Fine‐scale spatial pattern analysis of the herb layer of woodland vegetation using information theory

Abstract

The scale dependence of vegetation patterns and processes is generally accepted, but few long‐term studies have directly and explicitly considered scaling aspects in their sampling designs. In order to optimise data collection, particular emphasis must be given to the spatial scales at which maximum diversity, maximum heterogeneity, and maximum spatial dependence appear. The spatial pattern of the herb layer of woodland vegetation (old Quercus cerris coppice) in the central Apennines (Italy) was studied using information theory models. Inside a permanent plot, ground layers of vegetation were sampled along three circular transects. Data were analysed along a series of scales between 0.1 and 60 m. Detected patterns were tested against neutral models generated by Monte‐Carlo simulations. The maximum diversity of species combinations was found between 1 and 4 m, while the maximum heterogeneity and the highest degree of spatial dependence of populations appeared around 10 m. Complete randomisation detected much smaller scales for randomised communities with the same abundances of species, suggesting significant aggregations of individuals. However, no significant associations between patches of species were found by applying random shift algorithms. Our results suggest that vegetation is a multiscale phenomenon i.e. there is no single optimum plot size for permanent plot studies.

Key words:

• Aggregation
• association
• compositional diversity
• monitoring
• multi‐species patterns
• Quercus cerris community
• randomisation tests
• spatial scaling
• central Apennine