multi-country ecosystem management via interacting models of political and ecological processes...
TRANSCRIPT
Multi-country Ecosystem Management via Interacting
Models of Political and Ecological Processes
Timothy C. HaasSchool of Business AdministrationUniversity of Wisconsin-Milwaukee
[email protected]/~haas/ems-cheetah/
Outline
• Ecosystem management with interacting models of political and ecological processes
• Example: Management of cheetah across East Africa
System Characteristics
• Probabilistic models of groups and the affected ecosystem fitted to data
• Practical management strategies found from these fitted models
Group Models
• President, EPA, rural residents, pastoralists, and NGOs
• Groups act to reach economic, militaristic, and political goals
• Internal (distorted) perceptions of other groups and the ecosystem
Example: Simplified Group Influence Diagram
Endangered Species-Focused Ecosystem Model
• Latest population dynamics model
• Convert to a stochastic differential equation system to add uncertainty representation
East African Cheetah Management
• Presidents, EPAs, rural residents, and pastoralists of Kenya, Tanzania, and Uganda
• Conservation-focused NGO
• Cheetah population dynamics in each political region
Data
• Political actions from on-line newspapers over 1999-2006
• Artificial cheetah counts based on actual data from 1998-2000
Observed Group Actions for Kenya
Kenya Actions from Fitted Model
Data-Model Agreement
• 23% of observed action-target combinations matched by fitted model
• Error rate: 1 – 0.23 = 77%
• Blind guessing error rate:1 – 1/20 = 95% (with 20 decisionoptions)
Model-Based Most Practical Management Strategy Setup
• Specify desired future ecosystem state
• Example: 1000 cheetahs 50 years hence
Solution
• Find smallest change in group belief systems that will result in a sequence of group actions that lead to the desired ecosystem state
Conclusions
• A political-ecological model of ecosystem management decisions can be built
Conclusions continued
• This model can be calibrated to political-ecological data
• This fitted model can out-perform blind guessing of future group decisions