Example: Free-ranging cheetah population in Namibia

By: Johnson et al (2013).

This example illustrates the use of a Bayesian Network for modelling the dynamics of a vulnerable or endangered population, such as the free-ranging cheetah (Acinonyx jubatus) population in Namibia. This requires the synthesis of ecological, biological and management perspectives. The data available to the conservationist is often varying in quantity and quality and in some instances there will be complete data paucity. Yet the plight of an endangered species cannot afford the luxury of waiting for perfect and complete data to be available before management action plans have to be set in place, and critical decisions made under uncertainty. What is required is a modelling framework that can integrate all the available data from varying sources, and to refine the model and parameters as new research or information comes to hand. Bayesian Networks (BN) have successfully been applied to situations such as these, and this example illustrates how effective they can be as a tool for understanding, designing and evaluating the effectiveness of policies and management action plans in conservation. We use a specialisation of BNs, object-oriented BNs (OOBNs) which extend the functionality of conventional BNs to better cater for the unique challenges faced by conservationists, providing an integrated, parallel modelling framework.

The article looked at four different scenarios which may be run using the nodes below:

1. Cheetah Removal from farmlands ceases To simulate the removal of cheetahs, click on the Decrease state of the Cheetah Removal node in the set of Human Factors nodes. Observe how this affects the predicted viability of the cheetah population, and also the expected economic benefit for farmers. You should also observe a very slight drop in the abundance of Prey Availability

2. Increased farmer and environmental education To represent this scenario, select Yes for both the Farmer education and Environmental education nodes. This scenario results in only a small change in the viability of the free-ranging cheetah population. However, substantial improvements in the nodes Cheetah removal and Human cheetah conflict are predicted by this scenario.

3. Climate change To try and represent climate change impact on the viability of the cheetah population, set the Rain node to Lower and Plant biomass production to Insufficient. This scenario predicts a big drop in cheetah population viability. If we expect plant biomass not be affected by climate change, then the predicted drop in population viability is smaller.

4. Disease outbreak Although the health of the current free-ranging population is good, we can explore what effect a disease outbreak is predicted to have on the cheetah population in north central Namibia. Setting the node Health to decreased reflects this scenario and predicts a worsening effect on the population viability.

Model documentation

Below is a set of HUGIN widgets for interacting with the model (click on the probability bar to instantiate a node or remove evidence):

---

---

References

Johnson, S. and Marker, L. and Mengersen, K. and Gordon, C. H. and Melzheimer, J. and Schmidt-Küntzel, A. and Nghikembua, M. and Fabiano, E. and Henghali, J. and Wacther, B. (2013) Modeling the viability of the free-ranging cheetah population in Namibia: an object-oriented Bayesian network approach. Ecosphere 4(7):90. External link to the manuscript.

Useful references for those interested in Bayesian networks include:

Kjærulff, U. B. and Madsen, A. L. (2013) Bayesian Networks and Influence Diagrams: A Guide to Construction and Analysis. Springer, Second Edition.

Contact information

Sandra Johnson, Australian Research Centre for Aerospace Automation (ARCAA), sandra dot johnson at qut dot edu dot au

---