professional capacities as Executive Secretary of the NRC and as Chief of the Water Quality Division, and I hope to continue our person- able professional relationship in his new-rale as Chief William D. Marks 17 January 1983 p. 4 of the DNR's Environmental Services Division. Quiet men often go publically unpraised yet privately appreciated. Let me add well-deserved public praise for the technical excellence, good spirit and professionalism Bob brought and will continue to bring to state public service. Over the next few years many of the most fundamental issues to confront our nation and our world will be addressed in precedent-setting determinations by the WHO. The greatness of this country lies in its encouragement of freedom of expression and the exercise of the right to redress grievan- ces with one's government without fear of reprisal. I do not exercise such rights and freedoms Yet even more fundamental to the American way of life is the inalien- able right to life; a by our founding fathers in the Declarationec: Independence. The Foresight society'intends to wage a vigorous defense of the right to life of the statistical citizen before the WHO, not to disrupt or slow the process of issuing toxic substances water quality standards, but rather to assure the people that all the important issues were given fullest consideration in their development and that all requirements of due process of law were met in their promulgation. The issues of adequacy of risk estimate accuracy, adequacy of compensation for inherent risk estimate uncertainty and the adequacy with which a balance is struck between individual health and social benefit go to the heart of the democratic process. The outcome of the upcoming great debates on these issues will determine not only the fate of the right to life in our democracy but the fate of the Great Lakes resource and the unborn generations for whom its quality, productivity, utility and valuetare to be preserved and protected. It is a task too great for any one person or organization. I ask that you join with the Foresight Society in a spirit of cooperation to lay the foundation for a future upon which we will all be willing to bet our very lives. In Attachment y: I address thelimitations in cost?benefit analy- sis as a decision-making tool in the development of water quality standards. Thank you for considering my views. I ask that a copy of this letter be provided to each member of the NRC. ts. Si 'erely, 0 CC I "7 ?hm?n A I ?Larry- . Fink, M.S. Dired Governor Blanchard C. Hhiteley, Chair. NRC S. Aust, Chair, TSCC H. Cooper, Chair, MERE H. Tanner, Director. DNR K. Courckaine ATTACHMENT VI NCES WATER QUALITY-BASED EFFLUENT LIMIT PROCEDURES AND COST-BENEFIT ANALYSIS Cost-benefit analysis, as presently practiced, is a flawed methodology. Although.the list is by no means exhaustive, I believe that cost-benefit analysis ignores the following harsh economic realities in attempting to balance individual health and social benefits: 1. the prices of things and the taxes on things do not reflect enduring moral, social or environmental values; 2. the use of the Great Lakes resource as a garbage dump for toxic substances has no* price and is untaxed, at present; 3. while a human life cannot be purchased at any price, when life is wrongfully taken, jury awards of from 1 to 10 million dollars compensation are not uncommon; 4. the state does not require a deposit and degree of insurance fromrthe facilities it licenses to discharge toxic wastes to its waters sufficient to cover the potential liabilities incurred by those facilities; 5. the expression of propagated genetic damage, like compounded interest, grows exponentially, slowly at first, and then, as the genetic error principal passed on to reproducing pairs grows, with ever inr creasing swiftness; the risk of underestimating the incidence of gene damage-related health effects, and their attendant economic costs, to as yet unborn future generations outweighs the social benefits claimed for the proposed procedures; 6. the uncertainties in toxicity data, exposure and risk models and model calculations, like interest, are com? pounded by the modeling process; the accuracies of the econometric models projecting claimed benefits and the risk assessment models projecting claimed toxics risks are .nnquanxtiftble:balancing one meaningless set of numbers against another is a futile exercise and.intellectual charlatanism; 7. a fair and equitable schedule compensating the state for the decrease in the quality of life based on the increased risk of gene damage-related death and injury the state intends to accept on behalf of its citizens has not been drawn up. Jupiter. Shown in Fig. of Jupiter made by Rt Itiietnial emission fro the planet is clear! surrounding it is the emission from her i tion belts. These structur are a ogbus to the Van Allen radiation belts around the earth. Observations of radio events on the sun with the VLA have been carried out by groups from the California Institute of Technology. the University of Mary- land. Tufts University. and others. With the high resolution and high frequencies of the VLA solar radio astronomers can observe deep into the solar atmosphere to see radio emission associated with the sites of origin of major ?ares. A VLA radio map of a solar active region ob- tained by Velusamy and Kundu (9) is shown in Fig. 10. The radio contours are superimposed on an optical photograph in the hydrogen alpha line taken by R. Robinson at Sacramento Peak Observa- tory. Among the many complex problems being investigated by solar radio astron- omers with the VLA. there is one specif- ic theme that occurs with great frequen- cy. With the VLA one can make very good. high-resolution maps of potential ?are sites. This makes it possible to of locate and study radio emission from material participating in the motions and acceleration processes involved in the conversions of energy between titagnctic ?elds and plasma which are basic to the physics of active regions and ?are sites. A ?nal example solar system studies is the observation of asteroids. C. M. Wade. K. J. .lohn- ston. and P. K. Seidelmann are using the VLA to observe and track Ceres and other asteroids. This is one of few cases where both radio and optical emission are due to exactly the same (thermal) processes in the same physical regions. Thus successful simultaneous tracking ol' asteroids with the VLA and optical as- trometric telescopes will allow the radio and optical observing reference frames to be established with respect to each other to high accuracy. Future of Astronomy with the Very Large Array In the survey above I have had to neglect the vast majority of scheduled VLA observing programs. An outline of these programs summarizes the expect- ed role of the VLA in the coming dec- ades. I have not discussed observations of comets. moons around solar system Formaldehyde: A Question of Cancer Policy? Frederica Perera and Catherine Petito In 'what may constitute a test case for a new federal cancer policy. the Formalde- hyde Institute. an association of formal- dehyde producers and users. has advo- cated that formaldehyde not be regulated by the federal government despite recent studies showing that the substance causes tumors in animals and despite evidence that there is considerable hu- man exposure to formaldehyde. The in- stitute has argued that the animal data do not provide a sufficient basis to regard formaldehyde as a likely human carcino- gen and that federal regulatory agencies SCIENCE. VOL. 216. Ill JUNE should await the development of conclu- sive human (epidemiological) data before taking protective action. This position contradicts principles for assessing carcinogenic risk that have been widely accepted by the scienti?c community for over a decade and em- bodied in policies of regulatory agencies following deliberations of broad-based scienti?c panels. These principles assert that confirmed Positive .inimal data are presumptive evidence of carcinogenicity in humans: that with current information and methods it is not possible to estab- I 3- Jill/ll - a?womr Copyright I952 planets. ordinary stars. double stars. llare stars. pulsars. gaseous iiclnilas. iio- i'as. supernova x-ray sources. interstellar molecules. in- terstellar neutral hydrogen. the stiucture of nearby spiral galaxies. supernovas and gaseous nehtilas in other galaxies. or the full variety of radio phenomena in other radio galaxies and quasars. All of these have been and will continue to he observed by astronomers using the For astronomical observations at centimeter and resolutions front 0.05 are second to a few are min? utes. the VLA will probably continue to be the dominant instrument for at least the next two decades. References and Notes I. A. R. Thompson. B. Clark. C. M. Wade. P. J. Napier. J. Suppl. 44. [Si 2. J. Burns and W. Christinrtsen. Nature thtdiml 281. 20 1. S. P. Ewald. thesis. New Mexico institute of Mining and ?l?eehnology Hitli. R. T. Newell. thesis. New Mexico Institute of Mining and Technology twill. . R. M. Hjellining and K. J. Johnston. .-tstroplivs. 1. Leu. 246. Ll-ll i198?. . B. Mangon. Eric-m1 215. 247 ?982). . P. Bowers. J. Johnston. J. Spencer. iropln's. 1.. in press. . J. Roberts. Ci. Barge. R. C. Bignell. in prepara- tion. 9. T. Velusamy and M. R. Kundu. in Radio Pinni- iu of the Sim. M. R. Kumtu and T. Gergely. Eds. tReidel. Boston. l9ROi. pp. 10. The National Radio Astronomy Observatory is operated by Assocnized Universities. Inc. un- der contract With the National Science Founda- two. ?do LII 00 lish threshold or no-effect levels that can be reliably applied to the human popula- tion; and that positive human epidemio- logical data are not necessary to con- clude that a chemical substance poses a signi?cant human risk (I). In fact. feder- al agencies have regulated such Sub- stances as pesticides. hair dyes. food additives, and industrial carcinogens (for example. B-propiolactone and ethylene- irnine) in the workplace primarily on the basis of results in experimental animals (2). These principles are consistent with the accepted social policy that it is pref- erable to err on the side of caution in interpreting the available scienti?c data in order to avoid failure to regulate a serious health hazard. Thus. acceptance by federal agencies of the indUIilr'L' nosition regarding the risk Db by e~ E-isure to formaldehyde could overturn established procedures for assessing and regulating carcinogenic Frederica Perera is a staff scientist at the Natural Resdurees Dell-nae Council. New York l?lfi?. and assistant clinical Prolcwi'r. IJiVision of l-?nvironrnentat lleziltli .?ieienccs. sitv School of Public lle.ilth Catherine l'etito is a science associate at the Natural Detense Council. New York 1235 int. n. substances general. During the last 3 months. Protection Agency tl-ll?i?ti and (tcettpational Safety and Health Administration tl oili- cials have reVersed prior stall? recom- mendations to initiate regulatory action to limit human exposure to formalde- hyde (3-37). _ln February l?ttll. EPA de- clined to regard formaldehyde as a prior- ity candidate for regulation under the Toxic Substances Control Act (TSCA) it?ll. Concentrations of more than 8 parts per million t?Jt, (lit: to 4.2 tom; to 3.4 ilm have been measured in the wotkplacc. in mobile and in U.S. houses insulated with urea-forni- foam. respectively. while lev- els frequently range above tt.l in urban airtI-t-hottr average] The pres- ent U.S. occupational standard for l'orm- aldehyde is 3 (time-Weighted 8-hour average) on the basis that the animal data may not be relevant to humans. that there is an absence of positive human data. and that it has n0t been established that. at hu- man exposure levels. the risk of cancer is ?probable and would be high" Ac- cording to sources qu0ted in Inside EPA. not only does the agency's decision not to move quickly on formaldehyde reflect a "clear divergence from current federal policy." but Deputy Administra- tor John Hernandez has made tentative plans to totally revamp the agency's can- cer policy.? t4). Our purpose in this article is to review in detail the data on the carcinogenicity of formaldehyde in light of established guidelines for assessment of carcinogens ic substances to see the extent to which the recent federal agency decisions rep- resent a major policy change. Background Formaldehyde is a versatile chemical used in the manufacture of such products as particle board. ply? wood. paper. home insulation. material polymers and resins. leather and agricul- tural products. permanent-press fabrics. preservatives. embatming ?uids. drugs. and cosmetics. About 7 billion pounds of formaldehyde are produced each year. making it the 26th largest volume chemi- cal in the United States t7). An estimated L4 million people are exposed to formal- dehyde in the Workplace: ll million peo- ple may breathe vapors in the home released by construction and insulation materials: and \irtually the entire popu- lation comes into contact with the chemi- cal because of its ubiquitous presence in polluted air and in consumer products l2? Thus. in the fall of I980 there was widespread recognition of the signi?~ cance of an interim report from the Chemical Industry institute of Toxicolo- 8y (CIIT) that formaldehyde was carci- nogenic in rats tth. According to the ?nal report at the end ofa 30-month period squamous cell nasal carcinomas were observed in 103 of 232 rats exposed by inhalation to 14.3 01' formalde- hyde. in 2 of235 rats exposed to 5.6 of formaldehyde. and in 2 of 225 mice exposed to 14.3 of formaldehyde. No such nasal cancers were found in 236 rats exposed to 2 of formaldehyde or in the control animals. P'otypoid ade- nomas were reported in all exposure groups and in one male control rat. By February t98t. various groups ofexperts had reviewed the interim data. including a federal panel convened at the request of the Consumer Product Safety Commission (CPSC) under the aegis of the National Toxicology Program (I4) and the Environmental Cancer in- formation Unit of the Mount Sinai School of Medicinett?Sl. The NTP report stated that "Formaldehyde should be presumed to pose a risk of cancer to humans" in agreement with the Mount Sinai conclusion that HCHO is a carcinogen in rats and. data sug- gest. in mice at exposure levels comparable to those found in some home and work environ- ments. These ?ndings indicate that effective controls should be initiated to reduce or elimi- nate human exposure to HCHO [15. p. Meanwhile. experiments at New York University I7) showed that exposure of groups of tilt) male rats to formaldehyde and hydrogen chloride separately and combined. at average concentrations of and It) ppm. respectively. resulted in an excess of 7 icatly continued nasal inotnas in rats exposed one and none in the controls in exposed to alone. (?om positre to HCHO and HCI pi.- ahont the same number of ?him logically contirmcd nasal stntantom co" carcinomas as alone lib}. the Cll'l' study. no grossly visible spun? - taneous nasal tumors of this type but! been observed in control rats at th..i laboratory over a period of many yeait U7. in the spring of t98t. on the basis int ,thc Cll'l' study and a review of available data on torntaldehydc use and human exposure to the chemical ill/t. EPA drafted a Federal Rt'L?fot?l? notice under ?4tll of TSCA designating formaldehyde as a priority chemical for regulatory .tH- sessment The draft 4th notice stat- ed: EPA has determined that there may be .i reasonable basis to conclude that some expo- sures to formaldehyde present a signi?cant risk of widespread harm to humans. There- fore the Agency is initiating action to investi- gate those exposures ofgreatest concern ul?ll! determine whether they lead to unreasonable risks The notice was not signed by the Admin- istrator of EPA. Rather. during the sum- mer of l98l EPA Deputy Administrator John Hernandez convened a series ot unannounced meetings?termed ?sci- ence courts??primarily attended by EPA and Formaldehyde Institute repre- sentatives in order to review the scien- ti?c data on formaldehyde tlft. A con- gressional subcommittee was critical ot this signi?cant departure from the ac- cepted peer review process (3. 22). On 4 September l98l. the Natural Resources Defense Council re- quested an explanation of EPA's failure to act on formaldehyde under ?4t0 of TSCA and noti?ed the agency of its intention to seek judicial review of that failure under #530 of the act. On it Sep- tember l98l. a memorandum from Don Clay. Office Director of the EPA Ot?ce ofToxic Substances to John Tod- hunter. then Assistant Administrator Designate for Pesticides and Toxic Sub- stances. recommended against treating formaldehyde as a priority for assess- ment under ?4tft of pending addi? tional epidemiological information in parallel developments at OSHA. in July l981 an OSHA otlicial recommend- ed revcrsul of a prior decision to release a bulletin on formaldehyde jointly with the National Institute for Occupational Safety and Health tE-Jt. ?l'be NIOSH Current Bulletin had stated that formaldehyde should he SCIENCE. VOL. 2th handled as a potential cilttigclt and that ttp? should be imposed exposure (9). fly-M a. ficd a 26 October 1981 petition is United Auto Workers at for an emergency standard for for OSHA's ac- tion in putting aside new standards on formaldehyde and other substances trig- gered concern that new of?cials at OSHA were likely to revise the agency?s cancer policy (5). On August I9SI. Arthur Upton. Chairman of the NYU Medical Center Institute of Environmental Medicine. wrote to the heads of federal agencies that formaldehyde is "decisively carci- nogenic in animals? and "if the carcino- genicity of formaldehyde is ignored. it would mean that no agent could be re- garded as carcinogenic in the absence of positive evidence in humans" This letter prompted a response from Joel Bender. Chairman of the Medical Com- mittee of the Formaldehyde institute. that ?to regard formaldehyde as a likely carcinogen in man is not supportable" (26). By contrast. in October l981 a working group of the international Age n. cy for Research on Cancer tlARCl con- cluded. on the basis of the and NYU studies. that formaldehyde gas is carcinogenic to rats and should be con- sidered ?for practical purposes." in the absence of adequate data in humans. as if it represented a carcinogenic risk to man (27). On 29 January 1982. a detailed letter to .the assistant secretary for OSHA. the EPA administrator. and the chairman of the CPSC from Upton and l. B. Weinstein of Columbia University rec? ommended prudent measures to restrict exposure to formaldehyde: worker it has come to our attention that EPA. OSHA. and possibly other federal regulatory agen- cies. may be planning not to take immediate protective action on formaldehyde. in spite of substantial evidence for its carcinogenicity from animal bioassays. We are concerned about the possibility of such a departure from established public health policy. it would con- ?ict with the prevailing views of the scienti?c community and would set a precedent which could hamper future regulatory action on oth- er carcinogens. There is general agreement among experts in chemical carcinogenesis that a substance which causes cancer "1 signi?cant numbers of experimental animals in well-conducted us- says poses a presumptive carcinogenic risk to some humans. even in the absence of con?r- matory epidemiological data. While negative human data can de?ne the upper limit of risk to man. there is no recognized method as yet for establishing the existence of a threshold for a carcinogen in the human population. These principles. which are accepted throagh- out the world. have served for many years as l8 JUNE 1982 the hosts for sound health policy and regulatory action on carcinogens. To compare our Vic-?rs on this oibicct tutti those of otir colleagues. we have consulted several of the tuitltl's leading authorities on chemical carcinogenesis for their opinions. The replies we have received from them thus far are unanimous in supporting the principle that de?nitive demonstration of carcmogcnic- ity in well-Conducted animal bio-assays saf- ?ccs to provide evidence of presumptive car- cinogeuicity for the human population 12M. A week later. the American Cancer Soci- ety issued a statement urging regulatory agencies "to set appropriate standards to minimize occupational and public expo? sure to the chemical. its industrial prod- ucts and applications" t29i. On [0 February I982. Todhunter. EPA Assistant Administrator for Pesticides and Toxic Substances. formally recom- mended against considering formalde- hyde as a priority candidate for regula? tion. Characterizing formaldehyde as a ?potential animal carcinogen" be ob- served that concern about human carci- nogenicity should be "tempered" by the observations that quantitative and possibly qualitative results of exposure to formaldehyde appear to depend highly on exposure level. specres. and route: that rats seem to be particularly sensitive to formaldehyde: and that long human experi- ence does not seem to indicate any pressing concerns . . . By contrast. the CPSC voted on 22 February 1982 to ban urea-formaldehyde foam insulation Canada and the states of Massachusetts and Connecticut had previously banned the use of urea- forrnaldehyde foam Validity of the Data Questions have been raised about the validity of the animal data (26). Howev- er. the study was rigorously peer-reviewed and is considered to be valid tl4. 15. 27. 32}. The possibility of a viral reSpiratory infection confounding the data in the CIIT study was considered unlikely in the NTP and Mount Sinai reports tld. l5). Control animals had also shown signs of viral infection but did nor devel- op tumors. Further. in some of the rats. nasal cancers had formed by the time respiratory viral infection occurred In the NYU study con?rming the OFT ?ndings. :1 sample of the animals was tested for the virus and found to be negative in the Cll'l? study mice were not all'ccted by the virus yet they developed tumors. Although it is unlike- ly that the transient viral infection con- tributed to the carcinogenic response of people to the chemical may also c\pcricncc viral iti- fcctions of the upper respiratory tract l5). Addressing the criticism that the Cll'l' study is llaucd because in?ammatory lesions" present in nasal mucosa. CPSC stalf scientists have written that pathologists who examined the slides from the Cll'l? study did not observe such changes 135). The NYU studies provide con?rma- tion of the Off results in a different strain of rats tin. l7}. According to Up- ton li7). the studies "provide decisive con?rmation of the Chemical industry Institute of 'l'oxicology ?ndings that formaldehyde induces squamous cell carcinoma in rats." Partly because the type of tumor was not that associated with it is judged unlikely that the formation of BCME as a result of combination of formaldehyde and was responsible for the excess squamous carcinomas in the nasal cavities 19. H. 36). The fact that formaldehyde alone produced about the same number of tumors as when combined with also argues against an etiologic role for BCME. The prior reports of negative results in three long-term inhalation studies of formaldehyde do not detract from the signi?cance of the and NYU stud- ies. According to the NTP report tNl. all three had shortcomings in experimental protocols and execution (for example. high mortality. inadequate exposure. 0r de?cient histopathologyl. Bioassays in which other routes of exposure were used have been similarly limited; howev- er. some give de?nite clues that formal- dehyde may be carcinogenic to a variety of target tissues and animal species Formaldehyde Data in the Context of Established Cancer Policies A number of principles have been elaborated reports written during the last 10 years by various scienti?c com- mittees concerning the assessment of human risk l'rom environmental carcino- gens. Composed of scientists aililiazed with academic institutions. industry. and government. these committees were broadly representative of the scienti?c community. Their reports included those of the National Cancer Advisory Board. the lnteragency Regulatory Liaison Group thLG). the National Research Council the Food Safety Coun- cil. the Office Assessment ti). and the Occupational Safety and Health Administration as well as publications by the IARC and the l2!? 1-.- New York [\L?iltlumy l37?45l. Developed cooperatively by representa- tives of all federal agencies concerned with toxic substances control. the guidelines were based on an extensive review of the scienti?c literature on carcinogen assessment. EPA formally adapted (in l979l the IRLG policy for purposes ol' evaluating evidence regard- ing suspect carcinogens as a supplement to its own Interim Guidelines for Carci- nogenic Risk Assessment (46. 47). In particular. the report of the Toxic Sub- stances Strategy Committee published in 1980 is signilicant because it was based on a review of 23 major re- ports written betwecn 1956 and [979 (48). The TSSC. composed of represen- tatives from federal agencies with re- sponsibility for controlling toxic sub- stances, identi?ed ?principles and tech- nical considerations underlying federal policies for the identi?cation of potential human carcinogens?: Although they have been the subject of con- siderable public misunderstanding. these prin- ciples are widely supported in the scienti?c community and in the deliberations of rule- making and adjudicatory bodies. the courts. expert committees. and international agencies [48, p. Even more recently. the Of?ce of Tech- nology Assessment has reviewed and rea?irmed the basic principles of carci- nogenicity assessment (1). These princi- ples are as follows. I) Animal testing data from preperly designed and well-conducted tests are adequate for concluding that a chemical substance is a likely carcinogen in hu- mans. Over 30 chemicals or industrial pro- cesses are judged by the to be carcinogenic or probably carcinogenic to humans on the basis of epidemiological evidence Of those for which animal data exist. all (with two possible excep- tions) have been positive in experimental animals. The two possible exceptions are benzene and arsenic. However. there is evidence from two recent bioassays that benzene is carcinogenic in animals and that arsenic may be a cocarcinogen capa- ble of inhibiting DNA repair L50. Hence. the has concluded: In the absence of adequate data on humans. it is reasonable. for pracrical purposes. to re- gard such chemicals [for which there is sulf- cient evidence of carcinogenicity in animals] as if they presented a carcinogenic risk for humans [52. p. Thus: All Federal agencies accept a positive blous- say result in a single species as evidence that the substance is a potential human carcinogen ll. p- 13!- I288 at Animal testing at high tithe levels is a valid and necessary procedure for iden?i tifying potential human carcinogens. The exposure of experimental animals to tox- ic agents in high doses is a necessary and valid method of discovering possible carcinogenic hazards in man [39, p. Since carcinogenic response is usually dose related. the biological and statistical sensitiv. ity ol'a bioassay may be enhanced by increas- ing the exposure levels of the test substances rather than the less feasible alternative of increasing the number of test animals to match the human population at rislt [42. p. 5093]. The basis for such extreme doses. must sim. ply stated. is to maximize the sensitivity of the test and its capability of detecting irre- versible molecular events leading to neOpIas- tic transformations of cells which could also occur as the result of low level exposure [40. p. 127]. This method is valid as well as practical and necessary because The intrinsic carcinogenicity of a chemical does not depend on dose level although the proportion of animals developing cancers and the earliest time that tumors are detected are usually related to dosage [48. p. Therefore. were environmental levels considerably lower than animal dose lev- els. this would not invalidate the results of the testing for purposes of human risk assessment. However. in the case of formaldehyde. OTS sra?? (7), the Mount Sinai committee and the NTP panel (14) have all pointed out that the tumors have occurred in rats at levels compara- ble to those encountered by humans. b} Results in laboratory animals are qualitatively relevant to humans since the overall patterns of metabolism are generally similar. although the type and site of cancer induced may ditfer. Basic biological processes of those molecular. cellular. tissue and organ functions that con- trol life are strikingly similar from one mam- malian species to another [50. p. 85}. . . . [Metabolic studies have shown that most di??erences between humans and experimen- tal animals are quantitative rather than quali- tative and support the idea that animal results can be used to predict human responses p. 126]. For example. the large body of infor- mation on the metabolism of benzolalpy- rene shows that the overall pattern of metabolism in all species and systems tested is the same talthough the carcino- genic potency may differ] (53). It should be noted that several scienti?c panels have raised the possibility that humans may. in fact. be more vulnerable to cer- tain carcinogenicsuhstances than labo- ratory animals t39. 50. Weighing the usefulness of metabolic studies in human risk assessment. OSHA concluded: I "trol. lot the purposes of negating it .tlion or of potential .. . I carcinogens. Information on and pharmacokinctics is ot'extrcu metical value at the present time - . There is also scienti?c consensus It. a negative human ell'ect cannot be cm. - cludcd from evidence that the target . the carcinogen dill?ers in humans lt?ur . that in experimental animals. in expel mental carcinogenesis. the type and of cancer seen may or may not he ll?. same as that recorded in human SIUtllt' For example. indu. bladder cancer in man. monkeys. tit-g;- and hamsters but hepatic cancer in rat (43). Present knowledge indicates that . . . the 1.. spoosive target tissues or organs and types of tumors induced in different spctit may vary greatly. Therefore . . the ?ndine. negative results in some other species gCll'Jl ally does not detract from the validity t-l positive result as evidence of Carcinogenitn for the test substance [33. p. 39866]. Speci?cally. as regards formaldehyde the CPSC states: There is no evidence of biological difference- between the laboratory animals tested humans that would decrease the potential lw humans to develop cancer when exposed t. formaldehyde This is in agreement with the ?ndings the OTS staff and the NTP panel. illit! formaldehyde metabolism and its reac tion with cellular components is qualitn tively the same in all mammalian Speth- examined to date. Including man (7. The reports also concurred that. at though formaldehyde caused nasal can- cer in rats. this may not necessarily an. the site affected in human beings. 2) It is not now possible to extrapolatt from animal data a ?safe" population threshold for any carcinogen regardie? of the mechanism of action. [The] position that there is no presently .lL' ceptable way to reliably determine a thresh= for a carcinogen for any given population amply supported by the evidence present-:. and also represents. to a large extent. a eon sensus of scienti?c opinion [42. p. 5137}. Methods do not now exist for determinine safe threshold level of exposure to carcim- gens. The major obstacle to determinin- whether there are safe threshold exposun levels for carcinogens is the lack of data the effects of low exposure levels. . . . it cause there is not de?nitive evidence of It'- existence of thresholds and because not .- cancer variables have been identi?ed. pr!t deuce requires that no safe level thresht-h- be assumed to exist. . . . Exposure to .m amount of a carcrnogen. however small. mus be regarded as an addition to the total carciw genie risk [48. pp. The self replicating nature of cancer. It- multiplicity of causative factors to which im- SCIENCE. .. 2? . 1" L'it'll runny?. HIV "taint?. tutu possible ct in ol' cllects. and the wide range of tots -plibil- ities work together in rently unreliable to A ?t a . human popula . - has no ell?ect on . . Illt is gene ly accepted that a popula? tion threshold ould define a ?risk? free" dose fora pof people composed of diverse individuals. if it exists. cannot now be demonstrated ll. p. Ill. This principle applies to any agent that contributes to the carcinogenic process: "It would not be practicable or justi?able to establish dill'erent criteria lor the identifica- tion. classification. or regulation of initiating and promoting agents. OSHA agrees with the NCAB Subcommittee that "any factor or combination of factors nhich increases the risk of Cancer in humans is of concern regard? less of the mechanism of actiOn? [43. p. Still. The mechanisms by which individual carcinogens induce tumors are not easily understood. Even where it may be possi- ble to de?nitively classify a substance or agent as an initiator. promoter. or com- plete carcinogen. these distinctions are not practically useful for purposes of regulation.? This is because ofthe impos- ?sibility of identifying ?papulation? thresholds for any carcinogen regardless of't'hef'm??hhnism by which it operates. Under de?ned experimental conditions. a given substance may appear to show a threshold level (that is. a dose level below which it does net increase tumor incidence). However. as indicated by a large-scale bioassay in which a threshold could not be identi?ed for acetylamino- ?uorine (54). such observed thresholds may simply re?ect the limited ability of the test system to detect etfects at low dose levels. Even where experimental thresholds could be established with cer- tainty. it would be impossible with cur- rent information and methods reliably to predict from experimental data the threshold level in humans L55). Such predictions are precluded by the dif?cul- ty of obtaining quantitative data at very low dose levels in small numbers of animals. variations in human host re- spouses. and possible additive or syner- gistic elfects of Other agents that individ- uals might be exposed to. Thus. regu- latory agencies have sought to reduce human exposure to carcinogenic sub- stances to the lowest possible level con- sistent With relevant social and economic considerations. From the above discussion it is clearly not necessary that the mechanism of action be de?nitively estahhshed before formaldehyde is identiticu a carcino- gen. Howeverfthe Formaldehyde insti- tute (26) contends that formaldehyde is It! JUNE I982 lth'ly ll, (ILI mechanism?causing only at high doses through its cytotoxic or irritant. action?and is therefore probably a "threshold" carcinogen. This view is also implied in the memoran- dunt The epigcnetic mechanisms suggested include cell destruction and rapid cell proliferation triggered at high dose levels that prevent DNA repair and detoxi?cation systems front operating cf- fectivcly and induction of ulceration. irreversible hyperplusia. 0r metaplasia only at high dose levels which are prema- lignant in themselves or serve to pro. mote tumor formation to. 26). The OTS. the NTP panel. and CPSC stall have rejected this interpretation. citing the substantial body of evidence that shows formaldehyde to be genotoxic and the absence of factual support for the epigenetic or "threshold" theory (7. I4. 56). While the promoting ell'ect ot?l?orm- aldehyde may play a part in its overall carcinogenicity. formaldehyde is a po- tent all-tylating agent t57l: is a mutagen in a wide variety of test systems including microbial. insect. and mammalian sys- tems t8): induces sister chromatid ex- change in human and causes unscheduled DNA repair in HeLa cells Formaldehyde is able both to transform mammalian cells in culture at low concentrations and to initiate cell transformation in vitro (60. Formal- dehyde also enhances the genotoxic ef- fect of peroxides and radiation H4). Thus. with regard to the hypothesis that the ?carcinogenic etfects of formalde? hyde are indirect. termed epigenetic." 0T8 concluded. "There is . . . absolute. ly no scienti?c evidence for this hypoth- esis in the published literature" t7. p. A-6l. The NTP panel further noted that ?a number of agents were reported to in- duce epithelial hyperplasia in several types of tissues but they had no carcino- genic or tumor promoting activity associ- ated with them" p. 34). The panel ?found no evidence that the induction of irritation or. more speci?cally. ot'epithe- lial hyperplasia is a suf?cient coodition for the carcinogenic activity of an agent" (14. p. 34). The NYU studies (17) sup- port the NTP panel on this point. If severe irritation and resultant rapid cell turnover were either a sul?cient condi- tion or a necessary prerequisite for carci? nogenicity of formaldehyde. one would expect an increased eITect in animals exposed to a combination of formalde- hyde and a strong irritant. However. in the NYU study HCHO and were administered singly and combination: there was no increased response to the - I l\ I.?l 'ascotts irritant: nor did alone cause tumors although it did hyperpla- sta (In. Scientists from the (PST responded to the Todhunter memorandum. citing evidence against the ?threshold" theo- ries above and rejecting the assertion that the observation of endogenous lev- els of formaldehyde in animals without Spontaneous tumors indicated :1 thresh- old tdjl. They reallirmcd their prior con- clusion that ?there is no evidence dent- onstrating that there is a threshold for formaldehyde. or a dose level below Which it is certain that formaldehyde will not cause cancer? 3) Positive human data are not neces- sary to regard an agent as a likely human carcinogen warranting protective action. Although epidemiological evidence is a necessary prerequisite to actually call- ing a substance at ?human carcinogen." this is a point of terminology rather than a criterion for taking protective action. The most powerful reason is that the usefulness of epidemiology for the iden- ti?cation of carcinogens is limited by a number of constraints. These include cost. the usual long lag between expo- sure and appearance of cancer. the con- founding el?fect of multiple exposures to carcinogens. and dif?culties in identify- ing an appropriate control group I see {48. pp. A practical problem is that very large samples must be com- pared if the risk in the unexposed pepti- lation is low and the number expected to show the elfect is small. Because of the possibility of a false negative result. the. absence of positive results cannot prove an absence of risk: however. an absence of positive results may be useful in plac- ing upper bounds on the magnitude of the risk (64): . . . [Nlegative epidemiological data. ques- tionable because oflimitations in the power of detection of such studies. do not deny the conclusion of carcinogenicity on the basis of animal assays [38. p. 3987?. Thus: When a toxic substance is identi?ed in a mammalian test system tin which the criteria listed in the standards are usedt as a prudent health policy matter this substance is to be treated as posing a carcinogenic risk to human beings. . . . Because public policy mandates preventive health care. waiting for epidemio- logic data is unacceptable. since it means waiting to "count dead bodies" [41. pp. t4- Speci?cally. as regards formaldehyde: epidemiological studies completed to date have not speci?cally designed to evaluate the carcinogenicity dehyde in human populations; thus they I239 have been inCouclusive or suggestive of a positive ctl'ect rather than negative regarding the carcinogenicity of. formal- dehydc. These studies are inconclusive. because of small study site tthe three studies together cover fewer than 2.000 deathst. poor docu- mentation of exposure. possible multiple ex- posures. and study design limitations p. Epidemiological studies conducted to date do not permit a de?nitive evaluation of the carci- nogenic risk of formaldehyde to humans p. Citing the conclusions of the NTP panel. CPSC stated: The epidemiological studies presented at the conference and by the Formaldehyde Institute do not have suf?cient information to be considered conclusive because of the small population size. lack of exposure data. and other confounding factors [35. p. 88]. As has noted. :hree proportional mortality studies of workers exposed to formaldehyde had only very limited (7 to 12 percent) power to detect a threefold excess of mortality from nasal cancer (27). However. a number ofstudies have been suggestive of an increased cancer risk (19. 36. 65). The preliminary epidemiological ?ndings sug? gest that persons occupationally exposed to formaldehyde may experience elevated risks for certain cancers. notably of the pharynx. oral cavity. and hematopoietic sys- tem. brain and skin [65. p. Recently. there have been several re- ports of rare nasal tumors in workers exposed to formaldehyde I66. 67). According to the OTS. ?because of inherent limitations. studies in progress are not likely to resolve present concerns about formaldehyde safety? (7. p. 17). This is true of an ongoing NCI study that Will not be completed for at least 2 years (68). ?Analysis of the design of the stu- dy . . . suggests that it may not be pow- erful enough to assure that formaldehyde is not carcinogenic in humans" (7. p. 27). In summary. the has evaluated the laboratory and epidemiological data On formaldehyde gas and has stated: "There is suf?cient evidence that form- aldehyde gas is carcinogenic to rats. Epi- demiologic studies provide inadequate evidence to assess the carcinogenicity of formaldehyde to man." In accordance with its established policy (52) therefore. the working group concluded that in the absence of adequate epidemiological data. formaldehyde gas should be con- sidered. for practical purposes. as. if it represented a carcinogenic risk to man (27. p. 50). I290 Conclusion Over several decades. the results of scienti?c research have led to policies for assessing carcinogens that have been widely endorsed by the scienti?c com- munity and accepted by regulatory agen- cies in this country and abroad. These policies are based on the principles that valid positive animal data are presump- tive evidence of carcinogenicity in hu- mans and that population thresholds can~ not be identi?ed for carcinogenic sub- stances. These two principles are the heart of present federal programs to pro- tect human health from cancer-causing chemicals. and are consistent with ac- cepted social policy that makes a con- scious decision to err on the side of caution in addressing health risks. The present behavior of two U.S. regulatory agencies regarding formaldehyde sug- gests that they are disregarding the sub- stantial evidence and rationale on which the established policies relating to as- sessment and regulation of carcinogens have been based. Indeed. it appears that EPA is informally revising its cancer policy to decrease reliance on animal studies?a step that could have the etl'ect of substantially delaying or indeed bar- ring altogether protective action on sub- stances such as formaldehyde. pending the development of positive epidemio- logical data. References md Notes I. O?ice of Technology Assessment. Assessment of Technologies for Determining Cancer Risk: from the Enwronment IOTA. Washington. D.C.. June . H. H. Hiatt.l- D. Watson. J. A. Winsten. Eds. The Origin: of Human Cancer (Cold Spring Laboratory. Cold Spring Harbor. N.Y.. . M. Sun. Science ill-1. 525 II98II. . Inside EPA tl6 October 198?. p. IZ. . R. J. Smith. Science 212. I482 Il98ll. . Todhunter. "Review of data available to the Administrator concerning formaldehyde and di- memoran- dom to A. M. Gorsuclt. to February I982. 7. Environmental Protection Agency. Office of Toxic Substances. Options Paper on Funnelde- hyde (Of?ce of Toxic Substances. Washington. DC. II September 3. National Academy of Sciences. and Other Aldrin-dos (National Academy of I VJ 0?th Sciences. Washington. DC. l98 . 9. National Institute for Occupational Safety and Health. Current Intelligence Bulletin No. 14 (I5 April 198?. 10. K. C. Gupka. A. G. Ulsamer. P. Prettss. paper presented at the International Symposium on Indoor Air Pollution and Energy Conservation. Amherst. Mass.. l3 to to October I9Itl. II. This standard can be compared to threshold limit values lTLV'sl for Sweden. Norway and Denmark of I and for the U.S.S.R. of 0.4 ppm. In West Oermanv formaldehyde is listed as an occupational carcmocen IK. Hemminki. Institute for Occupational I-Ieatth. Finland. per- sonal communication. January I982). 12. I. A. Swenberg. W. D. Kurtis. R. I. Mitchell. E. G. Gralla. K. L. l?avkov. ('tmt'rr Her. 40. 3398 ?980}. Chemical Industry Institute of Toxicology. Fi- Report rm a ('I'tmnt?r' Study in Run and Hire Exposed to hyth' IBnttelle Columbus Laboratories. Colum- bus. Ohio. Data released to the Internaencv chcarch on Cancer llARi (it?ottp??xcarch I'nanplc I'arlt. . . (uric-tents . chairman. Rl'pt?l'f tiflht? I not Pour! (National Tour. ogy Program. Research I'rtanglc l'ark. .N.t November I?Jtml. . I. J. ('urt't'mret'ttir'itv of ("mm ("t'nttl Report [Report to the American - Cancer Society. 25 l9ttll. . R. Ii. Albert. R. Sallahnmar. S. Lasktn. Kuschner. N. Nelson. C. A. Snyder. J. Natl Cum-er Inst. 68. 59? ?982). . A. Upton. letter to V. dc Vita and? I7 Augmt - .Chetntcal Imlustrv Institute of Tostcolcxc . Ht?rt?rtrt?h Fine/ur- Dor't'rt't No. ICIIT. Research Park. C. 3 October I979l. . Environmental Protection Agency. View Lrt't't? ul? 'Iiu . Substances. I9 February 198?. . That section prot ldt)?. that hen the Adnunia' tor receives information indicating that circa . cal may pose a signi?cant risk of cancer. lum- defects. or gene mutations. EPA must inm. appropriate regulatory action or publish .. m- ing that the risk Is not unreasonable wnlun it?. days. . Memorandum from Chairman Molfett to them. bers of the subcommittee on environment. cnc? mi. and natural resources. 6 October l9tll. The House of Representatives subcommittee on environment. energy. and natural resources . . the Committee on Govemmenl Operationx Hearing. 2! October "Environment and? industry talks cnticrzed." New York Time: [22 October i9ttll. . Insidi- EPA September I980). pp. 4?5. . M. Cowan. memorandum to Secretary of Thorne Auchter. Health Saj'ery Left. (22 Jul.- 1981). . H. Young. United Auto Workers Petition for Emergency Temporary Standard on Formalde- hyde. sent to T. Auchter. 26 October 5981. . Bender. letter to Auehter. 28 August I?Jtt' The views of the Formaldehyde Institute lt-?c also expressed in S. .I. Byington. Formaldehyd. Institute. memorandum to I. C. Miller. Tas. Force on Regulatory Relief. 23 February 198: H. Demopoulos. testimony before the Depar: ment of Energy. 10 February I931: H. Dentu- peulos. I. Cimino. B. Wagner. "An academ--. review of the adverse health e?'ects formaldehyde." paper presented to the CPSC 20 March I981: and J. Ramcy. chairman or me Formaldehyde Institute. letter to I. Hernandez. EPA. August l98l. . Report of the IARC Working Group on me Evaluation of Caretnogemc Risks of Chemicals to Humans. .?lfonoter.. in press. . A. Upton and I. B. Weinstein. letter to Auchter. A. M. Gorsuch. and N. H. Steorts January I982). - . Statement adopted by the Board of Directors- American Cancer Society. New York City. February I982. . ?Product safety agency bans use of fortnalde~ hyde foam insulation.? New York .Tm" February I982). p. 15. . M. Sun. Science 213. I132 The Massa- chusetts ban has been set aside by the Superio- Court: that decision is on appeal. . Report of the Pathology Group Review of CIIT Studies on Formaldehyde Exposure -u Rodents tChemtcal Industry Institute of Tm: cology. Research Triangle Park. N.C.. Febmart I980): E. G. Gralla. H. d'A. Heck. L. Vv Hmbesh. G. W. Meadows. A Report of for Review of all: Erratum: tChenu- cal Industry Institute ot'Toxicolo CIIT Dock at No. 62620. Research Triangle ark. N.C.. l4 January Proceediues of the If Conference rm ormm- dehyrte Toxicity Publishing. Nev? York. in press). . R. E. Albert. personal communication. Septem- 35. her I98I. Consumer Product Safety Commission. foam insulation [imposed ban denial of petition." Fed. Regist. 46. ?188 1" February P. F. lnl'antc. A G. D. Groth. K. 4 Cho. 1. Ward. Lancet 1981-". 980 H9811. National Cancer Advisory Board. J. Natl. Inst. 58. 4M ?977:. Inter-agency Regulatory I unison Group. Fm Racist. 44. $9853 in July l97'Jt. National Research Canned. .S?ttitmum- Drinking lt?utrr and lNattonal Reseatczi Council. Washington. IJ.C.. I971). . Food Safety Council. Proposed System SCIENCE. VOL. al'ely Council. vim. l' ch on Cancer. l'ttr'tf?uh'h? lut?n Wit-.liuietiiu. . II. R. Ruttenltcie . b. C. Ha ix: I. J. Belikn?'. Amt. . 32?) I W791. entire volume. .- id. 3631193?. entire vol? - ion Agency. Ft?d. Rt'gtn'l. 41.14102 :25 -ay Iowa. .. lbi'u'. 44. 58642 October 1979}. See p. 581347 and Sitti?li. )i'oxic Substances Strategy Committee. Tm'r'r- Cltr'mr't?uls mu! l'rutrt'ttum I Wash- ington. D.C.. May 49. International Agency for Research on Cancer. dimmer. Suppl. Nu. ?9791. 50. D. P. Rall. Aim. Arm]. lie-i. .129. 8? 119791. 5i. in St'r. Pub}. 35 ?9791. 2. International Agency for Research on Cancer. (ARC 20. 7 ?9791. 53. P. Brooke: and M. E. Duncan. Food Science The Gulf Between The Brandt Commission ll) describes the gap which separates rich and poor nations as being so wide that at the extremes pepple seem to live in ditl'erent worlds. The contrast in life?styles is par- ticularly evident in the relative quality of their diets, to which food science has contributed so much for the richer and so little for the poorer. Food Science in Developed Countries Historically. food science has been devoted to an understanding of the bio-_ chemical and biophysical nature and composition of foods. the changes that foods undergo after harvesting. and dur- ing such traditional technological trans- formations as fermentation. milling. dry- ing. frying. baking. boiling. and other forms'ol? cooking. For people in devel- oped countries. food science combines the skills and knowledge of chemists. physicists. microbiologists. nutritional biochemists. engineers. and many other professions to provide the most varied range of wholesome diets in the history of mankind. From large grocery stores people can'choose several thousand dil'o SCIENCE. VOL. 216. Ill JUNE 4t! ll't't'll?. Suits. f'hi'ntria?l t'uriimn- t'i-m'u'i Mimi. k. \lnntesauit. . M. ?and. lids I you. I tame. l?l7-ltis". his . I'itl't'r t" t'lit' Press. New York. WWI 54. N. A. J. ll. hunter. 1). W, tiavltir. (i J. limiter". I'iuhiil. Tutti of. J. i? 55. This statement is based on the passages quoted above and on conversations ?Hit Burns. NYU: I. ll. Weinstein. Columbia University: and T. Slaga. Oak Ridge National Laboratory. 56. R. L. I?irt'krtm- nu Urt'it-Furuinldr'hvile Foam Insulation tConsum- er Product Sal'ety Commission. Washington. 0.0. February 1983. 57. R. J. Wilkins and H. D. Maelcnd. RM. 36. ll tl97nl: N. Magana-Schwenke and ti. Ekcr?t. i'hr'il?. It ?9731. 53. G. (Jhc and Beck. Drug Ale. Dr?pt'l?td. 4. I [979]. C. N. Martin. A. C. Mchi-mid. R.-C. Garner. Cum-er Rat. 38. 362! I 60. D. J. Brusiclt. B. C. Myhr. D. G. Stetka. J. O. Randell. mid Irirrts?irmirti: At'ti?t'ir?v iil? and Nutrition: Rich and Poor Joseph H. Hulse ferent food items at all times of the year. and many spend less than a quarter of their disposable income on feeding them- selves. As much as. ifnot more. than any other branch of learning. food science has made it possible for both parents in a l ittun Repurl. ltuini-tics. Mil. l?lh'llt bl. Raeau and J. lluicILu. (inn I at. U. .125 R. Albert. personal communication. Novem- ber I?ll?. 63. K. Gupta and M. ('ohn. Ill-aim .i'i'i'i'iiri-ii Amity- H't im the Him it! tt'onsumer Product Safety Commission. Wash. ington. l?l I'cbruary l?th?ll. M. C. S. Murr. NJ. Arm}. Sci. 329. 153 tl979t. 65. Blair, presentation to the National Cancer Advisory Board t5 October l981l. lib. lnfante. "Documentation of excess nasal cancer among workers exposed to formalde- hyde." memorandum to the Consumer Product Safety Commission. January 1932. 67. W. E. Halpcrin. M. Goodman. L. Staynor. L. J. Elliott. R. A. P. J. in preparation. 68. According to liltiir. National Cancer Insti- tute. the study it ill not be completed earlier than summer Wit-l lpersoniil communication. Decem? ber grown. and the methods by which they are harvested. stored. and processed. At all stages after harvesting. both during and after processing. changes take place involving so many diverse and complex chemical reactions that it is impossible to follow any one in isolation from the rest. Consequently. the extensive body of knowledge acquired has resulted from both biochemical and biophysical mea- surement and from empirical observa- tion. Many food plants. including those Widely accepted. contain substances un- suitable for ingestion. Some. for exam- ple. contain mycotoxins resulting from infection ofgrains in the natural environ- ment: others toxic substances that prorect them. Primitive people. probably by trial and error. found simple Summary. The people of economically developed countries benefit greatly trom modern food science. They are protected from food contamination. have access to a great variety of food. and need spend little time preparing it. The poor in developing countries enjoy few of the benefits of food science. Their diets are often nutritionally deficient and they spend many hours each day processing their food and searching for wood with which to cook it. In most tropical countries food losses between harvest or slaughter and eventual consumption are inestimable. Efforts to improve post- harvest food systems in developing countries require the attention and ingenuity of many scientific disciplines and the support of all deveIOpment agencies. household to pursue their careers with- out detriment to the adequacy or variety of their family's diet. The raw materials ot?the food scientist are more highly and uncontrollably vari- able than most of those used by inorgan- ic chemists. The properties and composi- tion of the seeds and trans of cultivated plants are in?uenced by genetic back- ground. the environmental conditions of soil and climate under which they are ways to eliminate. or reduce to relatively safe levels. naturally occurring toxins and nutritional inhibitors present in their staple food sources. Typical are the Joseph H. Hulse is Director. Agriculture. Food and Nutrition Sciences. international Development Research Centre. KIG Jill). this article was prepared on behalf of the organizing committee tit'llie ll RAWN ll international t'onlerence on ?Chemistry and World Food Supplies?Hit: New I'rtintiel?s? to he held in Manila. Philippines. 6 to It) December I?m. 1:1 [932 I29l M- . . ?hw?uu- ~rmv-n.- . 18 R. T. . I. F. Doolittle. M. Lewis. - - ., (London) 447 (1982). - 39. .. B. Muller-Hill. . - rad Sci. U. 41.79. 2427 (I982). 40. H. Nick 'al" Proc. Natl. Acad. Sci. U.S.A. . . Nick er al.. 1. Mol. Biol. IGI. "982): Arndt. H. Nick. F. Boschelli. P. Lu. J. Sadler. (bid. p. 439: R. Kaptein and K. uthrich. persoml communication. .A. Steitz. D. H. Ohlendorf. D. B. McKay. W. F. Anderson. W. Matthews. Proc. Natl. Acad. Sci. U.S.A. 79. 3097 (I982). 42. D. H. B. W. Matthews. unpub- lished date. 43. I. T. Weber. D. B. McKay. T. A. Steilz. Nucleic Acids Res. IO. 5085 (IQZ). 44. D. H. Ohlendorf. W. F. Anderson. B. W. Mat- thews. J. Mol. Evol. l9. l09 ?9113). 45. D. H. Ohlendorf. W. F. Anderson. M. Lewis. C. 0. Pnbo. B. W. Matthews. J. Mol. Biol. in press. 46. J. C. Wang. M. I). Barkley. S. Bourgems. Nature (Londoni 25]. 247 ll9'ldl. 47. T. Maniatis and M. l?tashne. Prm'. Natl. Arad. Sci. U.S.A. 70. l5.? (1973]. 48. N. C. Seeman. J. M. Rosenberg. A. Rich. ihid. 13. 804 C. Helene. FEBS Lett. 74. 10 ?977). 49. P. H. van Hippel. in Biological Regulation and Dn-elo menr. R. F. Goldberger. Ed. tPlenum. New ork. i979]. p. 279. 50. M. H. Caruthers. Arr. Chem. Res. 13. I55 ?93m. O. G. Berg. R. B. Winter. P. H. van Hippel. Biochemistry Ill. 6929 ?981}: R. B. Winter and P. H. van Hippel. ibid.. p. 6948. R. 8. Winter. 0. G. Berg. P. H. von Hippel. ibid.. p. 69?. Science, Risk, and Public Policy We are now in a troubled and emotion- al period for pollution control; many communities are gripped by something approaching panic and the public discus- sion is dominated by personalities rather than substance. It is not important to assign blame for this. i appreciate that William D. Ruckelshaus con?dence. The polls show that scien- tists have more credibility than lawyers or businessmen or politicians. and I am all three of those. I need the help of scientists. This is not a naive plea for science to save us from ourselves. Somehow. our Summary. A climate of tear now dominates the discussion of environmental issues. The scienti?c community can help alleviate this tear by making a greater elicit to explain to the public the uncertainties involved in estimates oi risk. Current statutory mandates designed to protect public health both demand levels of protection that technology cannot achieve and are uncoordinated across government agencies. A common statutory framework for dealing with environmental risks is needed. In addition. care must be taken to separate the scienti?c process at assessing risk from the use oi such assessments. together with economic and policy considerations. in the management of risks through regulatory action. people are worried about public health and about economic survival. and legiti- mately so. but we most all reject the emotionalism that surrounds the current discourse and rescue ourselves from the paralysis of honest public policy that it breeds. I believe that part of the solution to our distress lies with the idea that disci- plined minds can grapple with ignorance and sometimes win: the idea of science. We will not recover our equilibrium without a concerted effort to more effec- tively engage the scienti?c community. Frankly. we are not going to be able to emerge from our current troubles with? out a much improved level of public I 026 democratic technological society must resolve the dissonance between science and the creation of public policy. No- where is this more troublesome than in the formal assessment of risk?the esti- mation of the association between expo- sure to a substance and the incidence of some disease. based on scienti?c data. Science and the Law at EPA Here is how the problem emerges at the Environmental Protection Agency. EPA is an instrument of public policy. whose mission is to protect the public health and the environment in the man- 52. L. (iuarentc. S. Nye. A. M. Ptashnc. Prm'. Natl. Anal. 36!. USA. 79. 274 ?982]. . 5.3. A. N. Irwin. M. Milne. Cell 31. 54. D. K. Hawley and W. R. .J. Mol. Biol. 493 ?932); (all 32. 327 (I 55. We thank R. T. Sauer for co 'cating results in advance of publication and 0 . Pabo and M. Ptashne for helpful com 3. script. Supported in pan . - GMZBUR and (3me (Y. a grant from Charitable Trust. NSF .. a grant from the a Research Council through Protein Structure and Fu an NIH postdoctoral fellow . Address correspondence to . Matthews. ner laid down by its stat That man- ner is to set standards and'r?force them. and our enforcement poems are strong and pervasive. But the we set. whether technology- or a?hlth-related. must have a sound scien?t base. Science and the law arches partners at EPA. but uneasy partr??l. The main reason for the uneasiness ii. I think. in the con?ict between thM?Way science really works and the pul??s thirst for certitude that is written laws. Science thrives on uncert?ty. The best young scientists ?ock intdi?elds where great questions have asked but nothing is known. The gr?test triumph of a scientist is the crud experiment that shatters the certain!? of the past and opens up rich new p?Ires of igno- rance. QB But EPA's laws often *me. indeed demand. a certainty of prwtion greater than science can provi 'th the cur- rent state of knowledge. . . laws do no more than re?ect what' . public be- lieves and what it often from peo- ple with scienti?c cred on the 6 o?clock news. The pit thinks we know what all the bad utants are. precisely what adverse th or envi- ronmental effects they disc. how to measure them exactly a?oatrol them absolutely. Of course. public and sometimes the law are but not all wrong. We do know a deal about some pollutants and we controlled them effectively by using . tools ofthe Clean Air Act and the Water Act. These are the pollutants? which the scienti?c community can?t safe levels and margins of safety for! lations. If this were pollutants. we could (in both senses of the not so. more easily )1 but it is William D. Ruckelshaus is U.S. Environmental Protect' gave at the National Academy Sciences. Wash- ington. D.C.. on 22 June I983. ?ner-2. VOL. 22: More than 10 years ago. EPA had the Clean Air Act. the Clean Water Act. a solid waste law. a pesticide law. and laws to control radiation and noise. Yet to come were the myriad of laws to control toxic substances from their man- ufacture to their disposal?but that they would be passed was obvious even then. When i departed EPA a decade ago. the struggle over whether the federal government was to have a major role in protecting our health. safety. and envi- ronment was ended. The American peo- ple had spoken. The laws had been passed: the regulations were being writ- ten. The only remaining question was whether the statutory framework we had created made sense or whether. over time. we would adjust it. Scienti?c Realities Ten years ago I thought I knew the answer to that question as well. i be- lieved it would become apparent to all that we could virtually eliminate the risks we call pollution if we wanted to spend enough money. When it also be- came apparent that enough money for all the pollutants was a lot of money. i came to believe that we would begin examin- ing the risks very carefully and structure a system that would force Us to balance our desire to eliminate pollution against the costs of its control. This would entail some adjustment of the laws, but not all that much, and it would happen by about I976. was wrong. This time around as administrator of EPA. i am determined to improve our country's ability to cope with the risk of pollutants over where i left it l0 years ago. it will not be easy. because we must now deal with a class of pollutants for which it is dif?cult. if not impossible. to establish a safe level. These pollutants interfere with genetic processes and are associated with the diseases we fear most: cancer and reproductive disor- ders. including birth defects. The scien- ti?c consensus is that any exposure. however small. to a genetically active substance embodies some risk of an efo feet. Since these substances are wide- spread in the environment. and since we can detect them down to very low levels. we must assume that life now takes place in a mine?eld of risks from hundreds. perhaps thousands. of substances. We can no longer tell the public that they have an adequate margin of safety. This worries all of us. and it should. But when we examine the premises on which such estimates of risk are based. 9 SEPTEMBER 1983 we ?nd a confusing picture. in assessing a suspected carcinogen. for example. there are uncertainties at every point Where an assumption must be made: in calculating exposure; in extrapolating from high doses where we have seen an effect to the low doses typical of environ- mental pollution: in what we may expect when humans are subjected to much lower doses of a substance that. when given in high doses. caused tumors in laboratory animals: and ?nally. in the very mechanisms by which we suppose the disease to work. One thing we clearly need to do is ensure that our laws re?ect these seien~ ti?c realities. The administrator of EPA should not be forced to represent that a margin of safety exists for a speci?c substance at a speci?c level of exposure where none can be scienti?cally estab- lished. This is particularly true where the inability to so represent forces the cessa- tion of all use of a substance without any further evaluation. Functions of Regulatory Agencies it is my strong belief that where EPA. OSHA (the Occupational Safety and Health Administration). or any other so? cial regulatory agency is charged with protecting public health. safety. or the environment. we should be given. to the extent possible. a common statutory for? mula for accomplishing our tasks. This statutory formula may well weigh public health very heavily. as the American peeple certainly do. The formula should be as precise as possible and should include a responsi- bility for assessing the risk and weighing it. not only against the bene?ts of contin- ued use of the substance under examina- tion. but against the risks associated with substitute substances and the risks asso- ciated with the transfer of the substance from one environmental medium to an- other through pollution control prac- tices. recognize that legislative change in the current climate is dif?cult. it is up to those of Us who seek change to make the case for its advisability. But my purpose here is not to plead for statutory change: it is to speak of risk assessment and risk management and the role of science in both. it is important to distinguish these two essential functions. and I rely here on a recent National Academy of Sciences report on the man- agement of risk in the federal govem- ment. Scientists assess a n'sk to ?nd out what the problems are. The process of deciding what to do about the problems second proce- ader array of owani a deci- is risk ma dure invol discipline?; and is a sion about control. in risk management it is assailed that we have assessed the health ?ks of a suspect chemical. We must thei?lctor in its bene?ts. the costs of tho-?variOUs methods available for its eontr?.? and the statutory framework for decibn. The NAS report recommends that ?ese two functions?risk assessment ad risk management?he separated at?mch as possible within a regulatory This is what we now do at EPA am makes sense. it it Risk Assessment or,? as: We also need to strengthen-Mir risk assessment capabilities. We need more research on the health e??ecu of the substances we regulate. I intend to do everything in my power to trike clear the importance of this scientific analysis at EPA. Given the necessity ducting in the face of enormous scienti?c uncer- tainties, it is more important-tho ever that our scienti?c analysis bests-igorous and the quality of our data be?gh. We must take great pains not demislead people about the risks to their ?lth. We can help to avoid confusion both the quality of our seienct and the clarity of our language in expl?ng haz- ards. intend to allocate some EPA 5 increased resources to pursnh these ends. Our I984 request eonta?h signi?- cant inereases for risk assesl?nnt and associated work. We have requested million in supplemental apprwiations for research and developm?, and i expect that risk assessment s: more strongly supported as a re of this increase as well. 'r I would also like to revital' long- term research program to dev? a base for more adequately protectir'the pub- lic health from toxic pollutanml will be asking the outside scienti?c chmunity for advice on how best to focus those research efforts. so in the future. this being ana?tperfeet world. the rigor and thoroughn?e of our risk analyses will undoubtedly.? affect- ed by many factors. ineluding? toxici~ ty of the substances examined,? popu- lations exposed. the pressure regu- latory timetable. and the resou?es avail- able. Despite these often conflicting pressures. risk assessment at EPA must be based only on scienti?c evil?nee and scienti?c consensus. Nothing will erode $027 {49:5 nun Iv Sit?E - ., . he -I- Although there is an objective way to assess risk. there is. of course. no purely objective way to manage it. nor can we ignore the subjective perception of risk in the ultimate management of a particui lar substance. To do so would be to place too much credence in our objective data and ignore the possibility that occasion5 ally one?s intuition is right. No amount of data is a substitute forjudgment. Further. we must search for ways to describe risk in torois that the average citizen can comprehend. Telling a family that lives close to a manufacturing facili- ty that no further controls on the plant?s" emissions are needed because. according to our linear model. their risk is only is not very reassuring. We need to describe the suspect substances as clear- ly as possible. tell people what the known or suspected health problems are, and help them compare that risk to those with which they are more familiar. To effectively manage the risk. we must seek new ways to involve the pub- lic in the decision-making process. Whether we believe in participatory de- mocracy or not. it is a part of our social regulatory fabric. Rather than praise or lament it. we should seek more imagina- tive ways to involve the various seg- ments of the public affected by the sub stance at issue. They need to become involved early, and they need to be in- formed if their participation is to be meaningful. We will be searching for ways to make our participatory process work better. For this to happen. scientists must be willing to take a larger role in explaining the risks to the public?including the uncertainties inherent in any risk assess- ment. Shouldering this burden is the responsibility of all scientists. not just those with a particular policy end in mind. In fact. all scientists should make clear when they are speaking as scien- tists. ex cathedra. and when they are recommending policy they believe should flow from scienti?c information. 1028 What we need to hear more of from scientists is science. I am going to try to provide avenues at EPA for scientists to become more involved in the public dia- log in which scienti?c problems are de- scribed. Lest anyone misunderstand. 1 am not suggesting that all the elements of man- aging risk can be reduced to a neat mathematical formula. Going through a disciplined approach can help to orga- nize our thoughts so that we include all the elements that should be weighed. We will build up a set of precedents that will be useful for later decision-making and will provide more predictable outcomes for any social regulatory programs we adopt. In a society in which democratic prin- ciples dominate. the perceptions of the public must be weighed. Instead of ob- jective and subjective risks. the experts sometimes refer to "real" and "imagi- nary" risks. There is a certain arrogance in this?an elitism that has ill served us in the past. Rather than decry the igno- rance of the public and seek to ignore their concerns. our governmental pro- cesses must accommodate the will of the people and recognize its occasional wis? dom. As Thomas Jefferson observed. ?If we think [the people] not enlightened enoltgh to exercise their control with a wholesome discretion. the remedy is not to take it from them. but to inform their discretion." Interagency and International Coordination Up to this point [have been suggesting how risks should be assessed and man- aged in EPA. Much needs to be done to coordinate the various EPA programs to ensure a consistent approach. I have established a task force with that char? ter. I further believe we should make uni- form the way in which we manage risk across the federal regulatory agencies. The public interest is not served by two federal agencies taking diametrically op- posed positions on the health risks of a toxic substance and then arguing about it in the press. We should be able to coor- dinate our risk assessment procedures across all federal agencies. The risk man- agement strategies that ?ow front assessment may indeed di?er. depcn on each agency?s statutory mandat the judgment of the ultimate decis maker. . But even at the ??hnagement there is no reason the approat cannot be coordinated to achieve goal of risk avoidance or minimiza' with the least societ?fl?disruption po ble. 1 have been explo?g with the House and the Ot?Q-of Managem and Budget the po (ity of effect better intragovernme?l coordinatior the way in which we ?ess and man risk. A To push this one beli: it is in our nation's best interest to sh our knowledge of our appro; to managing them with the other det oped nations of the world. The enviri mental movement hasjaught us the terdependence of the'__World's ecos terns. ln coping with tb legitimate Ct cems raised by envir?mentalists. must not forget that ?cope in a W01 with interdependent denomies. If 0 approach to the management of risk not suf?ciently in hamlhiy with those the other developed Mons. we can save our health and rishLiour economy do not believe We need to abandon ther. but to ensure thatjh does not ha pen. we need to to sha scienti?c data and un??stand how harmonize our manage?hnt techniqu: with those of our sister-thations. In sum. my goal is a Memment-wit process for assessing an! managing env ronmental risks. this will tak cooperation and goodwill within EPA among Executive an?l'agencies, an between Congress and the Administrr tion. a state of affairs that?may partake the miraculous. Still. Ml worth trying and the effort is worth wholehearte support of the scienti?i?community. believe such an effort ?nches on th' maintenance of our cM??tt society. it which a democratic poll is grounded it a high-technology indu??l civilization Without a much more Messful way 0 handling the risks assodhted with ?it creations of science. I we will have set up for ourselves a grl?'jand unneces- sary choice between tllit fruits of ad- vanced technology and blessings of democracy. it mCE-t. VOL. 221 all 'I?uhi tit . Colim. 'l'mml ili' Fitoei'tuitit'u. .I .tt't'o tl?i??hl. 26. Pi. N. Collins. J. Aerii Rm. M91 37. F. U. Brieger. Art. hit . Super. am. . di- Uttii. 2. ll?N-ll. Hm- ,etitit?iit S. 659 P. C. But. "in Lt'lt? ?on (hair. I2. 33 ll?M?l. 29. H. Wilkes. that. 22. 297 ?9701. (ia- linnl. Mon. Aura (ftp. Stir. Hall 535 M). Weatherwas. Hu- or Ilu- .Umn- Plum (Univ. of Chicago Press. Chicago. H. D. W. (jttlinat and W. C. .Uiim' (Erin-t (imp. Ni?ii'il. 46. It? INTEL .12. K. V. Amm. Rt-t. dittlit'opoi. 2. I'll tl97ll. H. H. Schall'ner. Hot. (in: t('lmoeui 84. 4-10 ?9271: mid. 90. 279 HTML Hull. Turret But (hill 62. .187 ?9151; J. Heslop-Harrison. . Sm: London I71. ?Will. .14. H. Sitntiieshi-r. Alon]. Win. Wit-ti Moth: .Vittitrti?iss. Kl. Abl. I I tl'iltli. . 7.. AhthimmMontgomery. Pop .?ut ?on, Ml. t 9l fil. U. W. A. Uri-him ?on 5. 44. "1 11895). and A Naylol. \m I Hot lit W. Asclierson. Stummihu. lint. lt'rtmi I'tui Ht'ritiilr-ithtiije In 40. J. H. ls'eniptoii. Hit-ml I4. 34.? ii Montgomery. I'up \ii lion 79. Mb I WI 4: . Hie tops Nest Yotls. IVIIL ti. M. East. Pop hi ?on ill. :35 tl?Jllt. l-s' Biol. Ventinllil 30. NJ: (i Collins. J. ?null ml, S: t. Z. I WIN. 4-3 and {i Recu?s. It I A L'l'it. hip. Slit. Hull 5 .74 if. Anderson. Amt No. [for (ion! ?ll. ?25 and u. I. llroun. Hair! 35. 321 4ft Vi' ?In l'iitllI Hurt [Hit I7. ll?dit?il, Urns 42in Slit "it? Weattieruas. Hull. lam-i Hot lid: 45. lost tl?Hl'll. ?it? Vol Hi. 43 I- Randolph, lint. Ml. tl?J'r'hI. to ("out iittil ('mu (i lid l?icss_ Yoik Vii-5 . lh:J ('mitiill But [oh l'im Pu l. iltwli l' and Reese-s. \m 47. 235 H. Wilkes. 'l'i'mt'titt'. The ('losi?it of Main- tllussey Institute. Harvard University. ('ambrid e. Mass.. 1967!. SI. M. Lene an. The Reader tC?hirueoi. In Septem- ber I975. 53. W. (third. For. 9. 522 HIP-lot. A. Evolution 5. 1195?. ?4 In With. Montgomery Um derived maize and teosinte from a perfect?[loomed common ances- tor, In l9ll t-H. p. 3471. he suggested "that the ear ?its a development from the central spike of the tassel borne on a lateral branch of the Innate?! plant. the other branches of this tassel becoming abortive." In l9l3. he ?nally derived maize from teosinte [42. . Ill): ?in this evolu- tion the central spike olPthe lteosintel tassel developed into an ear [of matrcl" and went on to state that. in icosinte. "the terminal tassel- like structure . . . borne in a leafattil. surround- ed by :i kind of husk as is an ear of maize. and [bearingl only pistillate ?owers . . . is only a step in the production of an ear of maize. from teosinte. by a development of the central spike of the lateral [teosintel tassel into [a maize] ear." Lacking the clarity needed to convince his peers. Montgomery's I913 text has rarely been cited um and never evaluated. While Wilkes t5t)t lists the I920 revised edition of The (?uni from in his bibliography in the quoted passages. is identical to the edition l-?Ull. neither Montgomery nor his ideas are discussed. ?5 Torres 8. tor (i Barustat tlft'iiuit I. 3 tl?Hl'lt. in Maiigclsilorl'. Alli. (ii-net. l. S, Rogers. Ut?tti'tn 35. HI 5X. 1.. I: Randolph. ?our (imp. Nt'ii?tl. 46. it i Iti72t ?4 I) it langhani. (it'iuvti. 25. till Hi (It all postulated genes. only for alleles at the dominant in locus Ipodcorii. an ata- sistic abnormality ot no signi?cance to this controversyi is there evidence of monogenic Inheritance. til t'hailtilthyan .md N. Khryanin. in Plant 'itihstimi l: Skiing. lid tSpiinger-Verlag. Berlin. p. Ml. til l) Riches and Sprague. .?lttl. Nut Mi. Wilt (ii le?hc?t'scn. Aim Uri Hut Until 47. 24? ll?ilHll. 5. ll Rood. P. Harris. I) Minot. I?him? Phi sit-l hl. 79? I K. s- lsll. Dull Hut sit. I75. 3.thl%7i Skiing. f'i'uitt (itoii'tt'i Milhlom t?s llerltn. Wilt?. n5 1 l- Doehlet. personal communication 66 ('rouding oi .innual icostnte results In the dele- tion of the primary branch especially the loner and middle ones. to prostdc unbranclicd plants. Such deletion is ati either-or gl. Law and Science Policy in Federal Regulation of Formaldehyde Nicholas A. Ashfot?d. C. William Ryan. Charles C. Caldart Formaldehyde. one of the more widely used chemicals in modern industry. has recently become one of the more contro- versial as well. A plethora of lawsuits. congressional hearings. and scholarly analyses have centered on formalde- hyde. and more particularly on federal agency responses to new data indicating that it may he a carcinogen. These devel- 394 opments were sparked by an October l979 report from the Chemical industry Institute of Toxicology that form- aldehyde causes cancer in rats. 0n receiving the Cll'l? ?ndings. the Environmental Protection Agency the Occupational Safety and Health Administration (OSHA). the Consumer Product Safety Commission tliieslioiil phenomenon quite ilillcicnl ic diiction in branch internode length tcondensa- tiont discussed here. Despite claims to the con- trary t2]. .llt. neither teosinte ear clustering nor primary branch deletion acre involved in the early evolution of maize, h7. in mid ('mii humour- mm?. (L l- Sprayiue. Ed. tAcademic Press. New York. 89 Ml .l I Harper. Population Biology oi theademic I'ress. London. Mt. lohanneson. personal uimmitnicattoii I?im Am. Phillis .?itii I02. 454 1 Will]. Ii. Mayt. Populations. ?lm-it's. and Evolution Illelknap. ('amhridge. Mass . l9?ltti. 72. H. Waddington. 7hr .Slritti'et' o! the (hairs tAllcn Unit-in. London. I957). 73. Stehhins. Jr . Evolution in Plano tColttnibia Univ. Press. New York. Willi. . 74. J. l- IJoehley. M. M. Goodman. W. Stubet. .Snt. in press. 75. King and A. Wilson. Sunni- Ill? {l?il'l'SL 76. See K. Schumann. tn Faint-[trill Mr As: luv-- suit. I. Urban and P. (iraebner. Eds. {Born- traeger. Leipzig. I904). p. In: P. and F. Randolph. in (?um and (on: hupmre- im-m. F. Sprague. Ed. tAcailemic Press. New York. p. Ill: (3. N. Collins. Bull Tom-y Bot. ('litli 57. [99 (With; H, ('ullei'. Hut. Mm. 12. 357 tl946]. 77. H. H. Iltis. Ami. Mo. [tut third. 44. 7'7 7 . E. Ammo! .?ipi'i tt'\ mid Evolution [Bells- nap. Cambridge. Mass. I?i?oh. 79. With their often cone-like gynoccia. early angio- spermous [lowers could well lime evolved from loosely structured branch systems adapted to it'ind pollination sshich telescoped into the tight- ly focused reproductive efforts that animal-polli- nated ?owers represent and demand. J. Gould. The ?moth tNortoti. Nets York. 193m. til. W. H. Wagner. personal communication 82 leis-e special thanks to (J. W. Beadle for inviting me to participate in the I9TI NSF-supported MesIean "teostme mutation hunt"; J. F. Duch- ley and T. H. Allen for both enthusiastic criticism and support; J. H. Lonnquist for grow ing the indispensable mam: teosinte I'l'lhl'ltls. ll. Benz. 1. Densloss. [)tivtck. W. (ialinat. T. R. Soderstroni. M. Waller. H. (i Wilkes. and 5 Wright for helpful comments; Taylor for uid with illustrations; and especial- Iy A. for help ssith the manu- script. Supported by? NSF grants HMS 74-21th and [Mil-I tilt-12772: Pioneer Hi-bred Internation- al. Inch. Johnston. IowaAllen llerbaritini Fund. University of Wis- eonsin-Miidison. 8i . and other agencies undertook severaljoint actions. the most important of which was to form the Federal Panel on Formaldehyde. The panel was com- posed of top scientists from the federal government and was directed to evaluate all available information on the long- term ell?ects ot'exposure to formaldehyde and to assess the human health risks. In November I980 the panel pesented its report to the agencies. Based on its re- view of the available data. the panel concluded that ?formaldehyde should be presumed to pose a carcinogenic risk to humans? ti). Thereafter. CPSC issued a ban against the use of urea?formaldehyde foam insulation. EPA and OSHA. how- ever. declined to take regulatory action against formaldehyde. Nicholas A. Ashl?ord is assocutle professor of technology and policy and director ofthe Center for Policy Alternatives. Massachusetts Institute of technology. Cambridge 02139. William Ryan is on the research staff and ('harles Caldart is a stall" attorney at the t'enter for l?ultt;\ Alternatives. 0 222 Framework for Evaluat? Formaldehyde Decis'i The greahight I, I icial authority suggests that the appropriate legal stan- dard by which to evaluate agency deci- sions is whether the agency engaged in ?reasoned decision-making? t2). As re- cently articulated by the DC. Circuit Court (3). an agency practices reasoned decision-making when it (il takes a ?hard look . . at the relevant issues." deliberates "in a manner calculated to negate the danger of arbitrariness and irrationality." violates ?no law." and (iv) provides an ?articulated justi?- cation? that makes a "rational connec- tion between the facts found and the choice made." Applying the concepts of reasoned de- cision-making as an analytical tool re- quires a clear understanding of what a particular agency has and has not done. With health risk determinations. this un- derstanding often requires a technical knowledge of the underlying data and methodologies. Further. it requires an ability to distinguish between purely technical determinations and those based on the more subjective. science policy determinations. The term "science policy? denotes issues that are grounded in scienti?c analysis but for which technical data are insuf?cient to support an unequivocal scienti?c conclusion. The ultimate reso- lution of these issues depends on deter- minations of social policy. Distinguish- ing science policy determinations from those of a truly technical'nature is a central step in evaluating the adequacy of an agency's assessment of human health risks. Simply deferring to agency expertise on all determinations that ap- pear to be ?scienti?c" overlooks the subjective determinations at the heart of the agency's decisions. Such an ap- proach frustrates any e?'ort to measure agency decisions against the standard of reasoned decision-making. Acknowledging that science policy will often play a major role in agency assessments of human health risks. it must then be determined whether an agency has abided by the principles of reasoned decision-making. 'l?hese princi- ples impose three primary responsibil- ities on an agency assessing health risks: til it must adequately evaluate the tech- nical data. it must follow proper administrative procedures. and it must correctly carry out its statutory mandate. In practice. these functions overlap. An agency's interpretation ofits statutory mandate. for example. can in- ?uence both the nature of the technical 35 NOVEMBER I'Jtl.? data it examines and the manner in which it makes that examination. When completed agency decisions are ana- lyzed. however. the three elements are separable and provide a logical frame- work. Treatment of dam. In evalu- ating the technical data relevant to deter- mining health risk. an agency must delve deeply into scienti?c issues. The agency will ordinarily engage in two levels of scienti?c analysis. On one level. it will address "hard" scienti?c issues that can be resolved with currently available tries and interested members of the gen- eral public. As a matter of administrative procedure. an agency must adhere to its policy guidelines or identify and explain any change in. or departure from. those guidelines. To develop policy guidelines in the area of health risk assessment. the agency must adopt positions on science policy issues. In the absence of formal announcements ofchanges in these posi- tions. recognizing policy departures will require that one ?rst identify and under- stand the underlying science policy is- sues. Summary. An examination of the way in which the Environmental Protection Agency. Occupational Safety and Health Administration. and Consumer Product Safety Commission each responded to evidence of lormaldehyde's carcinogenicity in animal systems reveals the interplay between politics and science policy in regulatory determinations. In some cases there were significant and unjustified departures from reasoned decision-making. Agency decisions not to take action deserve special attention by citizens. the Congress. and the judiciary to ensure that federal regulatory agencies take the necessary steps to protect the public from significant health. safety. and environmental risks. methodologies. On a second level. the agency will confront various science pol- icy issues that cannot be answered solely on a technical basis. Thus a meaningful critique of an agency?s treatment oftech- nical data requires both an understand- ing of the relevant technical methodolo- gy and an ability to distinguish between ?hard? science and science policy deter- minations. In appropriate circumstances. of course. the agency may depart from sci- enti?c opinion on science policy issues. These issues do. after all. involve policy determinations. and accordingly should be made by the governmental entity charged with re?ecting the will of the people through the execution of a con- gressional mandate. Nonetheless. they are also determinations that should be properly based on a sul?cient under- standing of the underlying scienti?c evi- dence. When a majority position on a science policy issue has evolved within the scienti?c community. we believe the agency should not depart from that posi- tion without acknowledging and justify- ing the departure. to proccituml rcquirru meats. Science policy issues may also arise in the context of procedural mat- ters. Agencies often develop general pol- icy guidelines for their regulatory ac- tions. in the form of either formal generic standards. such as OSHA's. or informal statements of procedure. such as the Regulatory Council's. These guidelines not only promote regulatory continuity but also provide notice to affected indus- of statutory mandates. Fi- nally. the agency must act in accordance with its statutory mandate. This respon- sibility has two elements. The agency must carry out the speci?c duties of the particular statutory provisions under which it is considering regulatory action. At the same time. it must faithfully ad- here to the more general aspects of the congressional mandate underlying its en- abling legislation. Any analysis of agency decision-mak- ing must carefully consider both of these elements. In evaluating health risk deter- minations. particular attention must be given to the ways in which assessments of science policy issues re?ect an agen- cy?s interpretation of its statutory man- date. in close cases. for example. should the agency tip the balance in favor of ?nding a human health risk or in favor of deferring such a ?nding until additional data are available?? In developing a general analytical framework. we are not unmindful of the judicial deference traditionally afforded agency decisions not to act. In the past. legal challenges to agency decisions on health and safety have come primarily in response to speci?c regulatory actions. An agency's implementation of a stain- tory provision was challenged as either too zealous or insul?ciently protective. In the present antiregulatory climate. challenges to agency decisions not to act may assume greater signi?cance. Judicial deference to agency discretion in such situations is based largely on respect for agency expertise in matters 895 of resource allocation and eval- uation. Such deference is tsplaced. however. where the cloak of expertise serves to disguise inadequate technical analysis. improper decision-making pro- cedures. or statutory misinterpretation. An analysis of the formaldehyde deci~ sions demonstrates why. before defer- ring to an agency's decision not to take regulatory action to protect human health. the courts should ?rst study the agency's decision to determine whether such deference is. in fact. warranted. EPA's Decision Not to Designate Formaldehyde a Section 4m Chemical In March I981 deputy assistant administrator for toxic substances. in consultation with other EPA ol?cials. concluded that the agency was obligated to begin consideration of formaldehyde under section 4(0 of the Toxic Sub- stances Control Act. Section 4tf} pro- vides that once EPA receives informa- tion indicating that there ?may be a reasonable basis to conclude? that a chemical poses a signi?cant cancer risk. the agency has I80 days to either ?initi- ate appropriate regulatory action" or publish an explanation of why the risk ?is not unreasonable" t4). When newly con?rmed EPA adminis- trator Anne (jorsUch assumed of?ce in May I98l she delayed action on formal- dehyde pending additional review. Final. ly. on it) February 1982. EPA released a memorandum. written by new assistant administrator .lohn Todhunter. that ana? lyzed the available evidence on formal~ dehyde carcinogenicity and exposure and concluded that section 4th had not been triggered. In a sense. any discussion of decision-making process may be super? ?uous. Considerable evidence suggests that the incoming EPA ollicials had de- termined their policy on formaldehyde long before any "decision-making pro- cess" had been completed (5. Assum- ing. however. that the 'I'odhunter memo- randum does represent the culmination of a decision-making process. this process was nonetheless ?awed in numerous respects. Analysis ol'tt'rlinir'ul data. An agency decision must evidence a "rational con- nection between the facts found and the choice made" Health risk assess- ments thus require a careful analysis of the relevant technical data regarding both a substancc's toxicity and the ex~ tent of human exposure. Unquestion- ably. an agency is properly accorded some discretion in its treatment of tech- 896 nical data. Nonetheless. the agency's analysis must be free of overt errors in technical methodology or reasoning. treatment of the formaldehyde data aroused signi?cant criticism. A re- view of the agency's technical analysis reveals several examples ol'questionable scienti?c reasoning. The 'fodhunter memorandum appears to contain signi?cant lapses in hard sci- ence. In many cases it seems simply to ignore empirical data contrary to EPA's conclusions. or to rely on controversial factual assumptions without offering evi- dence in support of these assumptions. As has been discussed elsewhere (7). the questionable aspects of 'fodhunter's technical analysis include his reliance on methodologically inadequate epidemio- logic studies. his presumption of site speci?city and species speci?city for formaldehyde carcinogenicity. his argu- ment that the irritant properties ofform- aldehyde are the basis of its carcinoge- nicity. his assumption that humans will avoid formaldehyde exposures above 2 ppm. and his concIUsion that formalde- hyde exposure levels in homes with urea-formaldehyde foam insulation are no higher than those in other homes. In his reliance on the epidemiologic studies and his assumption of site and species specificity. 'I'odhunter also de- parted from the science policy conclu- sions of a scienti?c panel convened by the Interagency Regulatory Liaison Group thLCi}. Indeed. EPA's formalde- hyde deliberations reflected science poli- cy positions that represent the views of a minority in the scienti?c community (7). While these positions may not be "wrong" in a purely technical sense. they demand justi?cation. EPA neither acknowledged the need for such justi?- cation nor supplied any. Prim-dam! tts?pt'r'ts, The principle of reasoned decision-making further re- quires that an agency deliberate "in a manner calculated to negate the dangers of arbitrariness and irrationality" Lil. Depending on the particular agency ac- tion under review. courts have enumer- ated various speci?c procedural require- ments for reasoned decision-making. EPA's treatment of formaldehyde raises procedural questions. Several of these questions concern a possible bias of EPA with respect to industry. With the advent ofthe (jorsuch administration. EPA embarked on a de- cision-making process for formaldehyde that served to maximize input from the formaldehyde industry and to minimize input from other sectors. Regardless of the ultimate substantive validity of the agency?s section 4tl't determination. the appearance of its decision-making pro- cess lends credence to claims of impro- priety. The EPA held three meetings with representatives of the formaldehyde in- dustry in summer The initial impe- for the meetings apparently came from industry. Although EPA has subse- quently characterized these meetings as having been ?exclusively scienti?c in nature." the rosters of those present reveal that the sessions were dominated by the perspective of the formaldehyde industry. With the exception of one or two "neutral" scientists at each meet- ing. the Formaldehyde Institute selected all the non-EPA participants. Rather than solicit a variety of view- points. EPA closed the meetings to the public. Conspicuous by their absence were scientists representing groups that might be expected to oppose the formal- dehyde industry's position- Also absent were representatives of other regulatory agencies and the IRLG. Indeed. the agency reportedly refused to admit two scientists who requested permission to attend: Andrew Ulsamer of CPSC and Han Kang of OSHA. both members of the formaldehyde group. As other observers have suggested. these meetings may also have violated the Federal Advisory Committee Act That statute recognizes that agencies have come to rely on technical "advisory committees" as an aid to deci- sion-making. and establishes speci?c procedural requirements for such com- mittees. The Act requires that advisory committees be formally chartered. be composed of a "fair balance" among opposing viewpoints. give notice ofthcit? meetings and open them to the public. and maintain and detailed min- utes of those meetings. The deputy administrator of EPA. John Hernandez. has written that the primary function of those sessions was to allow him to meet with "scienti?c and technical experts" to ?discuss the scien- ti?c merits of the available information.? The meetings. be indicated. "were de- signed . . . to explore fully the scienti?c and technical issues." He later testified that the purpose of the meetings was to ?get all the scienti?c information per- taining to the exposure and toxicity of these substances in the open." lfthis characterization is accurate. the meet? ings would probably fall within purview. EPA departed from the Act?s procedural requirements in numerous particulars. Certainly. the narrow range of viewpoints represented at the industry meetings is inconsistent with the ?fair balance" requirement of the Act. SCIENCE. VOL. 222 substantial amount of?cgence indi- cates that before releasing 10 Febru- ary memorandum. Todhunter met on several occasions with John Byington. attorney for the Formaldehyde Institute. and Len Guarraia. then a director of the American industrial Health Council and director for government relations for the Organic Chemical Manufac- turer?s Association. Although the pre- cise nature and scope of these gatherings are dif?cult to deduce. it seems that Todhunter did meet with formaldehyde interests before drafting position paper on the section an determination. As the DC. Circuit Court has noted, ?[tlhe inconsistency ofthe ex parte con- tacts with reasoned decision-making and fairness to the public has been increas- ingly recognized in recent years" (9). Shortly after they took of?ce the new EPA administrators ceased to cooperate with the other federal agencies that were assessing formaldehyde carcinogenicity. EPA also isolated itself from its own science advisory board. On 29 October the board's executive committee recommended that EPA submit the formaldehyde issue to the National Academy of Sciences before it conclud- ed its section 4th determination. EPA instead permitted Todhunter to draft his technical memorandum on formaldehyde without such assistance. Other procedural questions concern the explanation given by EPA for its decision on formaldehyde. The courts have long required agencies to ?articu- late with reasonable clarity their reasons fora decision" In the Words of an oft-cited opinion ?the orderly func- tioning of the process of review requires that the grounds upon which the admin- istrative agency acted be clearly dis- closed and adequately sustained." formaldehyde deliberations fell far short of this standard. The most troublesome procedural problem lies in Todhunter?s failure to acknowledge his departure from prior agency positions on many of the science policy issues involved. Where an agency has changed a previously articulated pol- icy or departed from a relevant agency precedent. the courts have required the agency to provide a detailed rationale. Although EPA never promulgated a for- mal cancer policy. it published informal cancer guidelines in I976. endorsed the risk assessment document in I979. and participated in the Regulatory Coun- cil's September I979 policy statement on regulation of chemical carcinogens (H). Being most recent. the Regulatory Coun- cil?s statement was the logical founda- tion for EPA's approach to formalde- 25 NOVEMBER I981 hyde. yet 'l'odhunter?s positions dill'er from the council's in several areas. The guidelines specify that negative epidemiologic studies will not be pre- sumed to indicate that a substance is not carcinogenic. (ii) sites exposed by routes other than those tested will be presumed to be at risk. negative bioassay re- sults for some animal species. even in well-conducted tests. will not be said to detract from well-established positive evidence for other species. and (iv) a no- e?ect threshold level will not be assumed to exist for carcinogenic substances. Todhunter adopted a contrary position on each of these points. Todhunter also suggested that positive data on more than one species at more than one dose level should be a prerequi- site to a determination of human risk. The Regulatory Council?s guidelines re- quire only positive data in a single spe? cies at one dose level. Similarly. Tod- hunter discounted ?ndings of benign tu- mors in bioassay data and considered only veri?ably malignant tumors. while the council concluded that benign tu- mors should be considered evidence of potential malignancy. The council state- ment also indicates that agencies will attempt to estimate the maximum risk that could reasonably be expected. Tod? hunter consistently assumed policy posi- tions that minimized estimated risks. In its failure to explain or even acknowl- edge these policy reversals. EPA fell shon of its procedural responsibility. Finally. 'l'odhunter's memorandum was not reviewed by his scienti?c peers inside or outside the agency. The failure to garner peer review. especially on mat- ters so controversial. run counter to the professed goal ofthe new EPA adminis- trators to improve the scienti?c basis of the agency's regulatory decisions. It may also have violated internal EPA proce- dures. In January I982 EPA implement- ed a new internal policy governing the review of scienti?c. informational. and educational materials. The policy. which applies to ?any material prepared for distribution to anyone outside the agen- cy." requires that at least two specialists review all such materials HE). Nonethe- less. the Todhunter memorandum was released to the public without prior peer review. mandate. Congress de? signed section 4tfl as a mechanism for early identi?cation and regulation of those chemicals that were of particular concern becaUse they are likely carcino- gens. mutagens. or teratogens. EPA's current interpretation of that section. however. will frustrate this scheme. As noted. section 4th requires agency action if there "may be a reasonable basis" to conclude that a risk of harm exists. ln both common usage and judi? cial interpretation. "may" indicates the possibility of occurrence. Under the plain language of section then. EPA cannot delay its threshold determination until a risk has become certain or proba- ble. but rather must take action on learn- ing of a credible possibility of such risk. Todhunter?s memorandum is particu- larly noteworthy in this respect. in sum- marizing the formaldehyde data. Tod- hunter noted that "there may be human exposure situations . . . which may not present carcinogenic risk which is of signi?cance.? He thus stated the re- quired statutory ?nding in the negative. The logical converse of this statement? that there may be human exposure situa- tions that do present signi?cant carcino- genic risk?is precisely the ?nding that requires EPA to proceed under section an. The agency's failure to do so re? ?ects misinterpretation of statutory lan- guage. The agency's assessment of the kind of risk that it is to consider under section an may also be inaccurate. Once again. the statute itself provides relatively clear guidelines. Congress dealt with both short-term and long?term risk in 4m. which addresses chemical substances that either ?present" or ?will present" a signi?cant risk of harm. The Toxic Sub- stances Control Act does not de?ne the phrase "significant risk." but the con- text suggests that "signi?cance" per- tains to the likelihood of occurrence. By providing that the risk that may exist must be signi?cant. the Act seems to require only the possibility ofa probable occurrence. Evidence indicating the pos? sibility of a signi?cant risk thus triggers the threshold determination that compels EPA to assess the risk more precisely. in its risk assessment. the agency must consider both "serious" and "wide- spread" harm. By speci?cally distin- guishing between these two categories of harm in section 4tf). Congress clearly indicated that either one will trigger a threshold determination. One element focuses on the extent to which the chem- ical may pose a risk of serious harm. Here the concern is not so much the number of people who may be affected. but how severely they may be a?'ected. A low incidence ofa debilitating cancer. then. would suf?ce. The other element is the extent to which the chemical may pose a risk of widespread harm. Here a higher incidence is required. but the harm need not be as severe. After making a threshold determina- tion of possible signi?cant risk. EPA a 897 mub decide. within a pre?r' -d time period. whether regulatory-a is ap- propriate. If the agency determines that the risk is not unreasonable. it must subject this ?nding to public scrutiny by publishing it in the Federal Register. If. on the other hand. the agency does not conclude that the potential risk is not unreasonable. it most "initiate appropri- ate action . . . to prevent or reduce to a sufficient extent such risk.? While we express no opinion here as to the appro? priate regulatory response to formalde- hyde under the Toxic Substances Con- trol Act. it appears that section 40'] re- quires something more of EPA than the agency?s actions to date. Decision Not to Issue an Emergency Temporary Standard After receiving the results of the pre- liminary CIIT bioassay in late I979. both OSHA and the National Institute for Occupational Safety and Health began preparing ajoint current intelligence bul- letin (CIBI on formaldehyde. A pre-pub- lication version was made available to the public in December I980. The CIB recommended that "formaldehyde be handled as a potential occupational car- cinogen and that appropriate controls be used to reduce worker exposure." In March 98 Thorne Auchter as- sumed of?ce as assistant secretary of labor for OSHA. Soon thereafter he withdrew sponsorship of the C18. In October I981 the United Auto Workers and other major labor unions petitioned OSHA to set an emergency temporary standard (ETS) for formalde- hyde under section 6th of the Occupa- tional Safety and Health Act (13). Sec- tion 6th speci?es that OSHA shall promulgate an ETS if it determines that employees are exposed to "grave dan- ger" from exposure to a hazard and that an ETS is ?necessary to protect employ- ees from such danger.? In a letter dated 29 January I982. Auchter denied the petition. He stated that OSHA assess- ments indicated that risks at the current permissible exposure level of 3 were not suf?cient to warrant a ?nding of grave danger. and that current employee exposure levels were below 3 ppm. Auchter's decision to deny the union?s ETS petition was apparently based on two evaluations performed by agency personnel after Auchter took of?ce. The ?rst is a review of a formaldehyde risk assessment prepared by scientists at the Massachusetts Institute of Technology. and the second. the agency's preliminary H98 risk assessment. Positions expressed in each ofthese documents found their way into Auchter?s ultimate statement of ra- tionale. although the extent to which they contributed to the ETS decision is not altogether clear. In many ways. OSHA's deliberations on formaldehyde mirrored EPA's. Al- though departures from sound technical reasoning and established ad? ministrative procedure were perhaps less troublesome than EPA's. they were nonetheless signi?cant. OSHA's treat? ment of science policy issues provides an excellent example of how such issues can cut across all three levels of an agency?s administrative responsibility. Analysis oftec'lmical data. The review of the MIT study inappropriately relies on the formaldehyde epidemiologic stud- ies. on arguments of species speci?city. and on arguments of minimum exposure. The review also assumes that workers are exposed only to low levels of formal- dehyde. avoiding exposures above 3 because of their irritating effects. Like the Todhunter memorandum. the OSHA review cites no empirical evi- dence for this assumption and apparently ignores data on exposures above the 3 level. Furthermore. the review departs from prevailing scienti?c opinion on science policy. As previously noted. the review presumes species speci?city and relies on negative epidemiologic data. In an even more striking departure. the review does not merely question the way results of animal bioassays are extrapolated to humans. but rather argues that such ex- trapolation is meaningless: ?Because of the vast uncertainties in extrapolating from experimental rodent studies to man. such experiments do not and can- not predict or measure human risks." At best. this sweeping denunciation of ac- cepted science policy represents a con- troversial minority opinion. Procedural aspects. Perhaps the most signi?cant procedural de?ciency in OSHA's deliberations was the agency's failure to adhere to its own policy on cancer risk assessment. Although that policy was promulgated as a formal agency regulation. which is still in effect. Auchter?s denial letter does not ac- knowledge it. Indeed. Auchter?s treat- ment of the rat data and the agency's failure to consider benign tumor data con?ict with the policy?s plain language. The OSHA review of the MIT report also departs signi?cantly from the agen- cy?s cancer policy. again without ac- knowledging or explaining the departure. Although the Auchter letter contains a statement of rationale. that statement does not identify the agency "risk as- sessments" on which it says it relies. Does it refer only to the OSHA assess- ment made after the ETS petition was ?led. or does it refer also to the agency's earlier review of the MIT study? Indeed. the OSHA assessment appears to haVe been prepared in written form sometime after Auchter?s letter was delivered to the unions. This fact. along with the conclusory nature of Auchter's analysis. calls into question the letter's adequacy as a "statement of reasons." The possi- bility of post hoc rationalization looms large here. A ?nal procedural problem with OSHA's formaldehyde deliberations is that the agency disregarded and mischar- acterized the advice of its own scientists. Although Auchter publicly stated that the agency withdrew its support of the formaldehyde CIB because it "lacked con?dence in the data" on which the C18 was predicated. all the technical personnel in carcinogenicity as? sessment group supported both the C18 and the underlying data. Later. when Peter lnfante. director of OSHA's Of?ce of Carcinogen Identi?cation and Classi?- cation. wrote to the International Agen- cy for Research on Cancer rec- ommending that formaldehyde be classi- ?ed as an animal carcinogen. the agency took steps to have him dismissed. The matter became the focus of a congres- sional hearing. Subsequently. the dis- missal proceedings were canceled. Statutory tituttdate. The issuance of an ETS for a workplace chemical under section 6(c) depends on a ?nding that ?employees are exposed to grave danger from exposure" to that chemical. A re- view of OSHA's formaldehyde decision indicates that the agency may have adopted a more limited interpretation of secti0n 6tc} than the statute will permit. Clearly. cancer is a "grave? illness. The question is what degree of cancer risk constitutes a "grave danger" under section fate). The Third Circuit Court provided guidance in a I973 opinion stating that "[thile the Act does not require an absolute certainty as to the deleterious effect of the substance on man. an emergency temporary standard must be supported by evidence that shows more than some possibility that a substance may cause cancer in man" litalics added]. A review of OSHA records shows that such evidence was available during the formaldehyde deliberations. Extrapolat- ing from the rat bioassay results. the agency?s own risk analysis indicates that four formaldehyde-related cancer deaths per exposed workers would SCIENCE. VOL. 337. expected at the cu permitted eXposure level of3 pp ause OSHA estimated that the averagg~ occupational mortality rate for manufacturing work- ers. from all reported occupationally re- lated causes. is also four per thousand. Auchter concluded that formaldehyde does not pose a grave risk of danger. This comparison misses the mark. The question is not how the risk from formal- dehyde compares with the aggregate of all other risks. but how many lives can be saved by regulating formaldehyde ex- posure. Furthermore. if the agency?s es- timate of average aggregate risk is valid. the fact that exposure to formaldehyde alone presents a risk of comparable mag- nitude should give rise to considerable concern. This evidence of carcinogenic potency appears suf?cient to warrant the issu- ance ofan assuming that a determi- nation of ?grave danger" under section 6(cl may be made by extrapolating from animal data. Dicta from the Third Circuit Court again provide substantial guidance "Extrapolation from animal experi- ments may in appropriate cases be used to establish a suf?cient probability of harm to man.? Moreover. the courts have indicated that evidence of animal carcinogenicity is by itself suf?cient to justify a permanent standard. A greater burden would hardly seem appropriate for a temporary standard. In addition to contravening its own cancer policy and the IRLG guidelines on this issue. OSHA contravened its section file) man- date as well. remaining inquiry is' whether the evidence of worker exposure to formal- dehyde is suf?cient to warrant issuing an In a recent decision involving eth- ylene oxide. the DC. Circuit Court de- clined to compel OSHA to issue an ETS where the evidence indicated that only ?some? workers are exposed to ethy ~ ene oxide at levels that present a ?signif- icant risk" of "grave danger.? On the basis of this finding. however. the court ordered OSHA to expedite ongoing pro- cedures to set a permanent standard to reduce worker exposure to ethylene ox- ide. The data on formaldehyde exposure appear to be stronger than the data on ethylene oxide. both in the detail and reliability of the exposure information and in the~number of workers exposed. Indeed. the DC. Circuit's approach in the case of ethylene oxide may also be appropriate for formaldehyde. An order to commence a procedure to set a perma- nent standard for formaldehyde would set the stage for an objective appraisal of the cancer risk and of the need for fur- ther worker protection. 25 NOVEMBER Decision to Ban Urea-Formaldehyde Foam Insulation The Consumer Product Safety Com- mission received the results of the pre- liminary CIIT bioassay in late 1979. By this time. the commission had already begun to study the health problems asso- ciated with the use of urea?formaldehyde foam insulation (UFFI). In March I980. when the need for further study became apparent. CPSC organized the Federal Panel on Formaldehyde. In June I980 CPSC proposed a rule requiring UFFI manufacturers to inform prospective buyers of UFFI health effects. The com- mission received the panel report in No- vember I980. and in February it proposed to ban UFFI altogether. Ulti- mately it promulgated a ?nal rule ban- ning UFFI as of August 982. On 12 April I982 the Formaldehyde Institute ?led a suit to challenge the CPSC ban. On It) August of that year a federal judge refused to issue a tempo- rary injunction. and the ban took e?'ecl on that date. The Fifth Circuit Court vacated the UFFI ban on 7 April I983 and has recently reaf?rmed its decision. Especially when viewed in contrast to the EPA and OSHA deliberations. formaldehyde deliberations might be considered a model of reasoned decision-making. The commission was subject to more stringent statutory re- quirements than were EPA and OSHA. However. it not only met these require- ments. but exceeded them. Analysis of'tct'hniml dam. During the notice and comment rule~making period for the UFFI ban. CPSC responded to many comments that speci?cally ques- tioned its technical risk analysis. A search of the commission's stated ratio- nale. background documents. and com- ment responses reveals no clear errors in scienti?c reasoning. The commission conformed to prevail- ing scienti?c opinion on science policy issues. For example. it extrapolated ani- mal results to humans and high-dose results to low doses. Procedural aspects. In promulgating its UFFI ban. CPSC was subject to the procedural provisions of the Consumer Product Safety Act. which require an opportunity for notice and comment. In addition. as a commission headed by a ?collegial body." CPSC must comply with advance notice and open meeting requirements of the government in the Sunshine Act. CPSC appears to have conformed to these requirements and to have comported with the principles of reasoned decision-making. It established and maintained a decision-making frame- work that allowed for input from all interested parties. The commission's UFFI science poli- cy decisions were consistent with the Regulatory Council's policy on the regu- lation of carcinogens. which CPSC ex- plicitly endorsed. The commission also provided a detailed statement of reasons for its decision to ban UFFI. That state- ment explains the basis of the ban. cites speci?c sources of data. and responds to numerous comments. The Fifth Circuit Court Opinion Overturning the UFFI Ban The Fifth Circuit Court does not share our sanguinity regarding the CPSC can- cer risk assessment. The court criticized two aspects of the commission?s risk assessment: the manner in which homes were selected for measurements of in- home formaldehyde levels and the use of the CIIT data on rats to project human carcinogenic risk. The court indicated that handling of either of these factors would be suf?cient to warrant reversal of the UFFI ban. We disagree. The CPSC based its estimate of likely formaldehyde exposure levels on ll64 measurements from homes insulated with UFFI and on laboratory tests on UFFI panels. 0f the in-home measure- ments. 827 were conducted in residences whose occupants had complained about UFFI-related health problems and 337 in homes selected for other reasons. In concluding that these in-home measure- ments were an improper basis for the commission?s risk assessment. the court points to two ?signi?cant omissions" The Commission does not explain its reliance on a data base comprised largely of complain- ant houses. Nor does the agency justify its failure to conduct a study of randomly select- ed UFFI homes before issuing the product ban. In truth. however. the agency did ex- plain its willingness to rely on the in- home data. According to the ?nal risk assessment. all the in-home measure- ments (for both "complaint" and ?non- complaint? homes) were grouped ac- cording to the time that had elapsed between the date that UFFI had been installed and_the date that the measure- ment had been taken. The groupings were by [0-week periods over a period of 9 years. Average measurements for com- plaint homes were compared with aver- age measurements for noncomplaint homes within each of these [0-week peri- ods. but no statistically signi?cant differ- ences were found. Thus. the agency con- 1 R99 clu . there is no reason ve that forma ehyde levels in mplaint homes were signi?cantly higher than those in other homes insulated with UFFI. While the statistical comparison employed by the agency does not rule out the possibility that use of complaint homes did in?uence the data to a certain degree. the data indicate that the effect of any such influence on the ultimate cancer risk projection would be relative- ly small. The failure to Use a randomized sam- ple is perhaps a closer question. Al- though selecting a study population through random sampling would be sci- enti?cally preferred. failing to do so does not necessarily vitiate the value of a study. The obvious source of potential bias in the CPSC data?and the only one cited by the court?is the possibility that formaldehyde levels were appreciably higher in the complaint homes than in other homes. As noted. however. com- parisons between complaint and non- complaint homes revealed no statistical- ly signi?cant differences in formalde- hyde levels. There remains a possibility that the lack of randomization allowed some oth- er source of bias to in?uence the results. As the in-home measurements were largely consistent with the results ofthe laboratory tests. this possibility seems slight. Nonetheless. some uncertainty remains. In choosing to take regulatory action in the face of this uncertainty. CPSC implicitly made a policy determi- nation that the potential risk to human health from continued use of UFFI insu- lation did not permit it to delay action until a large randomized study of UFFI homes could be completed. In vacating the ban. the court has substituted its own policy judgment for that of the agency. As a matter of administrative law. a court may reverse an agency?s policy determination when that determi- nation con?icts with the statutory man- date. It may also remand the issue to the agency for reconsideration ifit ?nds pro- cedural irregularity. It may not. howev- er. merely replace the agency?s policy with its own. Does the Consumer Product Safety Act require the commission to conduct a controlled test from a randomly selected study population before imposing a prod- uct ban'.? Clearly the statute itself con- tains no explicit direction in this regard. Rather. both the statutory language and the legislative history indicate that c0n- siderable discretion is to be afforded the commission in its choice of study de- signs. so long as it bases its conclusions on generally reliable data. Declaring that ?it is not good science to rely on a single experiment." the court also found the commission's ?ex- clusive reliance" on the CIIT rat bioas- say in its projection of human cancer risk to be ?unsupportable? U4). The extrap- olation of animal data to predict human cancer risk is ultimately an issue of poli- cy: ?good science" is simply unable to provide a precise calculation of formal- dehyde?s carcinogenicity in humans. Consistent with the federal cancer policy set forth in the Regulatory Council state- ment. CPSC based its projections on a single. well-conducted animal bioassay. Although the court makes no mention of the Regulatory Council statement in its review of the commission's action. its rejection of that action is an implicit repudiation of carefully established fed? eral administrative policy. All indica- tions are that the court did not base this position on its reading of the commis- sion?s statutory mandate. but rather on its own understanding of scienti?c meth? odology. As such. the court has con- fused science with science policy. and has once again substituted its policy judgment for that of the agency. In sum. though we must emphasize that we express no opinion as to the other aspects of the commission's deci- sion. nor of the court's review thereof. we ?nd the Fifth Circuit's analysis to be unpersuasive in its evaluation of cancer risk assessment for formalde- hyde. Conclusion Reasoned decision-making has evolved as a common standard forjudi- cial review of agency action. and is a particularly appropriate criterion by which to evaluate the conduct of agen- cies responsible for protecting public health by regulating exposure to toxic substances. Although it was initially ap- plied to decisions to take regulatory ac~ tion. this standard has been increasingly applied to decisions not to act as well. This is a welcome development. In these antiregulatory times. decisions not to act are becoming more numerous. and ade- quate review is needed to ensure that the agencies adhere to their statutory man- dates. This review of the OSHA and EPA actions demonstrates the need to exam- ine scienti?c determinations carefully. lest social policy decisions be hidden in alleged assessments of technical or sci- enti?c fact. Finally. the formaldehyde case raises important questions about the proper status for an environmental agency. Of the three agencies examined here. only the one structured as a commission acted responsibly. Recent concern over the political manipulation of environmental agencies has prompted proposals to con- vert EPA into a "hybrid commission." Although structural changes can help insulate an agency from political in?u- ence. continued judicial and congres- sional scrutiny of Executive Branch agencies may?at least for the short term?provide the most practical check on agency impropriety. The courts and Congress thus must take a ?hard look" to ensure that agencies exercise rea- soned decision-making in their approach to toxic substance control. References and Notes I. Federal Panel on Formaldehyde. Environ. Health Perrpt't't. 43. 139 [1983]. . Crenter Boston Television Corp. v. Federal Trude ('mnmisritm. 444 Fed. Rep. 2nd set". 852 Circuit Court. I970). ccrtiorari dc. nied. 403 US. 923 See also: Motor J'tvlumttut'turerx Association v. State Farm Mutual Automobile insurance Co.. U.S. Luu' Week 51. 4953 and 4960 [1983). where the Supreme Court held that an agency must per- form a ?reasoned analysis." . W.W.H.T.. inc. v. Federal ('mmnunii?utions Commission. 656 Fed. Rep. 2nd scr. 807 and 817 Circuit Court. 1981!. . IE Code. ?2603tft tl97fit. . Todhuntcr himself has testi?ed that when he arrived at EPA in July he was informed that the agency would take no regulatory action on formaldehyde. 6. N. A. Ashfortl. W. Ryan. C. C. Caldart. Hurt. Environ. Rt?l?. 7. 197 I 7. . pp. 339-333; LES. Congress. Committee on Science and Technology. Review of the Scientific Busts of the Enrirotuni-ntal Protection Agency's ('urt?inot'rnit' Risk Asst-5s- men! {98th Congress. Ist Ses- sion. 1933]: F. and Pctilo. Swimm- 216. I2851I982). 5 U.S. Code. appendts I tl976t. . United States Limo. [or v. Federal Muritimt' Commission. 584 Fed. Rep. 2nd ser. Sl?) and 540. note it Circuit Court. 1978!. Ill St't'uritit's and lit'rlmnec v. er)? Corp. 3l8 LLS. ?ll and 94 I II. Regulatory Council. l?mi Rt?t?itt. 44. (10.038 ?979). II. "Review process for scienti?c. informational. and educational documents? tOrder 2200. Envi- ronmental Protection Agency. Washington. D.C.I. p. 3. 28 U.S.Codc. ?655th ?976). Dry Color Manufacturers Association. Inc. v. Department of Labor. 486 Fed. Rep. 2nd scr. 98 and I04 {Third Circuit Court. l973l. l5. Gulf South insulation v. Consumer Product Scurry Conunission. 70l Fed. Rep. 2nd scr. I I37 IFiflh Circuit Court. to. All the primary source material referred to in this article was cited in full in our detailed treatment of this to it: lot. For convenience. we included citations cre only where we quoted directly front the source material. or where we felt the citation to be particularly important. IJ 'a-I SCI VOL. 232