The performance, application and integration of various seabed classification systems suitable for mapping Posidonia oceanica (L.) Delile meadows

Abstract

In the context of current global environmental changes, mapping and monitoring seagrass meadows have become highly important for management and preservation of coastal zone ecosystems. The purpose of this research was to determine the numerical precision of various cost-effective benthic habitat mapping techniques and their suitability for mapping and monitoring of Posidonia oceanica meadows in the Croatian Adriatic. We selected ultra-high resolution aerial imagery, single-beam echo sounder (SBES) seabed classification system from Quester Tangent Co. (QTC), and surface based underwater videography as affordable, non-destructive and simple to use systems for data acquisition. The ultra-high resolution digital imagery was capable of detecting P. oceanica meadows up to 4m depth with 94% accuracy, from 4m to 12.5m depth the accuracy dropped to app. 76%, and from 12.5 to 20m the system was only capable of distinguishing seabed biota from substrata, though with 97% accuracy. The results of the QTC system showed over 90% detection accuracy for Cymodocea nodosa covered seabed, excellent separation capabilities (>92%) of different sediment types (slightly gravelly sand, gravelly muddy sand and slightly gravelly muddy sand) and reasonable accuracy for mapping underwater vegetation regardless of the bathymetric span. The system proved incapable of separating P. oceanica from dense macroalgae on the same type of substratum. Surface-based underwater videography demonstrated great potential for estimating P. oceanica cover in a sampled region using either a single human rater or a computer estimate. The consistency between two human scorers in evaluating P. oceanica bottom coverage was near perfect (>98%) and high between digital and human scorers (80%). The results indicate that although the selected systems are suitable for mapping seagrasses, they all display limitations in either detection accuracy or spatial coverage, which leads to a conclusion that suitable system integration is essential for producing high quality seagrass spatial distribution maps.