Magazine article AI Magazine

Qualitative Reasoning about Population and Community Ecology

Magazine article AI Magazine

Qualitative Reasoning about Population and Community Ecology

Article excerpt

Why use qualitative representations for ecology? A number of textbooks published recently (for example, Haefner [1996]; Jorgensen and Bendoricchio [2001]) show that ecological modeling is almost synonymous with mathematical model building. These models might be precise and sometimes closely mimic what we believe is happening in the field, but they often fail to capture the mechanisms that actually explain the observed behavior (Gillman and Hails 1997). Moreover, they require numeric data of good quality, and ecological data are often difficult to obtain because long-term observations are required, and experimentation with real systems is limited. Hence, ecological knowledge is heterogeneous, including both quantitative and qualitative aspects. It is imprecise, incomplete, qualitative, and fuzzy; is expressed verbally and diagrammatically; and is, therefore difficult to model using a mathematical approach. As noted by Rykiel (1989), ecologists have a considerable amount of knowledge "in their heads" and not many ways to make this knowledge explicit, well organized, and computer processible. In this article, we show how qualitative models can be used to address some of these problems and, thus, become a valuable complement for mathematical approaches to ecological modeling. After all, many questions of interest in ecology (especially to decision makers) can be answered in terms of "better or worse," "more or less," "sooner or later," and so on (Rykiel 1989).

Qualitative representations provide a rich vocabulary for describing objects, situations, relations, causality, and mechanisms of change (for example, de Kleer and Brown [1984]; Forbus [1984]). With this vocabulary, it is possible to capture commonsense knowledge about ecological systems and use this knowledge to automatically derive relevant conclusions without requiring any numeric data. Another important feature concerns the idea of using a compositional approach to enable reusability (Falkenhainer and Forbus 1991), which is achieved by constructing libraries of partial-behavior descriptions that apply to the smallest entities relevant within a domain. As larger systems are built from these basic elements, reasoning about the behavior of larger systems means combining the behavior of these elements. Thus, we avoid having to develop dedicated models for each system encountered. A third feature of qualitative models, relevant to ecological modeling, is their ability to provide causal explanations of system behavior. Deriving the behavior of a complete system from the behavior of its constituents facilitates an explanation of the overall behavior in terms of these constituents. Such explanations are considered insightful, particularly when the set of partial models captures a causal account for the behavior of the constituents (for example, Forbus [1988]).

An interesting problem to illustrate the potential of qualitative reasoning for modeling ecological knowledge comes from the Brazilian cerrado vegetation. The cerrado is a large biome consisting of a number of different physiognomies (well-defined communities). According to a widely accepted hypothesis, changes in the fire frequency influence the composition of the cerrado vegetation. This succession hypothesis states that bio-diversity is lost, and the vegetation becomes dominated by grass species when the fire frequency increases. When the frequency decreases, the vegetation changes into forestlike physiognomies. The hypothesis has received support from different studies (for example, Coutinho [1990] and Moreira [1992]) and become the basis for environmental education and management decisions about the cerrado. However, knowledge about succession in the cerrado is incomplete and imprecise. It mainly provides a conceptual description of the succession process. Our goal is to construct models and run simulations that express this conceptual knowledge.

In the following section, we first introduce our qualitative theory of population dynamics. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.