On the integrated interpretation of indirect site ordinations: a case study using semi-arid vegetation in southeastern Spain

The need for rigorous methods of interpreting indirect site ordinations is discussed and the problem of oblique habitat and phytosociological trends is emphasized. A sequential approach, termed integrated interpretation, is introduced to overcome this problem. It first involves a technique termed rotational correlation a linear form of trend surface analysis, which locates the best fit between a habitat variable and the coordinates of a pluridimensional ordination. The coordinates at positions of best fit are then used to produce two-way site-species tables (seriation arrays) which summarize floristic variation in relation to each major habitat trend. The seriation arrays can also be used to identify groups of differential species. Integrated interpretation is demonstrated using semi-arid vegetation from Murcia Province, SE Spain. A two-axis site ordination of vegetation data by non-metric multidimensional scaling is shown to have two oblique trends related to aspect-induced topoclimates and types of past anthropogenic disturbance. Rotation of the configuration to maximum linear correlation achieves very high levels of explanation for each factor. Rotated ordination coordinates from the correlation analysis are used to obtain two seriation arrays and floristic noda which relate directly to the two underlying habitat trends. A floristic discontinuity is revealed on the topoclimate continuum and it is hypothesized that temperature conditions are a major determinant of floristic composition, but that moisture partly determines vegetational cover and phytomass. Several patterns of past cultivation are detected and associated with five main kinds of geomorphological unit, together with some evidence that the rate of succession is partly determined by topoclimate. The integrated approach to interpretation is compared with visual methods and other multivariate techniques. A case is made for increased interpretational rigour in site coordinate studies, especially the discontinuation of trend-seeking using individual unrotated ordination axes.Nomenclature of vascular plants follows Tutin et al. (1964–1980) Flora Europaea 5 Vols. Cambridge University Press, Cambridge. Nomenclature of syntaxa follows Rivas-Goday & Rivas-Martinez (1967).I am indebted to J. A. Michel, Dr M. A. Abdelrahman and Monica M. Dargie for field assistance. Field discussion with Dr R. L. Wright, Dr M. P. Austin and Professor F. Esteve Chueca was most helpful; the latter assisted with confirmation of vascular plant voucher specimens. Non-vascular taxonomy was determined by Dr C. Humphries (mosses and one alga) and Dr O. L. Gilbert (lichens). An anonymous referee suggested an approximation of the analytical solution to rotational correlation, and Dr N. R. J. Fieller assisted with its confirmation. The project was assisted by a grant from the University of Sheffield Research Fund.