ISBN 82-7420-006-3. ISSN 0802-8478.
This textbook presents the conceptual basis for modern vegetation science. The gradient nature of vegetation and the environment is emphasized. Variation in species abundance along environmental complex-gradients is described, and species response models discussed. Advantages of using gradients scaled in units of compositional turnover for the study of species resposes are given. The classical continuum controversy in vegetation ecology is reviewed. The niche concept is also reviewed, with reference to gradient theory. The theoretical part concludes with an overview of processes structuring vegetation (interspecific interactions, destabilizing factors, stress and chance), and their relevance to major Fennoscandian ecosystems are discussed.
Major approaches to description of Fennoscandian vegetation, the Braun-Blanquet approach and the Northern traditions (the Uppsala school of phytosociology and the Finnish site-type approach), are reviewed. The applicability of phytosociological methods in a multidimensional continuum is discussed.
Methods in vegetation ecology are reviewed, with particular emphasis on techniques for gradient analysis. The phases of vegetation ecological studies are treated in the order: sampling, pre-processing of data (data manipulation), and gradient analysis. Sampling designs are reviewed and discussed with emphasis on suitability for different research purposes. The conflict between homogeneity and representativity of plots and its consequences for plot size is discussed. Species abundances measures are compared. Matrix models in vegetation ecology are described. The main types of data manipulation operations for vegetation data, weighting and standardization, are explained. Floristic dissimilarity measures are reviewed, and the conceptual shortcomings of all floristic dissimilarity measures as measures of ecological distance is emphasized.
Gradient analysis includes four types of techniques that assist interpretation of species-gradient relationships: regression and calibration (univariate approaches; one response variable), and ordination and constrained ordination (multivariate techniques; many response variables). The roles of these four types in gradient analysis are explained, with particular emphasis on ordination. The rationale of ordination is explained statistically and geometrically. Ordination methods that can be interpreted statistically may be explained as a process of recurrent regressions and calibrations. The most important ordination techniqies, PCA, CA, DCA and LNMDS are described in detail, their faults explained and their relative suitability for use with data sets with specified properties thoroughly discussed. No ordination method exists that never produces distorted or artifactual axes, but guidelines are given for the identification of such axes. Constrained ordination is explained and the relationship of this family of techniques to ordination briefly commented on.
Numerical classification techniques are briefly discussed in the concluding chapter.
Rune Halvorsen Økland, Botanical Garden and Museum, University of Oslo, Trondheimsveien 23B, N-0562 Oslo 5, Norway.