All comparisons of disturbed and undisturbed mangroves were based on the dry season survey of 2001. The sampling effort for that survey was double that for surveys in 2003–2005, as paired transects were sampled at each location. ANOVA for species richness, abundance and feeding guild data involved 4-factor, nested analyses with the factors disturbance, location, transect and assemblage, in which all factors were fixed except transect, which was random and nested in location and disturbance.