Scale issues are fundamental to all ecological investigations, and have become a central topic in ecology in recent decades with the increasing recognition of broad-scale environmental issues (such as global warming) and land-management problems, rapid development of digital technologies (remote sensing, GIS, desktop computers, etc.), and emergence of landscape ecology. However, there are still many ecological studies in which scale is treated simplistically or superficially. Scale issues remain a key challenge for ecologists in the 21st century.
Scale issues include conceptualization of scale, scale analysis and scaling. This paper focuses on the former two issues. A three-tiered conceptualization of scale is introduced: dimensions, kinds, and components. Dimensions of scale are most general, including space, time, and organizational levels. Kings of scales can be distinguished among phenomenon scale (including structure and process scale, also referred to as characteristic or intrinsic scale), observation scale (also referred to as sampling or measurement scale), analysis or modeling scale and policy scale. The most specific and measurable definitions are components of scale, including grain, extent, spacing (or lag), coverage, cartographic scale, and support. Scale is often expressed as grain and extent in landscape ecology. Because each type of scale concept includes multiple terms and definitions, it is necessary to carefully discriminate, classify and unify them.
Scale analysis is to analyze scale effects and identify multiple-scale spatial patterns (especially characteristic scale(s)). Scale effects may occur in each of the following three situations: changing grain size only, changing extent only, and changing both grain and extent. When scale of observation, analysis, modeling, or experimentation change, statistical results (for example, mean, variance and multivariate relationships) are expected to change. Scale influences the results of examining spatial pattern, such as spatial heterogeneity, spatial distribution, spatial autocorrelation, spatial anisotropy, and patch and gap sizes. When scale changes, new ecological processes and patterns may occur, and the rate or frequency of processes, control factors, and correlations between processes may also change. Some landscape properties (for example, landscape openness, equilibrium, predictability, species richness and diversity) also exhibit scale effects. Because of scale effects, it is extremely important to conduct research at multiple scales. Identifying characteristic scale(s) means examining hierarchical structure of pattern or process, and distribution of patch sizes and spacing between patches, which are the basis for studying scale effects and scaling.
Some specific methods need to be developed for scale analysis, mainly including spatial statistics methods, landscape metrics and fractal analysis. Landscape metrics are the most popular methods which are both simple and potentially misleading. Spatial statistics methods and fractal analysis have been developed and applied in ecology during the past decades, and have a great potential in scale analysis and scaling. Spatial statistics methods include semivariance, scale variance, lacunarity, wavelet, representative elementary volume analysis, and so on. Each of them has its own advantages and disadvantages in identifying characteristic scale(s) and analyzing scale effects, thus it is essential to compare and evaluate two or more methods in the same study.
In summary, to provide a reliable foundations for scaling, scale analysis must be conducted thoroughly first. Research focusing on characteristic scales of landscape pattern or process and establishing scaling relations should be followed. Finally, a better understanding and thus explanation of the dependence of pattern and process on scale may be achieved.