James S. Latimer Office of Research and Development

1. James S. Latimer Office of Research and Development National Health and Environmental Effects Research Laboratory Atlantic Ecology Division Narragansett RI Regional Modeling Seminar Sponsored by the Council for Regulatory Environmental Modeling and Region 2 March 25, 2004 Overview of ORD’sModel Development Approaches under the National Aquatic Stressors Nutrient Program

2. Acknowledgements For some slides/summaries: the Aquatic Stressors Nutrient Workgroup Richard Green (chair), Gulf Ecology Division Edward Dettmann, Atlantic Ecology Division Jan Kurtz, Gulf Ecology Division Jo Thompson, Mid-Continent Ecology Division John Morrice, Mid-Continent Ecology Division John Kelly, Mid-Continent Ecology Division Michael Murrell, Gulf Ecology Division Pete Eldridge, Western Ecology Division Richard Devereux, Gulf Ecology Division Robbins Church, Western Ecology Division Russell Kreis, Mid-Continent Ecology Division Ted DeWitt, Western Ecology Division Walt Nelson, Western Ecology Division Coauthors: Giancarlo Cicchetti, Darryl Keith, Steven Rego, Edward H. Dettmann For controversial and provocative remarks: Jim Latimer, Atlantic Ecology Division

3. Implementation Plan for Nutrients Research • …improving the science supporting nutrient criteria development Mid-Continent Ecology Division Duluth, Minnesota Field Station Grosse Ile, Michigan Field Station Newport, Oregon Atlantic Ecology Division Narragansett, Rhode Island Western Ecology Division Corvallis, Oregon Gulf Ecology Division Gulf Breeze, Florida

4. ORD Water Quality Multi-Year Plan ORD’s Goal 2 Water Quality MYP LTG 1: Provide approaches and methods to develop and apply criteria (e.g. nutrient criteria) to support designated uses (fishable, swimmable, drinkable) NHEERL Aquatic Stressors Framework and Implementation Plan for Effects Research Provide scientifically valid approaches for protecting the ecological integrity of aquatic ecosystems from multiple stressors (e.g., nutrients)

5. Section 304(a) of CWA requires EPA to develop water quality criteria (nutrient criteria is a subset) …reduce anthropogenic component of nutrient over-enrichment to levels that maintain designated uses or prevent nutrient pollution in the first place Agency Problem Nutrient over-enrichment is one of the most often cited causes of impairment (CWA 305b reports) in coastal waters

6. NHEERL Aquatic Stressors Researchon Nutrients • Focused on coastal receiving waters (Great Lakes, estuaries and near coastal waters influenced by large rivers) • Goal – improve the scientific basis for developing and supporting nutrient criteria in the Nation’s waters by defining nutrient load-ecological response relationships

7. March 2003 Depth (m) June 2003 Transect Distance (km) Focus on three critical endpoints Develop nutrient load-response relationships and tools useful for developing nutrient criteria Submerged Aquatic Vegetation Project Food Webs Project Dissolved Oxygen Project

8. Systems do not respond similarly tonutrient over-enrichment Many uncertainties exist – cannot predict when a given nutrient concentration or load will have measurable adverse impacts The need for functional, response-based classification schemes • Changes in Primary Producer Component • Phytoplankton, SAV, Macroalgal biomass • 1o productivity • Production/Biomass ratio • Species composition Nutrient load Direct responses Filter Indirect responses • loss of critical habitats (oyster reefs, SAV. etc) • increased organic flux to benthos hypoxia • food web changes • jellies and other “bad swimming partners” • loss of commercial/recreational fisheries First-order classification parameters - Ecoregion/Biogeography - Hydrography (residence time,dilution) - Water clarity/turbidity/light Adapted from Cloern, 2001

9. Research Elements of the Critical Path SAV, DO, and Food Webs Response models, measures, indicators & tools Functional, response-based classification scheme Nutrient loading models, measures, &/or indicators Methods & tools to assess and link loadings to responses Assessment & validation of load-response model for all classes

10. ORD’s Research Approaches Goal: Predictive Models that will relate the level of nutrient loading to ecosystem response for dissolved oxygen, SAV and food webs in coastal receiving waters of the Great Lakes and US coastal estuaries Multiple System Empirical Models (Holistic Approach) 1) Identify the study systems about which we want to make predictions: coastal receiving waters of the Great Lakes and US coastal estuaries 2) Observe the behavior of the systems and search for patterns of behavior (response variables and nutrient loading along a gradient) 3) Express the pattern(s) of behavior in a rigorous/quantitative way to produce an empirical model or theory • Mechanistic, Simulation, Process Models (Reductionist Approach) • 1) Identify all of the important spatial and temporal scales of ecosystem processes • 2) Measure capacities, rates and exchanges of relevant components within and between all important environmental compartments at appropriate time/space scales • 3) Synthesize capacities, processes, interactions for all the important subcomponents into system model

11. Dissolved Oxygen Project Plan Multiple System Empirical Models (Holistic Approach): measure or determine DO or DO indicator, nutrient loading (or index) and classifying factors (residence time, ecoregion, water clarity) in whole systems along a nutrient loading gradient to directly relate nutrient loading to dissolved oxygen extent and/or duration Mechanistic, Simulation, Process Models (Reductionist Approach): Are being developed for Gulf of Mexico Hypoxic Zone, Yaquina Bay Oregon and Pensacola Bay Florida to understand nutrient effects on DO dynamics

12. Un-impacted Nutrient Impacted SO4-2 SO4-2 Stress – Response Model HS-2 O2 and HS-2 other oxidants SAV Project Plan Multiple System Empirical Models (Holistic Approach): measure or determine SAV extent or extent indicators, nutrient loading (or index) and classifying factors (residence time, ecoregion, water clarity ) in whole systems along a nutrient loading gradient to directly relate nutrient loading to SAV extent Mechanistic, Simulation, Process Models (Reductionist Approach): Are being developed for three species of eelgrass to evaluate health and growth under a variety of water quality circumstances.

13. Food Webs Project Plan Multiple System Empirical Models (Holistic Approach): measure or determine sensitive food web indicators, nutrient loading (or index) and classifying factors (residence time, ecoregion, water clarity ) in whole systems along a nutrient loading gradient to directly relate nutrient loading to food web response Mechanistic, Simulation, Process Models (Reductionist Approach): Are being developed to understand food web responses to stressors and to develop food web indicators Big fish Benthic suspension feeders Gelatinous zooplankton Benthic deposit & interface feeders Planktivorous fish Mesozooplankton Microzooplankton HYPOXIA Benthic detritus bacterioplankton Phytoplankton Biomass, production, composition Seston Macrophytes SAV, macroalgae, epiphytes Water Clarity Nutrients

14. Holism and Reductionism • Description • Comparison • Interaction

15. Source of Concepts Robert Peters, winner of the 1991 Ecology Institute Prize wrote this Excellence in Ecology book (using notes from Frank Rigler) as part of the award for receiving the prize. • Particularly • Chapter VII: Reductionism versus Holism: An Old Problem Rejuvenated by the Computer • Chapter VIII: Sources of Ecological Creativity • Chapter IX: Empirical Limnology

16. Reductionism vs. Holism • Reductionism: • The proper approach to the study of complex phenomena is to decompose this complexity to simple components. In many cases these components can then be shown to be instances of general physical and chemical laws. • With the advent of the computer systems ecology has become the dominant reductionist approach. (R&P) • Holism: • Complex systems must be treated as whole systems. Reductionism dominates ecology because philosophically it makes sense and with the advent of cheap digital storage and fast microprocessing it is computationally possible to produce system-wide simulations

17. Will the Alarm Clock Wake you in time? • Two methods to determine or predict when an alarm clock will go off • Holistic Method: Look at the behavior of the clock under a variety of circumstances and look for regularities or patterns. • Reductionist Method: Take the clock apart, study each screw, cogwheel, and cog, to see how they are constructed, study the shape of the teeth, and the scratches on the casing, look at the structure of the brass, steel and lead. Then study the interactions between and among these parts and eventually, painstakingly, the clock mechanism will be understood.

18. Systems Models • If we know the quantities of all components • If we know the interactions of all the components • Then we can predict the state of the system at any time in the future Implicit: We can identify all important structural/functional components and interactions a priori and can measure them accurately. If we want to dismiss or aggregate certain structural/functional and interaction we need to study each first.

19. The Scope of Systems Analysis, A hypothetical example • System: temperate lake • 200 species of plants • 100 species of herbivores • 100 species of carnivores • 50 species of bacteria • 200 animal species • 100 species of detritivores • Total 750 species = n(n-1) interactions: 561,750 • Interactions Analysis • The following would need to be quantified under the entire range of environmental conditions • E.g., rates of birth, growth and death, biomass, predation, competition, etc • E.g., light, temperature, nutrients, etc

20. Is Reductionism Vanquished?(not if you have $ and time) • Insurmountable difficulties? • The description of the entire ecosystem is so great an effort that we will never complete the task. In fact, we won’t even try. The job is just too big. • (Rigler and Peters, Science and Limnology, 1995) • Why has reductionism persisted even flourished: • Philosophically it makes sense • Practical problems deemed surmountable if enough data (time and money) are available But will reductive models be developed for smaller systems with less resources? Are holistic approaches useful?

21. Some Solutions? R&P “Reductionism and [its fruit Systems Models] are not inadequate on philosophical grounds but purely due to practical limitations.” The solution is simplification. However, simplification itself leads to difficulties

22. Source: Dan Campbell, Atlantic Ecology Division Simplification • Simplify total number of components • But obscure components may be important! • Aggregate into functional groups • Functional similarity must be determined which requires effort • Simplify analysis of interactions • Apply Michaelis-Menton kinetics using constants and empirical relations In short we must take short-cuts and thus replace reductionist rigor with rules or empirical relations showing that pure reductionism is impractical

23. Can Holism and Reductionism Coexist? • Rigler and Peters identified 4 steps to holistic studies. • 1. Identify the system (s) about which you want to make predictions • 2. Observe the behavior of that system(s) and look for pattern • 3. Express the pattern quantitatively to yield an empirical theory • The last step is where Rigler and Peters noted that reductionism and mechanistic models can be used most effectively. He called these models explanatory models and not theory producing models: • 4. Explain why the theory works with an analytical or explanatory theory • So mechanistic/process models can be important to: • aid in the explanation of empirical models • aid in understanding and addressing outliers • identify other variables that should be measured/determined in empirical studies • others?

24. Successes of Reductionist Approach • In Use: • The Chesapeake Bay Model • Long Island Sound System Wide Eutrophication Model • Others? • Under development: • Gulf of Mexico Hypoxia Model • Others

25. Success of the Holistic Approach - Lakes Reprinted from Rigler and Peters (1995)

26. 100 80 60 % Fish C from Phytoplankton 40 20 0 0 2 4 6 8 10 12 14 16 18 20 Phytoplankton Fish Periphyton Nutrient Loading Index Success of the Holistic Approach - Lakes Unpublished from Michael E. Sierszen

27. Success of Holistic Approach – Marine Systems 10.0 7.5 5.0 DO Index R2= 0.59 2.5 0.0 0 20 40 60 80 mgN * m-2 *Tr 7.5 y = -0.1268x + 8.9983 2 R = 0.6457 7 6.5 6 DO (MG/L) 5.5 Normalized By Residence Time 5 4.5 4 15 17 19 21 23 25 27 29 31 RESIDENCE TIME NORMALIZED TN (uM/d) Rhode Island Coastal Embayments Gulf of Maine Embayments

28. Regulatory Success of Holistic Approach – Marine Systems Calculate TMAL = 92,000 KG N/YR Water Quality Standards 65 50 40 Using Residence Time Depth Tidal Range . ORW 50 mg m-3 Vr-1 SA 150 mg m-3 Vr-1 SB 300 mg m-3 Vr-1 APPLY TO A SYSTEM NEW BEDFORD HARBOR CLASS SB CURRENT LOADING = 174,000 KG N/YR Buzzards Bay Side Embayments

29. Bio- and Effects Based Criteria Reductionist Approach Nutrient Loading Based Limits (Criteria?) ? Endpoint response metric Holistic Approach Nutrient Load Conceptual Approach to Using Load – Response Models to Establish Nutrient Criteria/Limits

30. Latimer is never one to be politically correct