Paper/Poster/Presentation Title
Keywords
uncertainty, extinction, simple models, management
Start Date
27-6-2018 9:00 AM
End Date
27-6-2018 10:20 AM
Abstract
Managers make decisions on how to protect and restore environmental systems, along with associated ecosystem services. Ideally, these decisions are made with the support of predictive models, built on a sound understanding of how the system works, validated with data and accompanied by credible estimates of uncertainty. However, in many cases data are scarce or non-existent, understanding of key processes is limited, and disturbances such as land-use change or invasive species may be fundamentally altering the way the system functions. If there are insufficient data available for meaningful parameterization or validation, and no prospect of more data becoming available, how can uncertainty be assessed? Under what conditions is it no longer useful to even build a model? We consider these questions in the context of wildlife management on Christmas Island, an Australian territory in the Indian Ocean. The island has numerous invasive pest animals and diseases which threaten the unique and diverse native wildlife; feral cats are a major problem, and a cat eradication program is underway. However, it is known from other island ecosystems that eradication of predators can have unintended negative consequences for native species. Here we investigate whether converting total predation pressure to a measure of “cat-equivalents” could provide an early warning system for perverse outcomes from eradication of cats on Christmas Island. In developing our model in a data-poor system, we ask: how much information is needed to make a model useful?
How many cats (or “cat-equivalents”) does it take to cause an extinction? And how simple can you make a model before it becomes useless?
Managers make decisions on how to protect and restore environmental systems, along with associated ecosystem services. Ideally, these decisions are made with the support of predictive models, built on a sound understanding of how the system works, validated with data and accompanied by credible estimates of uncertainty. However, in many cases data are scarce or non-existent, understanding of key processes is limited, and disturbances such as land-use change or invasive species may be fundamentally altering the way the system functions. If there are insufficient data available for meaningful parameterization or validation, and no prospect of more data becoming available, how can uncertainty be assessed? Under what conditions is it no longer useful to even build a model? We consider these questions in the context of wildlife management on Christmas Island, an Australian territory in the Indian Ocean. The island has numerous invasive pest animals and diseases which threaten the unique and diverse native wildlife; feral cats are a major problem, and a cat eradication program is underway. However, it is known from other island ecosystems that eradication of predators can have unintended negative consequences for native species. Here we investigate whether converting total predation pressure to a measure of “cat-equivalents” could provide an early warning system for perverse outcomes from eradication of cats on Christmas Island. In developing our model in a data-poor system, we ask: how much information is needed to make a model useful?
Stream and Session
F3: Modelling and Decision Making Under Uncertainty