beyond compliance - plant pathology · • the standards and trade development facility of the wto...
TRANSCRIPT
Queensland University of TechnologyCRICOS No. 00213J
BEYOND COMPLIANCEA new project on
integrated Systems Approachto Pest Risk Management
Peter Whittle1, Megan Quinlan2
Kerrie Mengersen1, John Mumford2
1 Queensland University of Technology2 Centre for Environmental Policy, Imperial College London
Standards and Trade Development Facility of the WTO
CRICOS No. 00213Ja university for the worldreal R
The problem
• Requires pest risk mitigation (biosecurity) measures– Subject to international standards (ISPMs). Must be based on pest risk, scientifically
justified, proportional to risk and least trade-restrictive
Pest in exporting country desire
Importing country is free of pest
• Pest risk mitigation measures are usually single, e.g. pest area freedom or chemical treatment. These can:
• Be difficult (or impossible) to achieve• Damage the commodity• Carry health and environmental risks• Halt the whole trade on a minor failure• Convey a power imbalance between trading partners
CRICOS No. 00213Ja university for the worldreal R
The opportunity
• Integrated pest management can be more flexible and robust than single treatments– “Systems Approach” – ISPM No. 14– Two or more independent risk management measures
• Can be difficult to develop andnegotiate due to uncertainty– What are the treatments?– How do they interact?– Where are the data?– We often feel we have some knowledge …
• How to overcome this?– Persist with single treatments?– Don’t trade until research is done?
CRICOS No. 00213Ja university for the worldreal R
The solution - Bayesian probabilistic models• Statistical paradigm that is less constrained
and more intuitive
p() p(X|) p(X)
---- prior
likelihood
posterior
Contrast with frequentistapproach, where you can only use the data (likelihood), and not until you have it.
CRICOS No. 00213Ja university for the worldreal R
Bayesian networks (BNs)• Use BN software (e.g. GeNIe)
• Build and parameterise the model using “prior” probabilities
– Priors can be estimates based on expert judgement
– The process of eliciting the model and priors is part of the negotiation
• Clarify and accommodate uncertainty by simulating (risk analysis)
• Collect data as the system is operated
• Update the model (“posteriors”) and review Elicit expert judgements
CRICOS No. 00213Ja university for the worldreal R
The motivation
• All countries – fundamental need for better trade systems
• Australia – trials with BNs to find solutions to potential loss of malathion/fenthion for post-harvest control of insect pests
• Europe - changes to the EPPO scheme, from PRATIQUE project, include uncertainty and matrix models or BNs
• North America – new regional standard on pathway analysis
• BNs enabled by computing advances, now very commonly used:• Modelling knowledge in computational biology and bioinformatics (gene regulatory
networks, protein structure, gene expression analysis), medicine, document classification, information retrieval, image processing, data fusion, decision support systems, engineering, gaming and law - (Wikipedia)
CRICOS No. 00213Ja university for the worldreal R
Collaborative project – QUT / ICL / JPM / STDF
• QUT and ICL developed plans through PRATIQUE– Apply BNs in a Control Point framework with case studies in developing countries
in Southeast Asia, with aid funding
• The Standards and Trade Development Facility of the WTO (STDF) funded a project preparation grant to hold a workshop in Kuala Lumpur in August 2010
– Five countries attended andall signed up, with regional agreement on case studies
• Full proposal to STDF – Funded for 2011/12 and 2012/13– Seeking supplementary funds
CRICOS No. 00213Ja university for the worldreal R
Commodity Exporting country Importing country
Fresh produce (not rubber plants) that may carry South American leaf blight of rubber
Countries with SALB Regional
Oil palm planting materialCountries outside the region Regional
Dragon fruit Vietnam South Korea, Taiwan
Jackfruit Malaysia China, Australia
Orchid cut flowers Thailand Europe
Mangosteen, avocado Philippines USA
Case studies proposed during the PPG workshop
CRICOS No. 00213Ja university for the worldreal R
Management Structure by Work Package
Project Steering Committee
CABI Southeast Asia Centre (CABI SEA)(Project Manager Dr A. Sivapragasam)
Work Package Leaders
TECHNICAL FRAMEWORKDr Peter Whittle
QUT
GOVERNANCEDr Jon KnightImperial College
London
COMMUNICATIONSCABI SEA
ADMINISTRATIONProject Manager (35%)Project Assistant (50%)
CABI SEA
CASE STUDIES – NPPO counterpart fromMalaysia ThailandVietnam
PhilippinesIndonesia - observing
Regional case studies include:CABI SEA support (Dr Lum)
One private consultant
CRICOS No. 00213Ja university for the worldreal R
BN for Pest Risk Analysis along the commodity pathway
We are finding that the risk management section of PRAs is commonly “underdone”. The focus of PRAs tends to be the risk, not how it can be addressed.
CRICOS No. 00213Ja university for the worldreal R
Restructured into a Control Point template- revised in each case study by group
Building a BN of a host-pest-environment system can be complicated and potentially overwhelming.
The Control Point approach is focused on the key questions and makes the task more manageable.
CRICOS No. 00213Ja university for the worldreal R
Conditional probability tables are populated by expert elicitation
• Estimate conditional probabilities, i.e. Pr (C|A,B)• For all combinations of the states of a node. Simple
models are easier!• Can make key estimates and interpolate automatically,
then test and review
CRICOS No. 00213Ja university for the worldreal R
Probability estimates are computed
CRICOS No. 00213Ja university for the worldreal R
… enabling analysis, modification and decision-making
CRICOS No. 00213Ja university for the worldreal R
Project outputs• A review of pest risk management in Southeast Asian plant
commodity trade
• A conceptual framework for Systems Approach decision-making, using Control Points and probabilistic or rule-based modelling
• Demonstrations using Bayesian Networks and/or matrix models, through case studies
• Establishment of a SE Asian competency base with the methods
• A plan for a harmonised framework for the region, which can be presented to the IPPC
CRICOS No. 00213Ja university for the worldreal R
Project outcomes• More robust pest risk management in the region• Better protection from exotic pest threats• More confidence in trade negotiations• New trade opportunities• Greater inclusion of stakeholders in pest risk management planning• Increased engagement in compliance• A draft regional standard (RSPM) on systems approach
• BEYOND COMPLIANCE