Ninety-six pregnant sows, previously penned in individual stalls, were housed in groups of four in pens containing three zones: a) four individual feeding stalls without rear gates, b) a communal sleeping kennel and c) a dunging area between the stalls and kennel. The feeding stalls were either long (2m) or short (1m) with barriers made of wire mesh. The daily allowance of 2.5kg pelleted feed per sow was presented either all at once (dump) or at the rate of approximately 100g/min by a manual system (trickle). Dung disposal systems were either a pit füled with sawdust-based compost, a slatted floor, or straw on a sloped concrete floor. There were two replicates of the 2×2×3 factorial design. The location of sows was observed from time-lapse video recordings taken at regular intervals throughout the 31 day experimental period.

Use of the three zones of the pens was influenced by feeding method, barrier length and dung disposal system. Overall the feeding stalls were used more with trickle than with dump feeding (P<0.05); more with long than with short barriers (P<0.001), and most with slatted floors and least with the straw system (P<0.001). AU these treatments had the opposite effects on the use of the communal kennel The use of the dunging area was not affected by feeding method or stall length but was greater (P<0.01) on the compost compared with the other two dung disposal systems. Throughout the experimental period the use of feeding stalls decreased (P<0.001). Orcadian use of the pen zones interacted with barrier length and dung disposal system. Kennels were used most at night in pens with straw, exceeding 54minlh with half length (1m) stalls. The lowest night-time use of kennels, around 8min/h, was found in pens with full length (2m) stalls combined with either compost or slats. Ambient temperature did not have a major influence on use of pen zones.

The occupation of stalls, especially full length stalls, indicates that offering a choice between solitary and communal areas may improve the welfare of group-housed sows.

The use of farrowing crates is increasingly questioned from an animal welfare point of view. Even so, since a number of attempts to develop loose farrowing systems have been unsuccessful, leading to high levels of piglet mortality due in the main to crushing, many farmers remain sceptical as to whether or not alternative systems can be viable. On the other hand, several European countries have introduced legislation requiring loose farrowing systems, thus promoting research into this type of housing and allowing for performance studies based on large samples of commercial farms. As a consequence of these recent developments, we think it timely to reconsider the evidence available on loose farrowing systems. In our review, we first address the normal peri-parturient behaviour of domestic pigs, as well as studies comparing behaviour and stress physiology in sows kept in both crates and loose systems during farrowing. We then review approaches taken to develop alternative farrowing systems in different countries, and focus lastly on pen, piglet and sow characteristics that contribute to piglet survival in loose farrowing systems. Taking scientific evidence as well as practical experience into account, we conclude that piglet mortality in loose farrowing systems need not exceed that of crate systems. To obtain good performance results, sows due to farrow should be kept individually in sufficiently large pens, structured for preference into a nest area and an activity area. Furthermore, both management and breeding aspects, resulting in high piglet viability and good maternal behaviour, are essential to achieve high production in loose farrowing systems.

In the Family Pen System piglets and fatteners grow up in species-specific family groups that correspond to the normal social organization of domestic pigs. The feasibility of a technologically improved version of this alternative housing system, originally designed by Alex Stolba, was tested on a commercial farm for two and a half years. Eighty-one litters were born in three family groups within this period. Average cycle length was 170 ± 24 (SD) days, resulting in 2.15 litters per sow per year. All piglets were suckled for at least seven weeks. In 53.8 per cent of the cycles lactational oestrus occurred before the piglets were seven-weeks-old. The litters of sows which did not show lactational oestrus were artificially weaned and returned to the family group as soon as the sow had been served. At the beginning there were problems with piglet health and crushing, but in the last 21 months of the study there was a stable reproductive performance of 19.5 piglets (28-days-old) reared per sow per year (n = 53 litters). Sows that had been raised themselves in the Family Pen System reared 21.4 piglets per year (n = 25 litters). In conclusion, the technologically improved version of the Family Pen System was found to be practicable on a commercial farm.