Semi-Automated Data Processing and Semi-Supervised Machine Learning for the Detection and Classification of Water-Column Fish Schools and Gas Seeps with a Multibeam Echosounder.

Multibeam echosounders are widely used for 3D bathymetric mapping, and increasingly for water column studies. However, they rapidly collect huge volumes of data, which poses a challenge for water column data processing that is often still manual and time-consuming, or affected by low efficiency and high false detection rates if automated. This research describes a comprehensive and reproducible workflow that improves efficiency and reliability of target detection and classification, by calculating metrics for target cross-sections using a commercial software before feeding into a feature-based semi-supervised machine learning framework. The method is tested with data collected from an uncalibrated multibeam echosounder around an offshore gas platform in the Adriatic Sea. It resulted in more-efficient target detection, and, although uncertainties regarding user labelled training data need to be underlined, an accuracy of 98% in target classification was reached by using a final pre-trained stacking ensemble model.

ORMEF: a Mediterranean database of exotic fish records.

The Mediterranean Sea is recognized today as the World's most invaded marine region, but observations of species occurrences remain scattered in the scientific literature and scarcely accessible. Here we introduce the ORMEF database: a first comprehensive and robust compilation of exotic fish observations recorded over more than a century in the Mediterranean. ORMEF consists today of 4015 geo-referenced occurrences from 20 Mediterranean Countries, extracted from 670 scientific published papers. We collated information on 188 fish taxa that are thus divided: 106 species entered through the Suez Canal; 25 species introduced by shipping, mariculture, aquarium release or by means of other human activities; 57 Atlantic species, whose arrival in the Mediterranean has been attributed to the unassisted immigration through the strait of Gibraltar. Each observation included in the ORMEF database was submitted to a severe quality control and checked for geographical and taxonomic biases. ORMEF is a new authoritative reference for Mediterranean bio-invasion research and a living archive to inform management strategies and policymakers in a period of rapid environmental transformation.

An integrated approach to assess fish spatial pattern around offshore gas platforms: A pilot study in the Adriatic Sea.

Research on abundance and composition of fish assemblages surrounding offshore extraction platforms is essential to evaluate their impact as well as to understand relationships between natural and artificial habitats. Also decommissioning practice, which belongs to the lifecycle of these structures, can be encouraged or discouraged if fish school behaviour in the close proximity of the platform is well understood. With thousands of platforms to be decommissioned around the world in coming decades, understanding such dynamic interactions is key to improve spatial management of marine ecosystems. In this context, this study drafts a work plan that can be used to investigate fish presence and abundance, school movement and qualitative species composition around a platform over long time periods. It integrates fishing captures, multibeam echosounder (MBES) investigations, and drop camera shootings to overcome the limitations of the individual methods. Monthly samplings were conducted at a three-leg gas extraction platform placed at ~80 m depth in the central Adriatic Sea, for one year after its installation. MBES completely insonified the studied area, providing data on school shape, volume, surface area and position throughout the water column. Fishing captures were useful for MBES targets' identification by measuring the presence/abundance of nekto-benthic and pelagic species both in the nearby of the structure and in the open sea, while drop camera shootings added evidence of a few species in close proximity to the poles, which were not censused by the other methods. Results underlined the strong attraction exerted by the platform and the significant influence of the explanatory variable distance on the schools' nominal volume.