Mapping reef forming North Sea species

The aim of this project was to predict as accurately as possible where four reef-building species in the North Sea: Sabellaria spinulosa, Modiolus modiolus, Lanice conchilega and Ostrea edulis can develop stable populations, given the environmental gradients and the gradients in human use of the North Sea.

All four of these species can form biogenic reefs, which in turn can form hotspots for biodiversity as they provide attachment or hiding opportunities for other species. Biogenic reefs are recognized as habitats worthy of special protection in OSPAR (Convention for the Protection of the Marine Environment of the North-East Atlantic) and EU regulations. That applies in particular to Sabellaria spinulosa and Modiolus modiolus, where trawling is considered to endanger the special habitats created by the reefs. Lanice conchilega is a very common species that does not appear to be particularly threatened, whereas Ostrea edulis is practically extinct in the North Sea, but is currently the subject of intense restoration efforts.

The results are shown in several habitat suitability maps (link to the project map). Herman & van Rees (2021)  documents the compilation of the habitat suitability maps.

Methods, results, discussion

For the three species that currently have stable populations in the North Sea (Sabellaria, Modiolus and Lanice), spatial distribution data were derived from EMODnet Biology. In addition, for Sabellaria and Modiolus data of macro-benthic bycatch in fishing trawls of Wageningen Marine Research were used. Lanice was too sparsely distributed in these data to be considered reliable. For Ostrea edulis, historic data on the distribution in the nineteenth and early twentieth century in Dutch, German and Belgian waters was used from literature sources. Environmental data were derived from a recent compilation in the literature, and in addition from Deltares physical modelling.

The presence/absence of the species was regressed on the environmental data using two regression models: logistic regression and random forest regression. The results of both models were in close agreement for all species. The random forest models gave the finest-grained predictions and are taken as the final product of the project.

In the report the R code used for collecting all data, preparing GIS files of the data and performing the regression analyses is discussed. An accompanying QGIS project and all its underlying files, as well as a geo-pdf file that contains all the environmental layers, species occurrence data and regression predictions are available here.

The project was part of the MONS research programme, executed by Deltares on demand of Rijkswaterstaat.

MONS project id: 55