Discovering the ecological structure of different macrophyte groups in rivers using non-parametric and parametric multivariate ordination techniques.

This paper analyses various methods of ecological ordering that are often used in modelling the relationship between vegetation and habitat. The results of direct gradient ordination by Canonical correspondence analysis (CCA), which is based on correlation, were compared with Non-metric multidimensional scaling (NMDS), which is based on rank analyses. Both tools were also compared with Detrended correspondence analysis (DCA), which is a popular indirect gradient analysis method. The macrophyte assessment was conducted at 98 river locations in the lowland regions of Poland. Each of the surveyed locations falls within a consistent abiotic category: small to medium-sized lowland rivers with a sandy bottom. Habitat elements analysed included limnological variables and geographic parameters, and the botanical survey focused on submerged macrophytes, including vascular plants, as well as bryophytes and algae. Firstly, it was shown that various analytical tools for determining the importance of ecological factors (Monte Carlo test, BIOENV) identify slightly different significant factors responsible for the development of macrophytes in rivers. Secondly, considerable similarity was found in the structure of macrophyte communities generated on NMDS and DCA biplots, while macrophyte communities were presented very differently based on CCA. Thirdly, the ecological preferences of aquatic plants based on one-dimensional analyses primarily reflected the results of CCA, whereas they did not always follow the ecological pattern revealed by NMDS. Finally, by conducting separate studies for non-vascular plants and vascular macrophytes, it was confirmed that different ecological drivers are responsible for the development of particular groups of macrophytes.