MEPS 476:285-299 (2013)  -  DOI: https://doi.org/10.3354/meps10156

ABSTRACT: Turbid water conditions make the delineation and characterization of benthic habitats difficult by traditional in situ and remote sensing methods. Here, we develop and validate modeling and sampling methodology for detecting and characterizing seagrass beds by analyzing GPS telemetry records from radio-tagged manatees. Between October 2002 and October 2005, 14 manatees were tracked in the Ten Thousand Islands (TTI) in southwest Florida (USA) using Global Positioning System (GPS) tags. High density manatee use areas were found to occur off each island facing the open, nearshore waters of the Gulf of Mexico. We implemented a spatially stratified random sampling plan and used a camera-based sampling technique to observe and record bottom observations of seagrass and macroalgae presence and abundance. Five species of seagrass were identified in our study area: Halodule wrightii, Thalassia testudinum, Syringodium filiforme, Halophila engelmannii, and Halophila decipiens. A Bayesian model was developed to choose and parameterize a spatial process function that would describe the observed patterns of seagrass and macroalgae. The seagrasses were found in depths <2 m and in the higher manatee use strata, whereas macroalgae was found at moderate densities at all sampled depths and manatee use strata. The manatee spatial data showed a strong association with seagrass beds, a relationship that increased seagrass sampling efficiency. Our camera-based field sampling proved to be effective for assessing seagrass density and spatial coverage under turbid water conditions, and would be an effective monitoring tool to detect changes in seagrass beds.

KEY WORDS: Zero-inflated negative binomial · ZINB · Kernel density filter · Geographic information system · GIS · ARGOS satellite · Foraging ecology · Spatial pattern