2.4. Statistical analysis
One-way ANOVA were used to compare tree diversity, forest
cover, ocypodid crab density and the ratio of ocypodid crabs to
grapsid crabs across treatments as it was possible for these data to
be fit to a normal distribution. Tukey's post-hoc tests were conducted
on these ANOVA results to show pair-wise comparisons of
the different treatments. General linear models with a Poisson
distribution were used to compare total mudskipper density,
Boleophthalmus spp. density, Periophthalmodon spp. density, and
plant diversity across treatments as these data most closely fitted to
a Poisson distribution. Pair-wise comparisons of the different
treatments were also done for these data using Tukey's post-hoc
tests. KruskaleWallis analyses of variance were used to compare
the density of large and small crab holes, gastropod density, grapsid
crab density and tree and plant density across treatments, as these
data did not conform to any particular distribution. To make pairwise
comparisons for these data Wilcoxon RankeSum Tests were
used. Linear regressions were used to compare the relationships
between fauna abundances and forest cover and tree diversity. For
the mudskippers of the genera Apocryptodon, Periophthalmodon,
and Oxuderces insufficient data were obtained for statistical analysis.
All statistical analysis was conducted using R statistical software
version 3.0.1 (R Core Team, 2013).