Table 4. Constrained principal coordinate analyses (cPCoA).
Results from a sequence of constrained principal coordinate analyses (cPCoA), were the model terms were tested for significance using order dependent permutation based ANOVA. An alternative way to perform the cPCoA is to remove the effects of variables in the ordination model prior to the ANOVA, in this case location and species identity, using a condition, which makes it possible to test and visualize the independent effect of the remaining constraining variable, in this case pollution type. Each model is based on 354 Amplified Fragment Length Polymorphism (AFLP) loci, using different subsets of the data based on location (BP, Baltic Proper; ALL, Baltic Proper + West Coast; WC, West Coast (i.e., KRI)) and pollution type (STP, sewage treatment plant; REF, reference and HAR, harbor). The significant effects, after false discovery rate (FDR) correction, are indicated in bold (FDR = 0.05).
| Terms | ALL STP/REF P-value | BP STP/REF P-value | BPa STP/REF P-value | ALL HAR/REF P-value | BP HAR/REF P-value | BP HAR/REF P-value | WC STP/REF P-value | WC HAR/REF P-value |
|---|---|---|---|---|---|---|---|---|
| Location | 0.001 df = 4 | 0.030 df = 3 | Condition | 0.001 df = 4 | 0.067 df = 3 | Condition | – | – |
| Species identity | 0.690 df = 1 | 0.682 df = 1 | Condition | 0.938 df = 1 | 0.941 df = 1 | Condition | 0.699 df = 1 | 0.603 df = 1 |
| Pollution type | 0.011 df = 1 | 0.009 df = 1 | 0.016 df = 1 | 0.634 df = 1 | 0.942 df = 1 | 0.942 df = 1 | 0.823 df = 1 | 0.175 df = 1 |
| Species identity : pollution type | 0.466 df = 1 | 0.285 df = 1 | – | 0.246 df = 1 | 0.540 df = 1 | – | – | 0.530 df = 1 |
Notes.
Results from this model is plotted in Fig. 2.