Abstract: PATTERNS AND THEIR MEANING The pattern of a landscape is, in its full detail, exceedingly complex. It is generally impossible to interpret adequately the relations of species and stands to one another and the landscape by observation alone. It is consequently necessary to develop abstract representations of the pattern, representations which show some relations of communities and environments which are most significant in the landscape pattern, but show these in a form more easily comprehended and apart from the complexity of the whole. The most familiar such abstract representation is the ecological series. In the complexity of the landscape paetern, certain main-directions of vegetational and environmental change may be recognized (cf. Meusel, 1940). Recognition of major correlations of properties of vegetation with differences in environment is originally direct and intuitive, but is later influenced also by means of measurement and interpretations of the significance of factors which, like those of the soil, are not so easily observed. When a single gradient is chosen for study, stand 110 THE BOTANICAL REVIEW samples may be arranged in sequence along this gradient to form an ecological series and interpreted as a gradient of environments and communities, an ecocline. Stands may be chosen and arranged in relation to a single factor-gradient, but the ecological series shows their relation not to a factor-gradient alone, but to a complex-gradient of many correlated factor-gradients, or of characteristics of environmental complexes (Whittaker, 1956). Within the ecocline one may choose to distinguish the complex-gradient of environments and the corresponding coenocline or gradient of communities (Whittaker, 1960). Although the ecological series is an approach toward isolation of a factor and its effects, it represents the variation in certain observed properties of ecosystems as most or all of these change along the gradient chosen for study. By the ecological series, characteristics of communities may be correlated with factors of environment, but the relation need not be assumed to be one of effect and cause. Environments and communities are coupled and interacting aspects of the ecosystem; environment acts not simply on the community, but in and through the function of the ecosystem to produce observed differences in community characteristics (Whittaker, 1954b). The relation between environments and communities in an ecological series may, however, have these characteristics: (1) The environmental gradient exists and can be measured apart from the presence of the communities along it. The gradient may thus be in a sense external to or separate from the community, although the gradient as it affects organisms may be modified by the community and the function of the ecosystem. (2) The relation between the gradient and communities is consistent; similar communities are observed to occur in habitats having similar levels or intensities of the gradient. (3) The normal complexities of ecological relations, effects of other environmental factors, chance differences in communities at similar levels of the gradient, and effects of communities in modifying the gradient not correlated with the gradient, may reasonably be neglected or controlled by choice of area or stands to be studied. (4) There is reason in present ecological understanding to think that the environmental gradient has significance in relation to the functions of ecosystems, such that differences in the functions of ecosystems that develop at different levels of the gradient are expressed in observable differences in communities. When these conditions occur, the relation between the environmental gradient and the gradient of community characteristics partially approaches the ideal of the "cause CLASSIFICATION OF NATURAL COMMUNITIES 111 and effect" relation (cf. Bunge, 1961). The synecologist in this area of study is concerned in general not with cause and effect but with correlations--variables which change together through an ecological series and which are often interrelated in the functions of the ecosystems along the gradient. To some degree some of these correlations approach the special circumstances to which designation of one gradient as cause or independent variable and others as effects or dependent variables may be appropriate (cf. Major, 1951; Whittaker, 1954b). When several major gradients influencing community characteristics are recognized in a landscape, stands may be arranged into ecological series in relation to each of these. An abstract representation of the landscape pattern as a multi-dimensional coordinate system of intersecting ecological series results (Ramensky, 1930; Sukatschew, 1932; Ellenberg, i950a, 1952a; Goodall, 1954a, 1954b; Whittaker, 1956, 1960; Bray and Curtis, 1957; Curtis, 1959). This general approach to study of landscape patterns and other relations of ecosystems through ecological series and abstract patterns (or by formal statistics of correlations and factor analysis) has been termed gradient analysis (Whittaker, 1951, 1952, 1956). The term expresses the fact that this is an analytic approach to ecosystems through measurable isolates as variables, and that the basis of relating stands to one another and a principal objective of the approach is the study of interrelations of gradients of environment, species populations, and community properties. For the techniques of arranging stands in ecological series or coordinate systems, and by extension for the approach itself, the term ordination (Goodall, 1954a; from Ordnung, Ramensky, 1930) is also current. Implications of such research for problems of classification may be clarified through study of an abstract pattern based on two major complex-gradients, using these gradients as axes of a chart (Fig. 1). Properties of the pattern represented by such a chart cannot be directly identified with those of the landscape pattern. The chart is a simplification of the landscape pattern; it omits from consideration factors not fitting into the complex-gradients studied. Points in the chart may represent, not particular stands, but average or most probable stand properties at a given combination of the gradients studied. The gradients represented as continuous on the chart are frequently interrupted by edaphic and topographic discontinuity and disturbance in the field. The chart summarizes changes of stands along the full extents of gradients which may be somewhere observed in the field by walking 112 THE BOTANICAL REVIEW VEGETATION OF GREAT SlVIOKY MOUNTAINS PRTERN OF EASTERN FOREST SYSTEM