Such maps provide a different view of the spatial distribution of valuable seabed

areas as they do not necessarily coincide with the high catch areas of selected fish species. It is known that it can take more than 30 hours for prey to be digested (Macdonald et al. 1982), depending on the size of both predator and prey (Santos and Jobling, 1991 and Bromley, 1994) as well as on water temperature (Tyler 1970). Furthermore, the sustained speed of cod can reach 0.6–0.9 BL s− 1 (He, 1991 and Björnsson, 1993), meaning that 60 cm cod can swim for 38–58 km before their prey are digested. This shows that high catch areas of mobile fish whose stomachs are filled with benthic invertebrates do not necessary correspond to the good quality of the seabed, for there is no proof that the fish were caught in an actual feeding ground. Certainly, this is not the case with low selleck products mobility species like flounder and eelpout. On the other hand, these maps do not evaluate

the suitability of a given environment for fish species apart from the biomass distribution of prey items and their importance to the diet. It may happen that a prey biomass is very high but the fish has limited access to this environment or the environment may be unsuitable in the context of factors other than feeding. For instance, the eelpout is exclusively associated with coastal hard bottoms, so other areas (even of the highest quality) are irrelevant to this species. Nevertheless, if the quality map of feeding grounds were combined with find more fish distribution maps, it would elevate our knowledge to a different level. As in many other modelling

approaches the outcome of our method is dependent on the quality of the initial data. The type of data for the service user module can be selected according to the aim of a study (in our case relatively robust data were sufficient) and could range from several categories of importance based on expert knowledge to exact figures of prey numbers and their weight. Bumetanide For the service provider module of the best available data on both macrozoobenthos and predictors it would be advisable, for instance, to add other environmental parameters such as organic content and nutrient supply, which could obviously enhance the quality and applicability of models (Gogina & Zettler 2010). Furthermore, accuracy assessments have stressed that the different quartiles of a predictor range may be unevenly justified by macrofauna data, so the sampling strategy should take into account the spatial peculiarities of important predictors, especially that part of a range where significant changes in the characteristics of macrofauna occur. Our method may have many other applications. The data in the user module (in this case the feeding of cod, flounder and eelpout) could easily be replaced by different objects like the feeding of other fish species or even birds.