Differences in performance of four ordination methods on a complex vegetation dataset (original) (raw)

Ordination is a widely used method in describing main relationships of multidimensional data. Properties of different ordination techniques have mostly been tested with simulated data. Although simulations provide valuable information about the behaviour of different methods, they are likely to be too simplistic to be able to completely predict the outcome with real data. We used post-fire vegetation succession data to compare four commonly used ordination techniques: CA (correspondence analysis), DCA (detrended correspondence analysis), PCoA (principal coordinates analysis), and NMDS (non-metric multidimensional scaling). Fire intensity was used as a methodindependent criterion for comparing the performance of the different methods. Solutions produced by these methods were compared using Procrustean analysis. According to our results, the compared ordination techniques presented different aspects of the multidimensional species space. In general, metric scaling methods, particularly CA and DCA, were far better in reflecting the main gradient in numerical terms, as compared with NMDS. In contrast, non-metric scaling out-performed metric scaling in graphical terms. We conclude that none of the compared methods is perfect in reflecting a complex vegetation gradient. Also, the difference in their abilities makes it difficult to point out the most suitable method for our data.

Sign up for access to the world's latest research.

checkGet notified about relevant papers

checkSave papers to use in your research

checkJoin the discussion with peers

checkTrack your impact