Preferred prey items for flounder and eelpout were gammarideans and bivalves Macoma balthica, while priapulids Halicryptus spinulosus and soft-shell clams Mya arenaria were eaten only by flounder. Flounder had the most diverse diet composition (a total of eight prey items), while eelpout and cod preyed upon six and four prey items respectively.

Half of the prey items were eaten by all three species, while two items (H. spinulosus and M. arenaria) selleck products were exclusively fed on by flounder. Different weights were assigned to every fish species separately according to the occurrence and importance of prey items ( Table 3). According to the coefficient of variation of mean absolute deviation (Table 4) the most accurate model was obtained for blue mussel M. edulis (16%). Models of S. entomon, Gammaridea, H.

spinulosus and M. arenaria were also relatively accurate (< 50%). The model of M. balthica was less accurate (61%), and the accuracy was the lowest for both polychaete models (> 70%). The mean decrease accuracy (%IncMSE) was calculated for each predictor in order to evaluate its importance to the response variable (Table 5). The most important predictor was near-bottom oxygen concentration especially for deep-living species like M. balthica, S. entomon and H. spinulosus (28.7, 12.1 and 24.6 %IncMSE respectively). BMS 354825 Orbital velocity, salinity and sediments were also important: the biomasses of amphipods M. edulis were mostly dependent on sediments (9.3 and 34.8 %IncMSE respectively), while salinity had a major influence on both polychaete worms and M. balthica, and orbital velocity on H. spinulosus 5-Fluoracil molecular weight and S. entomon (12.7 and 18.9 respectively). Near-bottom current velocity was less important, while the halocline and thermocline were only of minor importance or of no importance at all in some cases. The map of seabed quality for the feeding of cod, flounder and eelpout is presented in Figure 3. The highest quality feeding grounds for all three fish species is the stony bottom in the coastal area situated in the northernmost part of LEZ. Other high quality

areas are located in the offshore zone: one in an offshore bank with heterogeneous sediments at 50 m depth (western part of LEZ), another in the soft bottom at 40–50 m depths (central part of LEZ). The accuracy assessment indicates that the most accurate areas of the approach are at 10–40 m depths. The low accuracy areas were justified by only 18% of total samples and were set in very shallow areas (down to 3 m depth) and for the deepest areas. Accuracy was moderate for offshore areas in the central part of LEZ and for the coastal area. More than half the samples were taken in the coastal area, but because of the rapid changes in some environmental parameters (especially salinity and near-bottom orbital velocity) the quartiles of these predictors were only moderately justified in terms of accuracy.