Spatial assessment of predator-prey relationships in the North Sea: the influence of abiotic habitat properties on the spatial overlap between 0-group cod and grey gurnard

• Published in: Fisheries oceanography 2013-05, Vol.22 (3), p.174-192
• Main Authors: Kempf, Alexander, Stelzenmüller, V., Akimova, A., Floeter, J.
• Format: Article
• Language: English
• Subjects: cod; Gadus morhua; generalized additive models; grey gurnard; Marine; regression kriging; spatial overlap; species distribution model
• ISSN: 1054-6006; 1365-2419
• DOI: 10.1111/fog.12013

The understanding of spatio‐temporal dynamics of marine ecosystems is crucial for ecosystem‐based fisheries management and climate change impact assessments. We quantified temporal changes in the distribution of 0‐group cod (Gadus morhua) and grey gurnard (Eutriglia gurnardus), a primary predator of 0‐group cod, with the help of regression kriging and assessed the temporal dynamics of the related spatial predator–prey overlap of these two species at different spatial scales. We analysed the robustness of relationships among abiotic habitat properties (temperature, salinity and depth) and abundance. Small cod was mainly found in low salinity areas of the Skagerrak but larger year classes were able to expand their distribution area towards the central and northern North Sea. In contrast, grey gurnard was mainly found in waters with salinities above 33 and temperatures above 14°C. This species has expanded its high density areas in the central North Sea northward over the last two decades. Recruitment success of cod was negatively correlated to a Moran's I cross‐correlation index, a proxy for the degree of spatial overlap between both species. Strong cod year classes overlapped less with grey gurnard at the large and medium spatial scale. In general, the relationships between abiotic habitat properties and abundance showed an increased inter‐annual variability, which was likely caused by underlying factors not taken into account in the distribution models. Thus assemblage modeling approaches combining the strength of different model types should be considered in the future to predict potential distribution patterns under climate change scenarios.