Predictive models using randomForest regression for distribution patterns of meiofauna in Icelandic waters

The exchange of cold polar waters from the north and warm North Atlantic waters from the south is limited by the Greenland-Iceland-Scotland Ridge (GSR) which then may act as a barrier for faunal exchange from the deep Artic and deep Atlantic basins. We investigated the meiofauna density and distribution patterns from different regions north and south of the GSR at water depths between 307 and 2749 m. A total of 84 multicore samples were examined collected during the IceAGE1 (Icelandic marine Animals: Genetics and Ecology) project in summer 2011. We used a gradientForest approach to assess the magnitude of compositional change and the thresholds of remarkable community changes along environmental variables. randomForest regression was applied to predict the meiofauna on a spatial contiguous scale with a set of 23 environmental variables. Meiofauna density ranged between 187 and 3185 individuals per 10 cm 2 with the highest densities observed north of the GSR. Nematoda was the most abundant taxon among the meiofauna community, followed by Copepoda and Nauplii. In the Irminger Basin, Gastrotricha was the third most abundant taxon, while in all other regions, Ostracoda was the third most abundant. Food supply, water depth, bottom water oxygen, and hydrographic activity being the most important variables for community changes explain up to 86% of the variation observed in the meiofauna communities.