Table 2. Permutational analysis of variance (PERMANOVA) results of benthic community structure (236 species).
Analysis based on fourth root transformed abundances and Bray-Curtis dissimilarities, plus live coral, macroalgae and cyanobacteria based on square root transformation of % cover and Euclidian dissimilarities.
| Factors | d.f. | Component | Benthic community structure | Live coral cover | Macroalgae cover | Cyanobacteria cover | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Pseudo F | P (perm) | Pseudo F | P (perm) | Pseudo F | P (perm) | Pseudo F | P (perm) | |||
| Distance | 1.38 | Fixed | 3.9697 | 0.0001 | 0.84515 | 0.3712 | 15.504 | 0.0006 | 1.0293 | 0.348 |
| Rugosity Index | 3.36 | Fixed | 0.9554 | 0.577 | 1.2897 | 0.2888 | 2.5808 | 0.0664 | 1.6139 | 0.1776 |
| Recreation | 4.35 | Fixed | 3.5968 | 0.0001 | 2.6421 | 0.0434 | 7.4505 | 0.0004 | 0.99765 | 0.3739 |
| Diadema Index | 3.36 | Fixed | 1.644 | 0.0024 | 0.30943 | 0.8311 | 0.94652 | 0.4264 | 1.5125 | 0.2099 |
| Location (Distance) | 6.32 | Random | 3.368 | 0.0001 | 8.0592 | 0.0002 | 3.4613 | 0.0105 | 2.3539 | 0.0446 |
| Rug × Loc (Distance) | 18.20 | Random | 1.6768 | 0.0001 | 3.215 | 0.0113 | 1.9026 | 0.0967 | 1.7082 | 0.1825 |
| Rec × Loc (Distance) | 6.32 | Random | 3.368 | 0.0001 | 8.0592 | 0.0001 | 3.4613 | 0.0094 | 2.3539 | 0.0435 |
| Dia × Loc (Distance) | 16.22 | Random | 2.0492 | 0.0001 | 3.282 | 0.0097 | 1.274 | 0.2964 | 1.486 | 0.2559 |
| Distance × Diadema | 5.34 | Fixed | 2.0386 | 0.0001 | 0.35696 | 0.8901 | 4.0011 | 0.0058 | 0.9672 | 0.4464 |
| Distance × Recreation | 5.34 | Fixed | 3.6386 | 0.0001 | 8.5719 | 0.0001 | 7.5387 | 0.0002 | 1.006 | 0.3967 |
| Distance × Rugosity | 5.34 | Fixed | 1.5953 | 0.0008 | 0.7612 | 0.5883 | 4.292 | 0.0038 | 1.2102 | 0.2759 |
| Rugosity × Diadema | 10.29 | Fixed | 1.2218 | 0.0376 | 1.3486 | 0.245 | 1.6478 | 0.1333 | 1.4517 | 0.226 |
Note:
Pseudo F statistics were calculated for each term using direct analogs to univariate expectations of mean squares. A conservative Type III sums of squares was used. P-values were obtained using 9999 permutations under a reduced model, with each term contributing either a fixed or random component to the overall model. P-values in bold are below 0.05.