Canonical Ordinations: Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA)
This Eco-Tool implements two forms of canonical ordination: Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA). The main input consists of a matrix where rows represent sites and columns contain either dependent variables (typically observations of species abundances) or candidate predictor variables (measurements of some environmental quality at each site). A second input is the number of predictor variables; this is used to divide the input matrix into its two components. A third input is the number of randomizations used to test the significance of the dependent variables in determining the observations. Optional inputs are row and column labels.
An example of an input matrix is given below. This is the made-up tropical reef dataset used in Numerical Analysis (Legendre and Legendre 1998, p590).
| 1 | 2 | 3 | 4 | 5 | 6 | Depth | Coral | Sand | Other | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 0 |
| 2 | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 0 | 1 | 0 |
| 3 | 0 | 1 | 0 | 0 | 0 | 0 | 3 | 0 | 1 | 0 |
| 4 | 11 | 4 | 0 | 0 | 8 | 1 | 4 | 0 | 0 | 1 |
| 5 | 11 | 5 | 17 | 7 | 0 | 0 | 5 | 1 | 0 | 0 |
| 6 | 9 | 6 | 0 | 0 | 6 | 2 | 6 | 0 | 0 | 1 |
| 7 | 9 | 7 | 13 | 10 | 0 | 0 | 7 | 1 | 0 | 0 |
| 8 | 7 | 8 | 0 | 0 | 4 | 3 | 8 | 0 | 0 | 1 |
| 9 | 7 | 9 | 10 | 13 | 0 | 0 | 9 | 1 | 0 | 0 |
| 10 | 5 | 10 | 0 | 0 | 2 | 4 | 10 | 0 | 0 | 1 |
There are ten sites, six species, and four predictor variables. Note that all predictors except Depth are numerical codings for a single qualitative variable one might call Substrate Type. Because these categories are exhaustive (meaning that every site must be in one of the categories), then using all three of the columns in the analysis is redundant (and will lead to numerical problems in the calculations). One option is to simply remove one of the columns, but this has the disadvantage that it is no longer possible to calculate the correlation of that environmental variable with any of the ordination axes. So, in this Eco-Tool we implement the option to keep redundant variables in the data, and simply specify which columns they are.