Transcript Slide 1
Use of ecological models for risk assessments of plant protection products in Europe Pernille Thorbek Outline of talk 1. Introduction: Why do we need ecological models? 2. EFSA Guidance on Good Modelling Practice 3. Case Study 4. Conclusions 2 Protection goals EFSA protection goals opinion: ● “To ensure ecosystem services, taxa representative for the key drivers identified need to be protected at the population level or higher.” ● “However, for aesthetic reasons (cultural ecosystem services) it may be decided to protect vertebrates at the individual level. “ ● “To protect biodiversity, impacts at least need to be assessed at the scale of the watershed/landscape.” EFSA Journal 2010;8(10):1821 3 Match between standard tests and protection goals Measured Protection goal Exposure Maintained or simple decline Realistic (often variable) Exposure (birds, acute) Gavage Natural feeding pattern Species Standard species Naturally occurring species Endpoints Individual level Population level (or higher) Spatial scale Lab/Local Watershed, landscape ● Higher tier tests (e.g. field and mesocosm) closer to protection goals – but expensive and time consuming 4 Risk at higher levels of organisation ● Effects of pesticides on populations of non-target organisms depend both on exposure and sensitivity to the toxicant, and on: - life-history characteristics - population structure - density dependence - landscape structure ● Higher levels - Interactions between species ● Ecological models can combine such factors 5 EFSA Opinion 1821 EFSA protection goals opinion: ● “Given that most of the services under the selected specific protection goals are performed by populations or groups of populations, development of appropriate population models for use in risk assessment is needed.” EFSA Journal 2010;8(10):1821 6 Ecological models (mechanistic effects models, MEMs) are not new 7 Outline of talk 1. Introduction: Why do we need ecological models? 2. EFSA Guidance on Good Modelling Practice 3. Case Studies 4. Conclusions 8 When is a model good enough? ● Good modelling practice - Scientifically sound - Practical - Transparent 9 Trust me! WHAT DO WE NEED TO BASE DECISIONS ON MODELS? ● What IS the model? Conceptually and formally ● Why has it been designed this way? ● Has it been correctly implemented? ● Has it been thoroughly analyzed? ● Are the main effects well understood? ● How sensitive is model output to changes in parameters and model structure? How uncertain is model output? ● What are the indicators that the model is a sufficiently good representation of its real counterpart? 10 PAGE 10 Important developments over the last 7 years dN BD I E dt http://cream-itn.eu/ 11 RUC 2010: INTEGRATING POPULATION MODELING INTO ECOLOGICAL RISK ASSESSMENT SETAC EU SAG MeMoRisk 12 Principles of good modelling practice Schmolke et al. 2010 Grimm et al. 2014 13 How to link current endpoints with protection goals Lab based endpoint on effects (e.g. LD50 or NOEC) 14 What will the effects on individuals be under field exposure? What impact will that have on the population level? What impact will that have on the ecosystem level? TK/TD modelling Population Models Ecosystem models •Exposure profile •Life history •Species interactions •Toxicokinetics •Population structure •Landscape structure •Toxicodynamcis •Density dependence What effect will exposure have on ecosystem services? Ecological Economic models •Value of the ecosystem service •Cost benefit analysis Outline of talk 1. Introduction: Why do we need ecological models? 2. EFSA Guidance on Good Modelling Practice 3. Case Study 4. Conclusions 15 BEEHAVE team and collaborators Juliet Osborne (now at ESI, Univ. Exeter) Matthias Becher (now at ESI, Univ. Exeter) Peter Kennedy (now at ESI, Univ. Exeter) Judith Pell (now at J.K. Pell Consultancy) Jack Rumkee Jennifer Swain Volker Grimm Juliane Horn Peter Campbell Pernille Thorbek David Chandler Gillian Prince Sally Hilton International Keith Delaplane advisors: Steve Martin Peter Neumann Thomas Schmickl 16 16 Becher, Grimm, Thorbek, Horn, Kennedy & Osborne, submitted Model development of BEEHAVE ● The BEEHAVE model is based on a review of existing honeybee models and bee biology ● It has been extensively tested (verification, sensitivity analysis, comparison with independent data, pattern oriented validation) ● Is freely available with full model description and user manual 17 Other models: Henry et al. 2012 Science ● Estimated homing failure of honeybees following exposure to neonics ● Used Khoury et al. 2011 PlosOne model ● Predicted widespread colony losses (but see Cresswell & Thompson’s comments) 18 Model design 19 Becher, Grimm, Thorbek, Horn, Kennedy & Osborne, JoAE Reran scenario with BEEHAVE ● High forage flow (no limitaitons to colony growth) or Low forage flow (keep the colonies at the threshold of survival). ● Normal or doubled probability of forager mortality following Henry et al. (2012) for a 30 day period (seperate runs for each month in the year) ● Only one forage patch in landscape ● All scenarios were run for 5 years, reporting colony sizes and losses at the end of each year. 20 Becher, Grimm, Thorbek, Horn, Kennedy & Osborne, submitted Colony size after 1 and after 5 years 21 Becher, Grimm, Thorbek, Horn, Kennedy & Osborne, submitted Colony losses (low forage only) 22 Becher, Grimm, Thorbek, Horn, Kennedy & Osborne, submitted Outline of talk 1. Introduction: Why do we need ecological models? 2. EFSA Guidance on Good Modelling Practice 3. Case Studies 4. Conclusions 23 Conclusions ● Different types of models answer different types of questions ● Good models improve understanding of what risk a toxicant poses to populations - Also aid design of mitigation measures ● Good models take investment - Need to capture the key drivers affecting the population dynamics - Need to capture exposure and effects ● Models that mechanistically capture main drivers of system can be used for much wider extrapolations than models that statistically relate cause and effect ● Once a model has been developed it is (relatively) cheap to use 24 Thanks for your attention! 25