A computer-based decision support system for welfare assessment in pregnant sows was constructed. This system uses a description of a husbandry system as input and produces a welfare score on a scale from 0 to 10 as output. Pregnant sows were chosen as a case in search for a formalised, i.e. structured, transparent, yet flexible procedure to 'objectively' assess the overall welfare status of farm animals in relation to the housing and management system based on available (and undisputed) scientific knowledge. The procedure to construct the welfare model and to calculate welfare scores is described. Decision making is based on the needs and distress of the animal thus from the perspective of the animal.
The sow welfare (SOWEL) model was validated using expert opinion in that there is a substantial agreement between pig welfare scientists and the model about the ranking of housing systems and, to a lesser degree, about the weighting of attributes of housing systems. The most important welfare-relevant attributes concern aspects of social contact, space, and substrate. The housing systems were roughly divided into low-, mid-, and high-welfare systems. Low-welfare systems were conventional housing in individual stalls and tethers, while high-welfare systems in our data set all provided substrate and outdoor access.
For practical applications further development of the decision support system is recommended, as well as ongoing validation, upgrading and extending of the model, e.g. to other species. The results show that integrated welfare assessment based on available scientific knowledge is possible.
Keywords : farm animal welfare assessment, pigs, applied ethology, housing systems, model, computer, knowledge base, expert system.