Five decades of stomach content data allowed insight into the development of consumption, diet composition, and resulting somatic growth of Gadus morhua (Atlantic cod) in the eastern Baltic Sea. We show a recent reversal in feeding level over body length. Present feeding levels of small cod indicate severe growth limitation and increased starvation-related mortality. For young cod, the low growth rate and the high mortality rate are manifested through a reduction in size-at-age. The low feeding levels are likely the result of a decrease in benthic prey abundance due to increased hypoxic areas, while decreasing abundances of pelagic species in the area of cod distribution have prevented a compensatory shift in diet. Our study emphasizes that environmental forcing and the decline in pelagic prey caused changes in consumption and growth rates of small cod. The food reduction is amplified by stunted growth leading to high densities of cod of smaller size competing for the scarce resources. The average growth rate is negative, and only individuals with feeding levels well above average will survive, though growing slowly. These results suggest that the relation between consumption rate, somatic growth and predatorprey population densities is strongly environmentally mediated.

Different ecosystem models often provide contrasting predictions (model uncertainty), which is perceived to be a major challenge impeding their use to support ecosystem-based fisheries management (EBFM). The focus of this manuscript is to examine the extent of model disagreements which could impact management advice for EBFM in the central Baltic Sea. We compare how much three models (EwE, Gadget and a multispecies stock production model) differ in 1) their estimates of fishing mortality rates (Fs) satisfying alternative hypothetical management scenario objectives and 2) the outcomes of those scenarios in terms of performance indicators (spawning stock biomasses, catches, profits). Uncertainty in future environmental conditions affecting fish was taken into account by considering two seal population growth scenarios and two nutrient load scenarios. Differences in the development of the stocks, yields and profits exist among the models but the general patterns are also sufficiently similar to appear promising in the context of strategic fishery advice. Thus, we suggest that disagreements among the ecosystem models will not impede their use for providing strategic advice on how to reach management objectives that go beyond the traditional maximum yield targets and for informing on the potential consequences of pursuing such objectives. This is especially true for scenarios aiming at exploiting forage fish sprat and herring, for which the agreement was the largest among our models. However, the quantitative response to altering fishing pressure differed among models. This was due to the diverse environmental covariates and the different number of trophic relationships and their functional forms considered in the models. This suggests that ecosystem models can be used to provide quantitative advice only after more targeted research is conducted to gain a deeper understanding into the relationship between trophic links and fish population dynamics in the Baltic Sea.

No abstract

A B S T R A C TLike other fisheries models, multispecies models are subject to various sources of error. However, with regard to their use for ecosystem-based fisheries management (EBFM) between model errors are likely to be most important. As multispecies models are by definition many-dimensional, comparing them is potentially a complex task. The paper uses a simple approach. This is to calculate the Jacobian matrix of long term steady state catch by species with respect to the fishing mortality relative to status quo levels on all species. This enables the comparison of the relative strength of species interactions among models both within and between regions. This Jacobian matrix approach to comparing multispecies models is applied to available models for the North Sea, the Baltic Sea and from Icelandic waters. Moreover, this information is used to provide the basis for estimating a multidimensional quadratic yield surface for each model in the near field. Used this way it is possible to compare different model estimations of fishing mortality rate changes needed to approach yield-related management goals. The results suggest considerable variation between models in their detailed results but more coherence in suggesting directions for changing fishing mortality rate. Thus the approach is of considerable importance in specifying the confidence with which it is possible to make multispecies predictions for EBFM.

No abstract