Edited by Roland H. Lamberson
Table of Contents
Purchase of the Special Issue
The Natural Resource Modeling special issue on Individual-based Models is now available in hardback. It may be purchased from the publisher, The Rocky Mountain Mathematics Consortium. Ask for volume 15 number 1. The price is $25.00.
Payment should be made payable to
Rocky Mountain Mathematics Consortium
and addressed to
Rocky Mountain Mathematics ConsortiumAdvance payment is required, and the RMMC does NOT accept credit cards.
Arizona State University
Box 871904
Tempe, AZ 85287-1904
**********************************************
Introduction to NRM Special Issue on Individual-based Models
What Does It Take to Make Individual-Based Models Realize Their Potential?
Roland H. Lamberson
Department of Mathematics and Environmental Systems Program
Humboldt State University
Arcata, CA 95521
This special issue is based on the symposium "Advancing the Individual-based Modeling Approach: New Tools and Concepts", at the annual meeting of the Ecological Society of America, Snowbird, Utah, August 10, 2000. The work presented at the symposium is published in this volume or previously by Grimm (1999), Railsback et al. (1999), Railsback (2001a, b), and Railsback and Harvey (in press).
Bottom-up simulation models, i.e. individual-based and spatially explicit models, can be powerful tools in ecology and natural resource management. However, in the past some people have concluded that individual-based ecology has been a failure because much of its initial promise has not been realized. Our goal is to reconsider the status of individual-based models, present some promising new approaches, and give some examples of successful new models.
Despite their promise, at this time individual-based models have made little contribution to basic ecology or natural resource management (Grimm 1999). One root cause of their failure to advance ecology is the lack of an established conceptual foundation. For example, the practice of forcing model individuals to reproduce higher level behaviors like density dependence or territoriality that are observed in real populations is dangerous when the model is applied to previously unobserved situations. A second root cause has been the lack of appropriate software tools. Because individual-based models use individual behavior to predict population responses, the modeling software must allow individual behaviors to be observed; otherwise, the model is essentially untestable and unlikely to advance the science. Finally, we lack even a list of design considerations for bottom-up models.
The new field of Complex Adaptive Systems appears useful as a conceptual foundation for individual-based models. Thinking about such issues as emergent vs. imposed behaviors, what kind of adaptation is appropriate, how individuals predict decision outcomes, and how fitness is evaluated can help modelers identify and address the subtle but important formulation decisions that determine model success. Failing to let individuals make decisions predictively makes realistic behavior unlikely. Giving individuals simple fitness-maximizing decisions rules, and the information about their environment necessary to predict decision outcomes, can cause many realistic behaviors to emerge naturally (Railsback et. al. 1999, Railsback 2001a). Movement rules are a critical component of many individual-based models because movement is the fundamental method by which mobile animals respond to changing environmental and competitive conditions. Railsback et. al. (1999) propose fitness measures and movement rules which substantially advance the ability of individual-based models to reproduce observed fish behavior. The rules require mature fish to move to locations where their expected survival to a given time horizon is maximized while immature fish move to jointly maximize their expected survival and likelihood of reaching reproductive size.
Software design is critically important for individual-based models since the results of an individual-based model are the emergent properties of a system of interacting agents that exist only in the software (Ropella et. al. 2002). These outcomes can only be replicated by exactly reproducing the original software implementation. In addition, outcomes from an individual-based model are expected to be complex and novel, which makes the recognition of programming errors more difficult. Wide review and careful management of the software, the use of well-tested software tools, and providing for thorough and pervasive validation and verification can give assurance of a successful model. New software tools for individual-based simulations can also help manage software concerns. Technologies like animation and "probes" allow individual behaviors to be observed, making models much easier to test and improve, communicate, and believe (Railsback 2001a, Ropella et al. 2002).
It is particularly critical that the problem of communicating bottom-up models be solved if they are to be seen as scientifically credible. One solution to this problem involves equipping simulation programs with a graphical user interface that integrates elements of conventional debugging with graphical representations of the model’s state variables (Grimm 2002). One benefit of this approach to communicating bottom-up models is that peers can download the executable program from the internet and test and analyze the model on their own, which will thereby contribute to establishing an improved culture of analyzing, understanding, and communicating bottom-up simulation models.
In classical theoretical ecology there are numerous standard models which are simple, generally applicable, and have well-known properties. These standard models are widely used as building blocks for all kinds of theoretical and applied models. In contrast, there is a lack of standard individual-based models, even though they are badly needed if the advantages of the individual-based approach are to be exploited efficiently. The recently developed ‘field-of-neighborhood’ approach is a potential standard for modeling plant populations (Berger et. al. 2002). In this approach, a plant is characterized by a circular zone of influence that grows with the plant, and a field of neighborhood that for each point within the zone of influence describes the strength of competition, i.e. growth reduction, on neighboring plants. Local competition is thus described phenomenologically.
In this volume the software engineering issues mentioned above are dealt with in articles by Ropella, Railsback, and Jackson and by Volker Grimm. These are followed by an article by Berger, Hildenbrandt, and Grimm in which they propose a standard building block model for plant interactions to be used in individual-based models. Following these are articles presenting examples of individual based models. In the first James Anderson describes and tests a game theoretic foraging model developed in an agent-based framework. The animal 's environment is described in terms of agents representing prey, predators, and the habitat. This is followed by Railsback et al’s individual-based model of stream trout. It is analyzed by testing its ability to reproduce patterns of population-level behavior observed in real trout. In earlier papers Railsback et. al. (1999) developed the fitness measures and movement rules used in this model, and Railsback and Harvey (in press) analyzed individual behavior in the model by comparing the behavior of virtual trout with observed behavior of real trout. The third example is an individual-based model used to predict the movement patterns of brown-headed cowbirds. This model has been incorporated in a successful management program aimed at reducing the impact of nest parasitism by cowbirds on two endangered bird species at Fort Hood, Texas. We conclude this volume with a paper by Chris Adami, which addresses issues of modeling simple ecosystems using methods developed in the study of artificial life. This paper provides an overview of artificial life, in which simple virtual ecosystems evolve in real time. There are many opportunities for synergistic collaboration between the fields of artificial life and individual-based ecology.
References:
U. Berger, H. Hildenbrandt, and V. Grimm [2002], Towards a Standard for the Individual-based Modeling of Plant Populations: Self-thinning and the Field-of-Neighborhood Approach, Natural Resource Modeling 15.
Grimm, V. [1999]. Ten Years of Individual-Based Modelling in Ecology: What Have We Learned and What Could We Learn in the Future? Ecological Modelling 115, 129-148.
V. Grimm [2002], Visual Debugging: A Way of Analyzing, Understanding, and Communicating Bottom-Up Simulation Models in Ecology, Natural Resource Modeling 15.
G.E.P. Ropella, S.F. Railsback, and S.K. Jackson [2002], Software Engineering Considerations for Individual-based Models, Natural Resource Modeling15.
S.F. Railsback, R.H. Lamberson, B.C. Harvey, and W.E. Duffy [1999], Movement Rules for Individual-based Models of Stream Fish, Ecological Modeling, 123, 73-89.
S.F. Railsback, [2001a], Concepts from Complex Adaptive Systems as a Framework for Individual-Based Modelling. Ecological Modeling 139: 47-62.
S.F. Railsback, [2001b], Getting "Results": The Pattern-oriented Approach to Analyzing Natural Systems with Individual-based Models, Natural Resource Modeling, 14.
S.F. Railsback and B.C. Harvey [in press], Analysis of Habitat Selection Rules Using an Individual-based Model. Ecology.