Vegetatio 68: 37-42, 1986 37 © Dr W. Junk Publishers, Dordrecht - Printed in the Netherlands Multi-species pattern analysis of Mediterranean pastures in three stages of ecological succession* Isabel Castro, Agustina Sterling & Eladio E Galiano Departamento de Ecologfa, Universidad Autdnoma de Madrid, 28049 Madrid, Spain Keywords: Grassland, Mediterranean pasture, Pattern analysis, Principal Components Analysis, Succession Abstract Multi-species patterns were analyzed in three stages of ecological succession in oligotrophic pastures in Central Spain using a new method of analysis. The method is based on plotting the species coordinates of principal components analysis undertaken at different block sizes. It allows the detection of any heterogenei- ty present and reveals both microtopographical and macrotopographical trends in species patterns. Introduction A plant community can be considered regarding two complementary aspects, floristic composition and structural variables. These variables, which in- clude alpha and beta diversity and the spatial distri- bution of species, may provide information con- cerning the nature of those processes which act within the community at specific spatial scales. For example: - many environmental factors present small-scale variations that govern species distribution over short distances; - the small scale pattern of vegetation provides information concerning the dispersion and vigour of some species. From the pioneer work of Greig-Smith (1952), Clark & Evans (1954) and Pielou (1959), a wide va- riety of methods have been devised to study spatial patterns in vegetation. See Greig-Smith (1979), Cormack (1979) and Ludwig (1979) for reviews both on the concept of pattern in ecology and the methods used for its detection. Most studies deal with species individually or in relation to some environmental factors, (see, for in- stance, Greig-Smith, 1964; Galiano, 1982a, b; Ford *Nomenclature follows Tutin et al., 1964-1980, Flora Eu- ropaea. & Renshaw, 1984; Renshaw & Ford, 1984). Others have investigated multi-species spatial patterns: Noy-Meir & Anderson (1970) and Walker et al. (1972) applied Principal Components Analysis (PCA) to covariance matrices obtained for differ- ent block sizes. Goodman (1978), Whittaker et al. (1979), Shmida & Whittaker (1981), Bouxin & Gau- tier (1982), Bouxin (1983), Bouxin & Le Bouleng6 (1983) and Galiano (1983) combined Correspon- dence Analysis (Reciprocal Averaging, R.A., Hill, 1973) with techniques of pattern analysis. The present paper presents a method of multi- species pattern analysis which consists of plotting the scores of species in PCA's carried out at differ- ent block sizes against block size. This method was applied to successional data from oligotrophic Mediterranean pastures in Central Spain. The process of ecological succession in these pastures has been described by Pineda et al. (1981) and is characterized by drastic changes in the floris- tic composition of the plant communities with time and by changes in the spatial structure (De Pablo et al., 1982; Sterling et al., 1984; Galiano et al., 1984). The problem was how different scales in an infor- mation analysis may offer different views on com- munity structure in a given successional phase, and how changes in pattern of those species present in different stages can be investigated.