Table 1. Effects of maintenance in relation to the exposure and location of the assemblages on the breakwaters.
| Source | df | MS | Pseudo-F | Pair-wise tests |
| Date = D | 3 | 24310 | 13.49*** | |
| Treatment = T | 1 | 68385 | 5.52** | |
| Exposure = E | 1 | 69816 | 6.49** | |
| Breakwater (T) = B (T) | 6 | 5415 | 3.01*** | D x T |
| D x T | 3 | 7912 | 4.39*** | May02: M ≠ C, Aug02: M ≠ C, |
| D x E | 3 | 7223 | 3.02** | Jan03: M = C, May03: M ≠ C |
| T x E | 1 | 21632 | 3.98** | T x E |
| D x B (T)a | 16 | 1802 | 1.93*** | Landward: M ≠ C, Seaward: M = C |
| E x B (T) | 6 | 4408 | 1.84* | |
| D x T x E | 3 | 1925 | 0.80 | |
| D x E x B (T)a | 16 | 2392 | 2.56** | |
| Residual | 420 | 934 |
The analysis is a PERMANOVA (17 variables, 4th-root transformed data) comparing assemblages between maintained ( = M) and non-maintained, control ( = C) breakwaters (Treatment, fixed factor), among dates of sampling (May and August 2002, January and May 2003, random factor; in May 2003, two control breakwaters were missing), between landward and seaward sides (Exposure, fixed factor) and among Breakwaters (4 levels, random factor nested in Treatment). a Term has one or more empty cells. * = P<0.05; ** = P<0.01, *** = P<0.001.