MEPS 522:79-95 (2015)  -  DOI: https://doi.org/10.3354/meps11133

ABSTRACT: In the northern hemisphere, eelgrass Zostera marina L. is the most important and widespread seagrass species. Despite its ecological importance, baseline data on eelgrass distribution and abundance are mostly absent, particularly in subtidal areas with relatively turbid waters. Here, we report a combined approach of vegetation mapping in the Baltic Sea coupled to a species distribution model (SDM). Eelgrass cover was mapped continuously in the summers of 2010 and 2011 with an underwater towed camera along ~400 km of seafloor. Eelgrass populated 80% of the study region and occurred at water depths between 0.6 and 7.6 m at sheltered to moderately exposed coasts. Mean patch length was 128.6 m but was higher at sheltered locations, with a maximum of >2000 m. The video observations (n = 7824) were used as empiric input to the SDMs. Using generalized additive models, 3 predictor variables (depth, wave exposure, and slope), which were selected based on Akaike’s information criterion, were sufficient to predict eelgrass presence/absence. Along with a very good overall discriminative ability (area under the receiver-operating characteristic curve ROC/AUC = 0.82), depth (as a proxy for light), wave exposure, and slope contributed 66, 29, and 5%, respectively, to the final model. The estimated total areal extent of eelgrass in the study region amounts to 140.5 km² and comprises about 11.5% of all known Baltic seagrass beds. The present work is, to the best of our knowledge, the largest study undertaken to date on vegetation mapping and the first to assess distribution of eelgrass quantitatively in the western Baltic Sea.

KEY WORDS: Species distribution · GAM · Habitat mapping · Habitat modeling · Depth limit · Patchiness