MEPS 460:35-47 (2012)  -  DOI: https://doi.org/10.3354/meps09779

ABSTRACT: Spatially explicit ecological modelling was used to predict the distribution of 4 benthic components (live coral, carbonate sand, macroalgae and dead coral) inside the Aldabra lagoon, southern Seychelles, western Indian Ocean. Both classic ordinary least-squares and spatial autoregression techniques were carried out on a field data set of 774 spatially referenced records and 3 satellite remote sensing images to define an empirical relationship between local environmental conditions (water depth and water level variation) and benthic cover. This relationship was then used to generate a synoptic model of the spatial cover and distribution of each benthic component at the landscape (i.e. whole lagoon) scale. Environmental conditions were estimated from satellite remote sensing data (water depth) and using GIS techniques (water level variation). By drawing on species−environment relationships applicable to many lagoons, continuous records of percentage benthic cover were derived for the extensive lagoon (174 km²) at a high measurement level (ratio) for use in conservation and resource management applications. The transition from the ordinary least-squares model to the spatially lagged model was accompanied by a marked growth in predictive power (R² = 0.25 to 0.79), indicating that neighbourhood context interactions play an important role in determining benthic cover of the Aldabra lagoon.

KEY WORDS: Spatial regression · Coral reef · Seychelles · Atoll · Autoregression · Autocorrelation