Predicting across Scales: Theory Development and Testing

Monica G Turner, Virginia H Dale, and Robert H Gardner. 1989. “Predicting across Scales: Theory Development and Testing.” Landscape Ecology 3 (3): 245–52.

Turner et al. synthesize discussion from a workshop entitled “Predicting Across Scales: Theory Development and Testing”, with the goal of defining key concepts, approaches, and directions for multi-scale research in ecology. While multi-scale research can be applied to many sciences, it is especially relevant to landscape ecology, which often involves large heterogeneous systems whose scale makes empirical studies intractable. Many of the concepts are similar to the Meier-Shellersheim et al. (2009) paper, but one that particularly stands out is the idea of a “critical threshold”, which is defined as an “abrupt change in some quality, property or phenomenon in the system”, at which transference of information from one scale to another becomes difficult. Although Turner et al. (1989) describe this a limit or boundary of the model, I think the idea could also be applied to help delineate scales within the model. According to the authors, there are four parts to predicting across scales: identification of scales, understanding the parameters at different scales, translating across scales, and experimental testing and sampling at multiple scales. The identification of scales is somewhat subjective on the part of the researcher, but must also be appropriate for the phenomenon of interest, and can be determined from scale-dependent changes in patterns. When understanding parameters at different scales, the measurement of interest is the relative importance of the parameters. For example, regional climate is certain affecting the dynamics of a forest patch, but its relative importance to other, more local parameters, may lessen its impact and lead a researcher to omit it from a model. Translating, or extrapolating, information across scales may be approached through a bottom-up or top-down perspective, both of which apply constraints, but at different levels. Turner et al. discuss the possibility of conducting multi-scale experiments, however there are sampling frequency and replication limits that reduce this techniques usefulness. Given the year of publication (1989), there is little mention of computational methods of answering multi-scale questions, although simulations are proposed as one method that has shown promise in ecology. Although this paper was published two decades before Meier-Shellersheim et al, it clarifies many of the same concepts and introduces the field of multi-scale modeling as it applies to ecological systems.