Evaluation of welfare and productivity in pregnant sows kept in stalls or in 2 different group housing systems

One hundred eighty pregnant sows (Sus scrofa), from first to ninth parity, on a commercial farm were selected and used in 3 different replicates (60 sows per replicate). Sows were housed from day 29 of pregnancy to 1 week before parturition in conventional stalls, in groups of 10 with trickle feeding or in groups of 20 with an unprotected electronic sow feeder (Fitmix; 20 sows per housing system per replicate). All the sows were equally feed restricted. Behavior, acute-phase protein (APP) plasma levels, and productivity were assessed. General activity and stereotypies were measured by scan-sampling observation at different times of the day on 11 nonconsecutive days. Overall, sows in groups with Fitmix showed an increase in resting behavior and a greater decrease in the frequency of oronasofacial stereotypies than sows in groups fed by trickle feeding (P < 0.001). Sows in the group-housing system showed a decrease in the frequency of sham chewing (P < 0.001). Gilts showed a lower general activity and frequency of stereotypies than older sows in all of the systems (P < 0.001). In general, APP levels and productivity measures did not differ across systems. However, lameness and vulva injuries were more often detected in group-housing systems. Removal of animals was more frequent in sows with Fitmix than in the other systems. In conclusion, well-managed, group-housed sows seem to increase resting behavior and decrease stereotypies at similar productivity and tissue damage levels. However, higher quality stockmanship may be required for early detection and resolution of problems in group-housing systems. Nevertheless, long-term effects of group-housing systems on welfare should be further assessed before recommendations are made.