I have three sets of community matrices of different organisms collected at sampling sites that differ in ecological health (poor, fair, good, natural) along with a comprehensive set of environmental variables. Two of the matrices have measures of relative abundance (ranked as 1,2,3) and one actual counts. My aim is to show which species within each group are most linked to changes in environmental variables, or, in other words, which env variables most impact species within each group. However, I am struggling to determine which ordination methods are preferable as the resulting plots have distinct differences. In particular, the two gradient methods (Canonical Correspondence Analysis (CCA) and Redundancy Analysis (RDA) give quite different outcomes, why would this be? I presume its something to do with how abundance is taken into account? I also though about fitting environmental gradients to a Detrended Correspondence Analysis (DCA) or Non-metric Multidimensional Scaling (NMDS), they certainly show stronger grouping between ecological health categories but the gradients are less clear. I was hoping there might be people on here that would be able to advise which method is best according to the nature of the data and desired objectives?
- Badges
- Science topic