Software for Modeling and Decision-Making

Many of the technical aspects of Adaptive Harvest Management are seemingly complex, and this apparent complexity may deter otherwise interested readers. We have developed a number of learning tools to help biologists grasp the basic concepts behind population modeling, statistical inference, and adaptive management, essential components of AHM. These tools were developed as part of instructional materials for two upper level courses at the University of Georgia: Applied Population Dynamics and a graduate seminar on Decision Making Under Uncertainty . Below we point readers to selected material from both of these courses; documentation is provided at the respective sites. As we develop modeling tools specific to the American black duck AHM project we will make these available as well.

Most of the population and statistical models are written in Java applets, which can be run from web browsers (i.e., do not require you to download a program.

Population Modeling

In modeling a population's response to harvest and other factors, it is fundamentally important whether the population is growing according to a Density independent or exponential model of growth, or follows a Density dependent or logistic model of growth. Under assumptions of logistic growth, one can also describe harvestable yield as a function of population size or density, which in turn can be solved for a theoretical Maximum sustained yield (MSY) . In practice MSY assumptions are not met, and alternative approaches have been sought for describing dynamic yield relationships. Specifically, much of the controversy over the effects of hunting on waterfowl and other wildlife populations is encapsulated by two alternative models of how hunting affects populations: the additive mortality hypothesis (AMH) and the compensatory mortality hypothesis (CMH). We illustrate these hypotheses in our applets on Additive and compensatory mortality.

Statistical Concepts

Much of Adaptive Management revolves around how management results in decision feedback, which is used to change the relative belief in alternative models (e.g., AMH vs. CMH). Several important statistical concepts are needed to appreciate how updating is accomplished. We provide applets to illustrate statistical distributions and sampling under a very simple statistical model (the binomial), and further illustrate the concept of a statistical likelihood via likelihood functions and maximum likelihood for the binomial.

These statistical concepts become part of the basis for our demonstrations of Decision making and adaptive management. Here, we use a graphically-based program for creating Bayes belief / Decision networks called Netica to develop simple decision problems that involve 1) sequential decision making, 2) a long-term objective (e.g., sustainable harvest), 3) uncertainty (both environmental and structural or model uncertainty), and 4) include a mechanism for incorporating information feedback via Bayes updating. Readers who are comfortable with these ideas are encouraged to consult the literature for more background and detail on AHM. An excellent review is provided by Johnson et al. in the online journal Conservation Ecology; readers will also wanted to consult many of the references cited in this article.

This page last updated on 10 January, 2002.

Send e-mail to the webmaster