did not result in a pattern indicative of trimming effects (Table 1).
Likewise, when examined separately, no effect of trimming was
identified on the density or biomass of benthic foraging fishes (H2;
Table 1), roving piscivorous fishes (H3; Table 1), or water column
foraging fishes (H4; Table 1). In no case did we record a loss or addition
of species from the community after trimming. Retrospective
power analysis indicated that the ANOVAs performed to test H1–4
suffered from statistical power being below the desired level of 0.80
(Table 1).
Fig. 2. Themeandifference in fish density (a) and biomass (b) between paired (treatment
and control) sampling plots (December 1998–November 2000). The vertical
line represent the division between the pre- and post-trimming periods.
Fig. 3. Nonmetric multi-dimensional scaling output based on Bray–Curtis similarities
in site-specific yearly mean densities of fish. Each point represents a site: the
dark circles signify the experimental plots after trimming had been performed, triangles
represent control plots, and the squares are the experimental plots prior to
trimming.
Community-level analyses were generally supportive of findings
described above for subsets of species. That is, we were also
unable to detect a change in fish species composition that could
be attributed to mangrove trimming (H5). Nonmetric multidimensional
scaling did not indicate that the species composition of the
trimmed plots was readily differentiated from the control or pretrimming
plots (Fig. 3; ANOSIM p-values = 0.80–0.99). Similarly,
two-way repeated measures analysis of the Bray–Curtis study plot
similarities provided no indication that mangrove trimming had
impacted the species composition of the fish community (Table 1).