Complexity in plant communities: The notion and quantification

Although an intuitive notion of community complexity is well established in ecological theory, its quantitative definition in other than just surrogate terms continues to elude the practitioners of the art. We examine the notion in its broad sense and develop a new measure based on the average length L(S) of the communication-theoretical parsimonious code required to describe the community (S). We use data from different model and natural communities to study the behaviour of L(S). Interestingly, the disorder-based entropy quantity H(S) is a lower-bound of L(S) and thus the difference Delta(S) = L(S)-H(S) is a quantity of potential theoretical significance. We show that Delta(S) can be substantial quantitatively and meaningful ecologically. Specifically, it displays patterns of sensitive behaviour vis-ci-vis species richness, level of disorder, vertical community layering, growth-form richness and nonlinearity in species-response. Delta(S) measures a component of community complexity distinct from the disorder-based component H(S). As such, we refer to Delta(S) as ''structural complexity'', H(S) as ''disorder-based complexity'', and L(S) as ''total complexity''. Since L(S) responds not only to disorder which arises from chance sorting, but also to emergent structures which issue from organization, it is more in line with our intuitive notion of community complexity than the surrogate measure H(S) previously used in ecology. We present examples and elaborate on general implication. (C) 1996 Academic Press Limited.