In addition, the accuracy is assessed to identify the reliability of the maps. The data in the service user module are not directly related to the service provider module and can be modified in accordance with the aims of the study (i.e. feeding grounds of a single fish species). This study was carried in the Lithuanian Exclusive Economic Zone (ICES subdivision 26), south-eastern Baltic Sea. Of the available environmental predictors known to be important for the distribution

of macrozoobenthos (Olenin, 1997, Bučas et al., 2009 and Gogina and Zettler, 2010, Reiss et al. 2011), eight were selected for the modelling of prey biomass: salinity, minimum near-bottom oxygen concentration, near-bottom current velocity, wave generated orbital near-bottom velocity, depth, sediment types, areas with the presence or absence of the thermocline PD0325901 datasheet and the areas above and below the halocline. Quantitative environmental parameters were tested for collinearity and predictors

were removed from models if variance inflation factors (VIF) were > 3 (Quinn & Keough 2002). Depth was highly collinear with the wave-generated near-bottom orbital velocity, near-bottom oxygen concentration and salinity. These three predictors are direct environmental factors for the distribution of macrofauna, whereas depth is a cumulative and indirect effect of them (McArthur et al. 2010) and was therefore omitted. The layer of sediments was derived from geological charts (Repečka et al., 1997, Gelumbauskaitė et al., 1999 and Bitinas et al., 2004). Sediments were classified into four types: boulders, cobbles/gravel, sand and silt (Wentworth 1922). The wind wave orbital velocity data layer Panobinostat nmr isothipendyl was derived using the SWAN model (Booij et al. 1999) based on 2008–2009 wind data. National marine monitoring

data was used to derive the salinity and thermocline/halocline layers (MRC, unpublished: 2003–2008 and 1998–2006 datasets accordingly). Minimum near-bottom oxygen concentrations (2000–2006) and annual mean bottom current velocity layers were obtained from datasets produced by the BALANCE project (Hansen et al., 2007 and Bendtsen et al., 2007). Data on the feeding habits of Baltic cod, flounder and viviparous eelpout of different body length were collected in the spring-autumn seasons of 2000–2010 during quarterly trawl surveys. Stomach contents were analysed by standard numerical and gravimetric methods (Hyslop 1980). To assess the diet composition of fish 1425 digestive tracts were analysed (empty tracts excluded): 300 digestive tracts of Atlantic cod (from 39 to 80 cm in size); 1000 digestive tracts of flounder (the size ranged from 15 to 40 cm); 125 digestive tracts of eelpout (sizes from 25 to 30 cm). Food items were identified to the lowest possible taxonomic level. In total, data from 640 benthic samples taken at 224 sampling sites during 1998–2010 were used to model the biomass distribution of the macrozoobenthos (Figure 2).