Using multivariate analysis to assess effects of fluvial type on plant species distribution in a Mediterranean river

We propose a method to assess the distribution of plant assemblages along rivers delimiting homogeneous fluvial types using cluster analysis applied to morphological features, quantified with aerial photos and geographic information system software. A stratified random sampling design along the elevational gradient was used to analyze riparian plant species. Multivariate statistics were applied to detect patterns of variation in the species data, and among functional and ecological groups. Cluster analysis identified four fluvial types: headwater, low sinuosity, braided, and incised. Canonical Correspondence Analysis, INdicator SPeciesANalysis, and partial ordinations all suggested that fluvial types were characterized by well-defined indicator species. The differences found by Redundancy Analysis and nonparametric analysis of variance for functional and ecological groups also supported the fluvial type division, showing different distributional trends for annual and woody species, and the greatest ecological distance between braided and headwater types. Based on the ordination results, dividing fluvial types according tomorphological features was justified by environmental and floristic differences, although plant species variability was only partially described. Our results illustrated that the fluvial type classification created using this methodology was consistent with natural plant species distribution patterns.