Actions promoting production animal welfare are often seen as investments without reward. By definition, high welfare is characterized by low levels of stress (Broom and Johnson 1993). As stress is shown to have adverse effects on reproduction (Dobson and Smith 2000), a low level of welfare may be suspected to have similar output. Many effects of management on pig reproduction are basically understood. These factors influence the wellbeing of animals, and when combined they can be used to assess welfare. There are probably no studies published on effects of animal welfare on productivity, if welfare is considered a state comprised of several contributing factors as housing, stockmanship etc. Poor health or productivity in animals can be used as an indicator of poor welfare. Good health and especially productivity can, however, not be taken as conclusive evidence of good welfare (Scientific Veterinary Committee
1997). Detrimental effects on reproductive performance are to be expected only when welfare is considerably compromised, because coping systems of animals have evolved to minimize effects of stress on reproduction (Dobson and Smith 2000). The Scientific Veterinary Committee states that ‘If there are differences between systems, even a small effect on reproduction may indicate considerable welfare problems’ (Scientific Veterinary Committee 1997). The Animal Needs Index TGI 35L (Tiergerechtheitsindex) is a scoring method developed to certify the level of pig, cattle and lying hen welfare on farms (Bartussek 1999). Points are assigned to easily identifiable attributes of the housing system. Management, health and behaviour parameters are noticed superficially. Based on knowledge of the needs of the animals more points are given for favourable circumstances. All points are summed for a final score that judges the prerequisites for animal welfare. The assessment of welfare by TGI 35L is classified as indirect, because most aspects are environment-based (Johnsen et al. 2001). Objectives of the study were to investigate the relationship between different parameters describing on-farm sow welfare and reproductive performance. Welfare was assessed using a modified TGI 35L–Animal Needs Index. The modified index had a maximal total score of 100. It was comprised of six categories: ‘locomotion’ (maximal score 21 for dry sows and 11 for lactating sows), ‘social interaction’ (12/8), ‘floor quality’ (16/9), ‘stable climate’ (16/21), ‘feeding’ (16/23) and ‘health and stockmanship’ (19/29). Index scorings were performed separately in farrowing, breeding and gestation units on 28 representative Finnish sow farms. One-year production parameters were collected. Multiple linear regression was used for statistical analysis. Total ANI-points varied between 36.5–68.0 for lactating and 39.5–86.0 for dry sows. Litter size increased with increasing scores for ‘feeding’ in the dry sow unit. Controlling for breed, high scores for ‘health and stockmanship’ and ‘floor quality’ shortened the reproductive cycle, probably because of good leg health. The number of weaned piglets per sow per year (PSY) increased with increasing scores for dry sow ‘health and stockmanship’, ‘floor quality’ and an interaction of ‘feeding’ in the farrowing and mating units. PSY increased with decreasing scores for farrowing pen ‘climate’. It was concluded that high-quality floors and stockmanship appear to correlate positively with reproductive performance in the sow; however, the effects of a welfare-promoting feeding strategy on reproduction are contradictory.