June 12, 1998

Friday Afternoon
Breakout Sessions

Objectives:

Need to focus on

• Measuring and implementing adoption of IPM practices.
• Risks of pesticides.
• Measuring success.

• Data - access/existence/quality.
• Bridging the reality of the complexity of ecological systems into manageable measurements.
• What constitutes a significant difference between two scores (environmental boundaries around predictions)?
• How do we interpret numbers?
• Identifying resources.
• Translating results into action.
• Threshold levels - when do people change their behavior?
• Under what conditions can we use proxies as indicators?

As can be seen by the List of Participants, Group 1 was highly diversified, offering a nice blend of interests and expertise concerning decision tools that support measuring the adoption of IPM practices and pesticide risk reduction. Discussions were noteworthy for both inclusiveness of multiple view points and unanimity in participation. After a brief round of the table for introductions, the Chair initiated discussion of the necessary components for a measurement decision tool, but upon advice of the workshop facilitator, he shifted discussion to three areas -- Data challenges, Information Sources, and How to bridge from the complexity of ecosystems to a manageable measurement decision tool-- all of which must be addressed in evaluating decision tools for measuring adoption of IPM practices and pesticide risk reduction. The group dynamic was essentially one of brain-storming/discussion of these three topics.

DATA

[Note: "*" indicates priority placed by group] Data accessibility

• proprietary/confidentiality
• international comparability
• multiple names for chemicals
• a.i.s linked with tox data made publicly available
• need for standardized formats/measurement units/filling data gaps*
• a.i./formulations/inerts
• importance of context/data interpretation
• need meta data (info about how data collected, etc.)/confidence limits*
• how do we prioritize data for use in decision making

• completeness of data/data gaps
• source of data
• data quality*

• need to know audience for data

• how to deal with new data from industry?
• how can we influence industry to give us best data (protocols, guidelines, Data Quality
• Objectives)?

• EPA/OPP Tox 1-Liner databases going on Web: evaluation of toxicology studies for registration of pesticides; reference dose (RfD), and Cancer classification (Q* data bases) are available on diskette.
• Dutch endpoints are published
• Pesticide Environmental Stewardship Program's Partners? risk reduction strategies go on EPA/OPP website.
• Fate-related Models such as Gleams, Prism, Exams on web.
• ARS Beltsville database on web on half-life and KOC
• Illinois Critical Trends Analysis Project B III Natural History Survey
• NRCS soil data
• IOBC data from ring tests (same pesticide on different targets)

Communications and Buy-in by Intended Users

• Who are the stakeholders/potential users of indicator data and systems' outputs?
• Data and systems? outputs must be understandable to users "Around at the end."
• Stakeholders must agree on indicators.
• Simple may not be precise.

Characteristics of Indicator Tools
Possibly may have same analytical tool but different coefficients so that data is weighted in accordance with concerns, depending on site specifics, routes of exposures, subpopulations, goals of measurement, etc. *

Risk metrics (combining data with different units of measure by evaluating and converting to scale of 1 to 10, for example) plus value metrics (what matters to stakeholders) included in measures. *

Validated. Tools should be validated by evaluating monitoring data on GW/SW concentrations, beneficial populations, etc., particularly as you move from complexity toward simplicity. *

Role of Peer Pressure and Education in Driving Adoption

CHALLENGES

Important to maintain the diversity of the "IPM Tool box." As a last resort, some hazardous chemicals with, for example, high toxicity/short half-life (residue), will be needed to manage pest outbreaks.

EPA should emphasize standardizing data formats to support data use by public.

Leadership on risk reduction/IPM: OECD, EPA, USDA, ISO, IPCS: problem for each is lack of resources. Need "real money" for IPM.

IPM credibility

Opportunities that FQPA offers to emphasize pesticide risk reduction and IPM.

Stay focused on IPM's balancing of short term and long term profitability: economics are always there.

Measurement community needs to be more supportive of efforts to develop measures:

• As systems emerge, don't just be critical of them; appreciate and support the efforts as opportunities to build, to extend knowledge base for measures.

• Recognize people that have good data; make sure their bosses know. Recommend establishment of award for innovators in the field of measurement for example, "Carol Browner innovation award."

• Identify stakeholders and build coalitions to better support consistency in interpreting data. Work with stakeholders to identify needs; how people can use data.

• Insist that assumptions and manipulations of data bases are transparent.

• Support communications/connectivity between measurement developers by supporting bulletin boards and information "clearinghouses" on measures.

• Conduct a conference in Washington, DC to call attention to the needs for measures, involving stakeholders and coalitions as sponsors to focus on the needs for resources and federal agency support.

• Make environmental concerns "sexy" to the public.

Q.: Are you (EPA) planning to publish the PESP (Pesticide Environmental Stewardship Program) strategies on the internet? I'm concerned about putting data from projects on the Internet because I don't think that the partners would agree or be comfortable with that. [PESP strategies are the PESP Partners' plans for promoting pesticide risk reduction and the implementation of IPM.. They are required to develop these strategies to be a Partner, and the strategies are reviewed and negotiated with EPA.] What do you plan to do with the information provided?

A.: Partners are informed that strategies will go on the Web, and there had been some expression of concern from some Partners. Other Partners, on the other hand, see the publishing of strategies as a positive thing that would inform the public of what they are doing to reduce pesticide risk. If a Partner had a concern about their strategy going on the Web, however, that concern could most certainly be negotiated with EPA. The purpose of putting the strategies on the Web is to promote an understanding of the PESP program and what Partners are doing to achieve risk reduction and to provide models for other Partners or potential Partners to consider in preparing strategies. The use/risk data that Partners submit in their annual reports are for both the Partner and EPA, to understand how well strategies are performing in achieving risk reduction. EPA has no plan to publish data from the Partners on the Web, and for sure Partners would have concerns about such data being published. Perhaps PESP Partners might support a Partners' clearinghouse for such data to assure they are used responsibly to support pesticide risk reduction. But would you support such a clearinghouse?

* There are many connections between the groups. Nomenclature needs to be addressed by official bodies as a better system of common names. Using official EPA name and number is not recognized internationally.

Our group had a good mix of supply side (those with access to data) and demand side (assessment system architects who need data), and a good mix of people from within Federal Agencies and without, as well as a good mix of US and foreign input. We began with in depth introductions. The session was then roughly divided in half, first discussing DATA ISSUES and then the "BRIDGING QUESTION."

DATA ISSUES include:

• identifying resources

• routes of access to data

• perceived gaps - with some challenges to the significance of these gaps

The BRIDGING ISSUE is essentially how to encapsulate a complex reality in manageable assessment tools which can be used by policy makers. Some attention was paid to specifics, such as what constitutes a significant difference in the interpretation of data.

Our European colleagues explained the policy context that has been the impetus (at least in terms of funding) for much of their work: European farmers are subsidized (at times with subsidies > 50% of income), and there has been increasing discussion about tying the subsidies to environmentally-friendly behaviors. Thus there is demand for assessments/indicators of the ENVIRONMENTAL IMPACTS of agriculture and specific agricultural practices/strategies.

DATA RESOURCES AND ACCESS
As we began the discussion of data, the question was raised (by a colleague from Europe) why EPA is not doing the work of data presentation and analysis, since the agency handles so much of the relevant data. The following points were brought out in response, or later in the discussion:

• EPA has a legal mandate for pesticide review and (re)regulation, from which they are backlogged and ill-equipped to take on additional responsibility

• there are statutory limitations on the work which EPA can undertake

• Much of the data relevant to comparative risk assessment are drawn from the gray literature, and may likely not be peer reviewed for the application to which this risk assessments are putting it-so it is unlikely they would get a government seal of approval (We discussed the point that comparative risk indicators and assessments the success and adoption of IPM programs often must use data collected for other purposes, e.g.: as basis for deciding label restrictions or tolerances for a single pesticide).

• most OPP personnel do not have access to the internet.

With this in mind, we exchanged information re: data sources, data needs, data manipulation:

1. EPA EcoTox Database (Brian Montague). It has > 10,000 records, one for each test on each test organism. Question of how best to utilize data, especially when there are multiple data points for the same taxonomic group: median values, most sensitive values, ...?

2. EcoTox out of Duluth EPA (unclear what form or how to gain access)

3. EPA Fate Database-not currently very interactive

4. Cancer Database (new version just released) and Tox 1-liner (access is difficult)-both managed by Rick Whiting

5. Reduced Risk Chemicals--information packets compiled by registrants have not been reviewed by EPA, and entries may change following Agency review-esp. carcinogenicity potential. Doug Urban will talk to Peter Caulkins and Alan Dixon, who are responsible for Reduced Risk registrations, about access to R.R. packets of information. Summary packets contain easily comprehended comparative risk data for the new pesticides vs. the pesticides they would replace.

6. PHED--exposure database for studies of occupational exposure used in risk assessments. The database may be proprietary and therefore unavailable to the research community.

7. Data are missing for older (particularly element-based) chemicals and less data are available for biopesticides.

8. Earthworms: Lois asked about indicators using earthworm data because has not found a consistent set that reflects field conditions. Volmar Gutsche reported that Germany requires earthworm data for pesticides registration, and he drew on this for his assessment. Doug Urban reported that EPA has not found earthworm lethality to be a sensitive indicator - response is erratic (Hayo van der Werf reported the same re: France). Thus need either to develop a non-lethal earthworm endpoint or else to look to other indicator species. EPA is therefore thinking more about beneficials.

9. Beneficials: of increasing interest worldwide. H. Vogt has developed the Ring DB.

10. Soil impacts: need to develop good indicator. Microbials are difficult to evaluate out of context-pesticide impact may increase or decrease a given spp. Increased biomass may result from decreased spp diversity-unclear which is more important. However, perhaps impacts on microbials could be incorporated into an expert system decision tool format.

11. Avian: Chuck Benbrook has been working with a group of ecologists and others (including the Canadian avian specialist Pierre Mineau who has developed a model for OP impacts on birds) to address the question of how to interpret ecological information for specific taxa.

12. Fish: Eric Hesketh and NAPRA group is looking for good data on long term impacts of pesticides on fish, in order to determine maximum concentration levels

13. NASS Usage and Practices Data: hard to know how to access NASS (National Agricultural Statistical Service) data-need to be SAS expert.

14. Role of GIS in usage and environmental assessments

15. Aggregating data where there are multiple data points (as with Eco Tox)-Doug Urban pointed out that it is always accurate to use median values, whereas the geometric mean and probabilistic assessments assume a random distribution, which one rarely has, and therefore the application is statistically invalid.

"BRIDGING QUESTION"-HOW TO BRIDGE BETWEEN THE COMPLEXITY OF REALITY AND THE SIMPLICITY OF INDICATORS

Doug Urban described the evolution in his approach to comparative risk assessment from (a) projections based on impacts of maximum rates of use on the most sensitive species to (b) an effort to consider multiple types of input data-e.g., a range of sensitivities (most sensitive, median sensitivities, etc.), at different use rates (maximum and typical), and different chemical forms (granular, emulsion...). He then puts together from this a suite of risk quotients which can be compared with a comparable suite for other chemicals.

The caveat is that this model is generally applied to specific risk situations (e.g., impacts of a pesticide in a given river). It might be instructive, however, to summarize this model and contrast it with alternative model(s) used for policy-relevant indicators, in order to assess + and minuses of each approach:

ONE APPROACH: At each stage utilize full data complexity and base analysis on this robust array of data. Have complex/comprehensive analysis behind the simplicity of the index.

ANOTHER APPROACH: Simplify at all stages of analysis.

Other issues:

• We moved into the inevitable discussion re: WEIGHTING OF FACTORS:
  Eric put out the challenge that "truth"-i.e., real impacts/risks and the "importance weights" given by stakeholders sometimes have nothing to do with one another. Discussed ramifications of this, and value of different approaches to weighting of factors.

  Decide Right software by Avantos enables sensitivity analysis in judging of factors. Based on a linear hierarchical analysis, enables entry of rationale for weighting.

  One of the values of this and other formal systems for weighting values is that it is a learning experience to see how results change with a change in values.

• CONCEPT OF THE RISK INDICATOR AS A LEARNING TOOL, A PROCESS: This perhaps is more relevant for risk management decision-making (tools for pest managers), than for risk analysis and assessment (tools for policy makers). For a POLICY TOOL, probably want to be more directive.

• In creating POLICY TOOLS want to highlight the "central tendency" in impacts on multiple species across an array of circumstances. "Policy makers don't care if you kill daphnia."

• For POLICY TOOLS: Know what key piece of information is compelling to the specific audience. To do this NEED A SUITE OF TOOLS. If the objective is to provide legislators with summary trends, they don't want much information. IPM coordinators require a greater level of substance.

• SIGNIFICANT DIFFERENCES: how large a difference in results is significant?
  Chuck Benbrook: don't draw conclusions from anything less than a magnitude of order.

  Hayo: the relevance of an order of magnitude depends on the scale of analysis.

  Lois: Assessment should provide guidance for inferring relevant and significant differences in impacts. Aim to base these significance breakpoints on differences in output or impact, rather than simply use numerical breakpoints (e.g., many assessment set thresholds so that LD50 &grtr; 1000 is in one hazard/risk category, 100-1000 in another, and &less; 100 in another; but these are just numbers, not differences in impacts.)

Questions & Comments:

• The idea of using data for active ingredients versus formulations needs to be considered. LD50 value might be very high while for GATT might be very low. Any system tends to be very low and has to be taken into account.

• There are many maligned policy types and risk managers. In last few years there has been an aversion to looking at central tendency reliability.

• Regulators and risk assessors have had a mandate to develop guidelines to see if there might be some effect (such as maximum dose). Risk assessment may need to determine, under proposed uses, the impacts on non-target species. People answering different questions, i.e. the overall impact on farm, watershed, state and national scales, will not be making fine-tuned decisions. In other words, people need different sets of data and new tools to use data in different ways. EPA has lots of data and risk assessment (such as the impact on birds and fish). There need to be attempts to mitigate risk by changing use specifications on pesticide labels.

• In terms of the bridge between complexity and simplicity, robust technology and policy makers, there has to be a logical extension - it is a communication issue.

• With materials that are highly toxic, the bridge to reality is exposure and that is not always considered. Do we in fact think that policy makers consider daphnia? Ecological risk has held up the replacement of pesticides because of ecological concerns in registration. It is a very complicated set of analyses done by the agency. While we do want to look at central tendency, we need to realize that this is a highly political issue.

• The challenge is one of scale, sensitivity and access. We need to build a model whose outcome is scale communication; these are mutually exclusive categories. Who cares? More importantly what is an outcome of this workshop? What is the most useful tool? We are talking about cross-purposes. We need to put boxes together so we can pull it out.

• We can do all three.

• Observation: We are making complex decisions based on laboratory data. When people look into system and find different answer, it bothers me because there is too much uncertainty.

• Data not from the laboratory is what people are trying to put into models. We have knowledge based and field-based data. We should not assume every risk indicator is being built-in to laboratory tests. If we look over end indicators, they are much more robust.

• EPA/OPP has never made a decision to register or not to register a pesticide based on laboratory tests. The first tier is to proceed to experimental use permits. The process of registration includes field studies.

• Most studies in the literature seem to be driven by industry or laboratory values.

• Risk assessment never uses death as endpoint.

• I would like to reinforce what I said earlier. Cruder use tools are useful for setting benchmarks; these tools are not useful for regulation but are useful for setting benchmarks over time.

• There are many objective for using tools.

• I cannot even think of any part of EPA regulation decisions that are based on these tools.

• Databases are no more than pointers.

1. Pesticide databases

Not maintained, updated.
Many are not accessible, publicized.
Missing data; single values rather than ranges that reflect uncertainty or responsiveness (pH, average temperature).

--> Need additional resources, must be web accessible, need organized and centralized sources [uninteresting (not sexy!) job, but requires highly trained/experienced people (QA)].

2. Appropriate use of indicators and indices must take spatial and temporal scales into account.

field/farm level, project/program evaluation, state/national risk or hazard assessment.

3. Problem: lack of site/(context) specificity of pesticide environmental rating systems.

Indices should be adapted to local conditions, information.
Add or integrate measures of environmental vulnerability (e.g., groundwater sensitivity).
Local/regional validation.

4. Issue: Outreach/delivery systems.

5. Need long-term commitment of resources for project evaluation.

However, because of the many variables that influence pesticide selection and management decisions (the context is not stable), it will be much easier to demonstrate changes in practices or environmental impacts than it will be to demonstrate that such changes are due to project intervention.

[Exception: dramatic changes-environmental crises, regulatory pressure, substantial positive incentives, large project/program resources]

Challenge: Many/most projects lack environmental monitoring component: Do the changes affected by a program result in actual changes in environmental endpoints?

Need field studies in order to validate pesticide environmental impact assessment (EIA) tools, specifically tools which address exposure factors. (for example: for groundwater or surface water compartments, if given a predicted environmental concentration/likelihood of exceeding a tolerance, or a relative likelihood for contamination . . . . .)

6. Problem: the simplification of complex issues/systems (Validity of pesticide environmental ratings).

There is a tendency to neglect exposure and focus on toxicological effects
There is a tendency to combine unrelated effects or effects best considered separately, into single category or equation, for example, environmental effects and efficacy such as chronic and acute toxicity.
Therefore we need appropriate toxicological endpoints based on the application/question.

7. Toxicological data: appropriate use as indicators in indices:

• Example: Use of oral LD50 where dermal LD50 is more appropriate (farm workers).
• Example: Simplifying information on endocrine disruption (ED) and carcinogenicity into qualitative categories to which numerical values are assigned for incorporation in multi-attribute indices.
• Also, what if a chemical's ED affects are evidenced as cancer? (are we not counting twice for the same result?)
  Thus, we need to reevaluate how to score carcinogens and ED's.

8. Need to maintain management flexibility: pesticide selection and management environmental trade offs.

Question and Comments:

We need to look at all tools in the IPM tool box and we also need to be flexible. There should be more consideration as to which toxicological end point we use. We also need to consider what the goal of the environmental indicator is. Maybe we only need a relative ranking and general trends. We also need to take into account exposure when trying to measure pesticide toxicity.