The interpretation of process from pattern using two-dimensional spectral analysis: modelling single species patterns in vegetation

Two-dimensional spectral analyses of spatial patterns are made using (i) the autocorrelation function, (ii) the periodogram, and its polar summary (iii) the R-spectrum and (iv) the -spectrum. Together these give a sensitive analysis of both the complete range of scales of pattern and directional components which exist in data sets and we illustrate how the significance of observed spectral features can be assessed.We investigate the spatial pattern of Calluna vulgaris in a regenerating woodland and of Epilobium angustifolium spreading in a woodland following the thinning of trees. Evidence in the spectra is found for directional, clumping and inhibition patterns, and is discussed in relation to spectra obtained from simulations of known pattern generating processes. Hypotheses about the important biological, environmental and management influences on the structure of the communities are examined.We gratefully acknowledge the Swedish Coniferous Biome Project and Dr P. J. Diggle for access to the heather data.