Skip to main content
No AccessArticles

Benthic diatom assemblages as indicators of water quality in the Everglades and three tropical karstic wetlands

Department of Biological Sciences, Florida International University, Miami, Florida 33199 USA

Limestone-based (karstic) freshwater wetlands of the Everglades, Belize, Mexico, and Jamaica are distinctive in having a high biomass of CaCO3-rich periphyton mats. Diatoms are common components of these mats and show predictable responses to environmental variation, making them good candidates for assessing nutrient enrichment in these naturally ultraoligotrophic wetlands. However, aside from in the Everglades of southern Florida, very little research has been done to document the diatoms and their environmental preferences in karstic Caribbean wetlands, which are increasingly threatened by eutrophication. We identified diatoms in periphyton mats collected during wet and dry periods from the Everglades and similar freshwater karstic wetlands in Belize, Mexico, and Jamaica. We compared diatom assemblage composition and diversity among locations and periods, and the effect of the limiting nutrient, P, on species composition among locations. We used periphyton-mat total P (TP) as a metric of availability. A total of 176 diatom species in 45 genera were recorded from the 4 locations. Twenty-three of these species, including 9 that are considered indicative of Everglades diatom flora, were found in all 4 locations. In Everglades and Caribbean sites, we identified assemblages and indicator species associated with low and high periphyton-mat TP and calculated TP optima and tolerances for each indicator species. TP optima and tolerances of indicator species differed between the Everglades and the Caribbean, but weighted averaging models predicted periphyton-mat TP concentrations from diatom assemblages at Everglades (R2  =  0.56) and Caribbean (R2  =  0.85) locations. These results show that diatoms can be effective indicators of water quality in karstic wetlands of the Caribbean, but application of regionally generated transfer functions to distant sites provides less reliable estimates than locally developed functions.