Reproductive Issues in Welfare-Friendly Housing Systems in Pig Husbandry: A Review


Author’s address (for correspondence): NM Soede, Adaptation Physiology Group, Department of Animal Sciences, Wageningen University, PO Box 338, 6700 AH Wageningen, the Netherlands. E-mail:


In Europe, housing conditions of sows are currently changing, related with a larger emphasis on pig welfare. As a result, sows are and will be less kept in crates, but more so in loose housing systems (farrowing, lactation) and group housing systems (pregnancy, lactation, weaning-to-oestrus interval). These changes in housing conditions may affect reproductive functioning of the sows. Group housing of sows may decrease farrowing rate and litter size when stress levels rise or when feed intake in early pregnancy is not fully secured. Loose housing during farrowing results in an improved farrowing process, but may increase piglet mortality by crushing during early lactation. Further, group housing during lactation may increase the risk of lactational oestrus. Thus, new – welfare friendly – housing systems require increased attention to management to ensure optimal reproductive performance.


In most commercial pig production systems in the world, sows are individually housed in stalls or crates for most of their reproductive life; during pregnancy (∼115 days), during lactation (most commonly 20–30 days) and also between weaning and oestrus (4–7 days). The housing systems are mostly slurry based, with (partly-) slatted floors and provide limited or no bedding. The short lactation periods optimize the number of litters per sow per year from an economic point of view because sows generally remain anoestrous during lactation.

Besides economic advantages, these systems allow easy control of individual animals for the aspects like feed intake, health and oestrus.

However, from a welfare point of view, these systems are far from optimal. Gestation stalls and farrowing crates restrict freedom of movement and may increase skin lesions because of prolonged contact with hard surfaces (Bonde et al. 2004; KilBride et al. 2009) Farrowing crates have been developed to protect piglets from being crushed by the sow. However, crated sows show increased heart rate and cortisol concentrations around farrowing as compared with loose-housed sows, indicating that this restrain is stressful for the sow (Cronin et al. 1991; Jarvis et al. 2001; Oliviero et al. 2008). Crated sows normally are also unable to perform nest-building behaviour, which is a behavioural need for the pre-partum sows that affects the farrowing process and, as a consequence, piglet vitality (Wischner et al. 2009). During lactation, sows in farrowing crates and pens are continuously exposed to their piglets, whereas in more natural systems [e.g. get-away systems (Pajor et al. 2000)], sows can leave the pen. Finally, if lactating sows are crated, piglets have limited possibilities to learn from their mother, for example, how to eat solid feed. If sows are loose housed during lactation and environmental enrichment is provided (e.g. straw, branches) piglets have a better feed intake, fewer maladaptive behaviours (including tail biting at a later age, Munsterhjelm et al. 2010) and better performance before and after weaning (Oostindjer et al. 2010, 2011). Group housing of sows and piglets during lactation may further stimulate this. These welfare issues in commercial pig production systems have triggered societal call for alternative pig production systems. Various countries have developed legislation that resulted in changes in pig production systems, such as, for instance, group housing for pregnant sows in Europe. Moreover, alternative systems have been developed for specific markets (e.g. organic farming, outdoor pig production), and the welfare implications of these systems have been reviewed (e.g. Baxter et al. 2012, 2011). However, there are also reproductive issues associated with these new systems. This review will shortly introduce various management systems that are (being) developed and discuss the reproductive issues associated with these systems.

The Pregnant Sow


For pregnant sows, housing conditions vary largely. Sows can be full-time housed in individual crates, or they can be housed in groups, whereby group sizes may vary from small (4–5 sows) to very large (up to 250 sows). Also feeding systems for pregnant sows vary greatly, which partly coincides with the differences in group size. Sows are mostly restricted fed during pregnancy, but also ad libitum feeding is used, sows can be fed using, for example, feeding troughs, trickle feeders or electronic sow feeders (ESF), they may receive wet or dry feed, and they may eat together or one by one. Table 1 summarizes the current use of these systems in a number of countries across the globe. A note should be made that in the EU, from 1 January 2013 onwards, pregnant sows may only be individually housed during the first month of pregnancy. Depending on national legislation, sows in some countries (e.g. Finland, the Netherlands) need to be group housed during complete gestation. In Sweden and the UK, sows not only need to be group housed during pregnancy, but also during the weaning-to-oestrus interval. However, many farmers still need to implement group housing during pregnancy. Apart from the mentioned differences in group size and feeding system, other aspects may also differ in group housing systems [flooring (bedding), room temperature, feeding level, indoors/outdoors etc.].

Table 1.   An impression of the variability in group housing systems during sow pregnancy in several countriesa,b
Feeding systemUKc,dSwedencFinlandNetherlandsDenmarkBelgiumFranceAustraliaUSA
SizeFarms (%)SizeFarms (%)SizeF arms (%)SizeFarms (%)SizeFarms (%)SizeFarms (%)SizeFarms (%)SizeSows (%)eSizeFarms (%)
  1. aExperts from several – mostly European – countries have contributed; UK (A Cliff), SW (AC Olsson), FI (O Peltoniemi); Denmark (LU Hansen); Belgium (FAM Tuyttens); France (S Boulot; Courboulay et al. 2010) Br (F Bortolozzo); Australia (P Langendijk; APL, 2010); USA (WF Flowers).

  2. bNot included: Brazil (virtually all sows confined in pregnancy).

  3. cK and Sweden: grouping from weaning onwards.

  4. dUK: Relates to indoor sows only (60% of sows).

  5. eAUS: Sow-based percentages because of skewed farm size.

  6. fAUS: Individual housing for at least 8 weeks of pregnancy (APL 2010).

  7. gSweden: Always includes a separate lying area for the sows.

  8. na, not applicable.

Individual housing throughout pregnancy
Group housing – individual feeding
 Stalls4–30276 to >2095g408010–502630–5027?2010–2031 ?20–253
 Electronic sow feeder30–100+15?11001040–100+3650–15028?430–15013 ?20–252
Group housing – group feeding
 Floor10–100+30na0 0<15615–2010?8??54–108430–405
 Trough5–10254–12210207–15515–2010?6–944 30–406
 Trickle feeder4–1034–6220108–20<110–15<1?   ?na0
 Ad libitumNa0Na04058–20<1na0 4   ?na0
Straw bedding 90 1004060 15  ?? 17 <10xxx0

The choice for a housing system is usually not based on fertility criteria, but may be based on, for example, welfare legislation, trade mark production, building costs, labour requirements or level of control of the animals (Tuyttens et al. 2008, 2011). Nevertheless, several aspects of the housing system may affect reproductive parameters such as farrowing rate, the number of empty days (affected both by the percentage of sows returning to oestrus and by the efficiency of tracing non-pregnant sows) and litter size, and possibly have carry-over effects into lactation, affecting, for example, the parturition process or piglet survival.

Factors that may affect fertility

Several studies have compared group-housed sows with individual-kept sows and concluded that reproductive performance, measured as farrowing rate and litter size, may be at risk in group-housed sows (e.g. McGlone et al. 2004; Kongsted 2004a). Only few studies have compared reproductive functioning in different group housing systems. Broom et al. (1995) found similar performance in a dynamic 38-sow Electronic Sow Feeder (ESF) group compared to a five-sow group with feeding stalls, Courboulay and Gaudré (2002) compared six-sow trough fed groups with 12-sow ESF groups and found a tendency for lower litter size in ESF sows. In a study of Van der Peet-Schwering et al. (2003), sows in a static ESF system (25 sows in the group) had a higher return rate than sows in groups of 12 with free access stalls. Van der Peet-Schwering et al. (2003) suggested that either the lower feed intake or the high level of aggression around the feeder caused the higher return rate in ESF sows.

More recently, Van der Peet-Schwering et al. (2009) investigated success factors of group housing systems for sow welfare and performance by evaluating both performance, welfare, management and housing of sows on 70 farms with group housing during the complete pregnancy (<4 days from insemination). It was concluded that in ‘each system of group housing, adequate reproductive results and animal welfare can be achieved’. The system in itself was not a major success factor. This is illustrated in Fig. 1, which shows the variation in farrowing rate between the group housing systems in the study. Similar results were found for other production measures (number of piglets weaned per sow per year; culling rate) and animal measures (backfat, claw lesions, skin lesions). Similarly, in more than 250 farms, Boulot et al. (2011) found that farms with group housing during pregnancy had similar reproductive performance as farms with individual housing during pregnancy, seeing a large variation in the performance within the systems. They concluded that gilt management, time of grouping and ESF-management are critical success factors for group housing. Van der Peet-Schwering et al. (2009) concluded that successful group housing needs famers that have an animal-directed approach, as shown, for instance, by their attention to individual sows in feed allowance, and also several aspects of gilt rearing, such as space allowance and feeding strategy. Only few specific housing conditions appeared to be important. For instance, on the 25% farms with the lowest farrowing rate (<83%), sows had a lower space allowance than on the 25% farms with the highest farrowing rate (<83%: 2.1 m2, 83–89%: 2.2 m2 and >91% 2.6 m2; p < 0.10).

Figure 1.

 Variability in average farrowing rate (2005/2006) of Dutch farms with group housing within 4 days from insemination onwards, in relation to the applied group housing system; electronic sow feeders (ESF) without straw (1), ESF with straw bedding (2), free access stalls (3), trough feeding (4) and other (5). Adapted from Van der Peet-Schwering et al. (2009)

The question, obviously, remains which specific aspects of group housing systems affect reproductive functioning. Based on extensive literature reviews, both Kongsted (2004b) and Spoolder et al. (2009) concluded that (variation in) feed intake and stress may be major factors affecting reproductive performance in group housing systems during pregnancy. These factors will be shortly reviewed below.


Stress factors such as stocking density, grouping of unfamiliar animals and associated aggression, competition for food, poor environments, thermal extremes and poor human-animal handling may all affect reproductive functioning (e.g. Varley and Stedman 1994). During pregnancy, such factors may lead to embryonic mortality, especially when (the combination of) such stressors results in chronically elevated cortisol levels. In the majority of group housing systems, grouping takes place during pregnancy. Both Van Der Mheen et al. (2003; dynamic 52-sow ESF groups) and Kirkwood and Zanella (2005; static 15-sow groups with floor feeding) concluded that grouping preferably takes place immediately after insemination and should not take place around the period of maternal recognition of pregnancy (2–3 weeks of pregnancy). Van Wettere et al. (2008); six-gilt groups with floor feeding) did, however, not find effects of timing of grouping on embryonic survival. This suggests that other factors may overrule the potentially negative consequences of grouping on embryonic mortality.

Body condition/Feed intake

A series of Swedish experiments have shown that short-time food deprivation during the first weeks of pregnancy may affect embryo functioning in different ways. For example, food deprivation at 2–3 days after oestrus delayed ova transport through the oviduct (Mwanza et al. 2000; Razdan et al. 2004), food deprivation at 10–11 days immediately decreased systemic progesterone and oestradiol levels (Tsuma et al. 1996) and food deprivation at days 13–14 decreased day 30 allantoic progesterone levels (Mwanza et al. 2000; Razdan et al. 2004). Although in these experiments embryo survival was not affected, they suggest that reproductive performance in group housing systems may be at risk when sows have to compete for their food. Indeed, Kongsted (2004b, 2006) observed eating behaviour in group-housed sows in 14 herds and concluded that sows that spent less time eating had a higher risk of returning to oestrus. Also, sows with a lower backfat gain during early gestation had a lower farrowing rate and litter size. In these sows, performance was not related with ‘fear’ and ‘social stress’ scores, which suggests that the consequences of a low feed intake may not be stress related. Other studies have also found that relatively high feeding levels during early pregnancy may be beneficial for litter size in sows (e.g. Sørensen and Thorup 2003; Hoving et al. 2011), although the physiological mechanism has not been elucidated.


The change from individual to group housing for pregnant sows poses new challenges to sow management. Group housing systems vary largely, in many different aspects. To achieve good reproductive performance, specific attention needs to be given to stress levels (related with unfavourable social, management and climatic conditions) and feed intake during early pregnancy.

The Lactating Sow


Various systems are used to improve welfare of sows and piglets in lactation, which can be arbitrarily divided into three main categories.

Loose-housed sow

Some systems are characterized by loose housing of the sow during parturition and lactation. Around parturition sometimes some bedding material is supplied to support nest-building activities.

Get-away systems

More freedom for the sow is allowed in systems where sows are allowed to get away from the piglets after farrowing (e.g. by stepping over a barrier). When not with the piglets, sows usually meet in a communal area in smaller or larger groups. Pajor et al. (2000) showed that sows will progressively stay away longer from the piglets when lactation progresses. Time spent away from the piglets, however, varies considerably between sows. Weary et al. (2002) described a system in which sows and piglets are both allowed to go to a communal area from day 12 (sows) or 14 (piglets) of lactation onwards. At the end of lactation (day 27), these sows spent over 14 h per day away from there piglets and nursed approximately 40% less as compared to control sows in farrowing crates.

Group lactation systems

Sows may also be grouped during (part of) the lactation. Group sizes and moment of introduction in the group after onset of lactation vary for the different systems [e.g. Hultén et al. (2006)]. In general, these systems are normally combined with longer lactation lengths (6–8 weeks).

Reproductive issues

Farrowing process and piglet survival

Recently, Pedersen et al. (2011) investigated causes of piglet mortality in crates and indoor pens and concluded that the odds for stillborn, crushed or dying of starvation were similar in these conditions. Moreover, piglet traits for pre- and postnatal survival were also similar; the chance of being stillborn increased when piglets were born late in the birth order, after a long inter-birth interval and with a lighter piglet birth weight. Further, lighter piglets had a higher chance of being crushed or dying after starvation. They concluded that the micro-climate at farrowing and the heat-preserving properties are more important than the housing conditions at farrowing. Others reported, however, that loose-housed sows with the opportunity to perform nest-building behaviour have a shorter farrowing duration than crated sows (Thodberg et al. 2002; Oliviero et al. 2008, 2010). Interestingly, Thodberg et al. (2002) found this shorter farrowing duration only in first parity and not in second parity sows. Oliviero et al. (2008) correlated the shorter farrowing duration with increased oxytocin concentrations. Oliviero et al. (2010) also confirmed earlier findings that a shorter farrowing duration is related with a lower stillborn rate, but the authors also state that it is difficult to establish cause and effect. Preliminary analyses in a Dutch study on farrowing behaviour in crated and loose-housed sows show that loose-housed sows have fewer posture changes during farrowing, accompanied by a lower chance of piglets dying during parturition (A.M.E. Raats, A.I.J. Hoofs and N.M. Soede, unpublished results).

Thus, loose-housed sows may be more relaxed during farrowing, especially when they have had the chance to express nest-building behaviour. This may result in a shorter farrowing process and thus fewer still born and more vital piglets.

It is generally assumed that loose housing increases piglet crushing in the first days of lactation. Marchant et al. (2000) found a dramatic increase in mortality rates in free crates (26%) and pens (25%) compared to conventional farrowing crates (13%). In loose farrowing systems in Switzerland, however, on average only 11.8% of the live born piglets died (Weber et al. 2009). In another study, Weber et al. (2007) found that total piglet mortality was similar in loose-housed and crated sows, although more piglets died from crushing in loose-housed sows. Similarly, KilBride et al. (2012), based on data from 112 sow herds in England, concluded that loose housing increased risk of crushing, but decreased the risk of death from other causes, resulting in similar overall pre-weaning mortality. So, indeed, piglet crushing needs extra care in loose housing systems. Damm et al. (2010) found that the number of crushings and near crushings were lower in penned sows that had access to straw.

Lactation oestrus

In get-away systems and group lactation systems, there are several aspects that pose a risk for lactation oestrus. Sows normally remain anoestrus during lactation as a result of suppression of GnRH/LH release from the hypothalamic-pituitary system, thereby preventing outgrowth of follicles to ovulatory sizes. This suppression is suckling-induced and is mediated by opioid peptides (reviewed by Quesnel 2009). As previously mentioned, in get-away systems, the suckling frequency at the end of lactation is much lower as compared to conventional systems, which means that the suppressing effects of endogenous opioids on GnRH/LH release may be counteracted. Moreover, Pajor et al. (2002) showed that in these systems sows may lose less weight during lactation. In sows with high weight losses during lactation, GnRH/LH release will not restore sufficiently as compared to sows with low or no weight loss. In group-housed sows, the presence of foreign piglets may affect the nursing behaviour of the sow and, in this way, also affect the chance of lactational oestrus.

Pedersen et al. (1998b) used a ‘get-away’ system where sows had access to a communal area, area for four sows and piglets of these sows could mingle from an age of 7 days onwards. The presence of foreign piglets had a detrimental effect on nursing as competition for milk increased thereby reducing the sow’s motivaton to nurse and accelerating the weaning process. This may contribute to the induction of lactation oestrus. Weary et al. (2002) allowed piglets to mingle from an age of 14 days. The incidence of cross-suckling was low in this study, which may be explained by the fact that piglets at that age may have developed a sufficiently strong attachment to their own mother (Vanheukelom et al. 2012).

As mentioned before, in the studies investigating loose house systems, lactation lengths are generally quite long (6–8 weeks). Milk production and suckling frequency decrease after the third week of lactation, and piglets become less dependent on milk. As a result, sows may become anabolic during lactation, which poses another risk factor for lactation oestrus. There is limited information on the occurrence of ovulation and oestrus during lactation in get-away systems and group-housed lactating sows. Hultén et al. (2006) studied ovarian activity and oestrus signs in group-housed lactating sows. Sows were in groups of four sows with piglets from 3 to 7 weeks of lactation. On average, 47% of the sow showed a lactation ovulation, but in only 50% of these sows standing oestrus was detected. In an earlier study, Hultén et al. (1995) found an ovulation frequency of 28% in farms with a similar set-up. In both studies, older parity sows showed higher ovulation frequencies during lactation then first parity sows. The moment of ovulation appeared quite unpredictable. In the study of Hultén et al. (2006), approximately 50% of the sows ovulated in the last week of lactation, but also in 3–6 weeks sows ovulated. Sows showing lactation oestrus where not inseminated and therefore weaning-to-oestrus intervals were prolonged. In an attempt to synchronize lactation ovulation and inseminate sows during lactation, Kongsted and Hermansen (2009) performed a study in organic sows that were individually housed until day 35 of lactation followed by group housing and introduction of a boar. Piglets were weaned at 8 weeks. All sows showed lactational oestrus, and 84% of the sows showed oestrus within one week after grouping and boar introduction; 84% of the sows were diagnosed pregnant at 5 weeks after oestrus.

Another approach to stimulate and control lactation oestrus and ovulation in sows is by intermitted suckling. In this approach, sows and piglets are separated for 10–12 h per day starting at 2–3 weeks of lactation [reviewed by Kemp and Soede (2012)]. Using intermittent suckling, lactation oestrus and ovulation can be induced. Timing of oestrus is very much synchronized, at 5–6 days after start of intermittent suckling and oestrus induction rates vary between 50% and 100% depending on parity and breed. Reproductive results in terms of ovulation rate, embryo survival, farrowing rate and litter sizes were not compromised when start of intermitted suckling is after day 19 of lactation. However, embryo mortality may increase when lactation continues for 23 day into next pregnancy.

Litter size

Over the last years, average number of live born piglets in Dutch organic farming was consistently higher (0.5–0.7 piglets per litter) than in conventional pig production (J.I. Leenhouwers, personal communication). This may be related to the improved energy balance of organic sows in the course of the 40-day lactation that results in improved follicle development and ultimately a higher litter size (as discussed by Wientjes et al. 2012). Unfortunately, higher litter sizes usually also imply lower average birth weight and increased variability in birth resulting in less viable piglets; Wientjes et al. (2012) found average litter sizes of approximately 17.0–18.8 piglets, an average birth weight of 1260 g and pre-weaning mortality rates of 21–31% in organic sows with an average lactation length of 41 days.


Farrowing pen design influences the farrowing process and (early) piglet survival. Loose housing systems on the one hand ease the farrowing process, resulting in lower piglet mortality at farrowing but on the other hand, may increase the risk of piglet crushing. Optimized pen design and management are crucial to reduce this risk. Group housing may increase the risk of lactational ovulation. This risk may be bent into an advantage, when stimulating ovulation induction and oestrus expression, since insemination during lactational oestrus can result in good reproductive performance.

The Weaned Sow

In some European countries, such as, for instance, Sweden and the UK, all sows are group housed from weaning onwards. Effects of group housing of sows may affect oestrous characteristics in sows, as has been reviewed by Kemp et al. (2005). Unfortunately, recent literature is lacking. In modern genotypes, timely oestrus onset is less a problem than it used to be. Therefore, older literature investigating oestrus characteristics in group-housed sows may not be relevant to our modern genotypes.

Reproductive issues

Onset of oestrus

Some studies in sows suggested an advancing effect of group housing on onset of oestrus after weaning while others find no effect or even a negative effect of group housing. For example, Pearce and Pearce (1992) found that, compared to isolated sows, housing sows adjacent to anoestrous sows increased weaning-to-oestrus intervals and housing adjacent to oestrous sows reduced and synchronized weaning-to-oestrus interval. They suggest that, as in other species, oestrous sows may release pheromones partly explaining the oestrus advancing effect. The lengthening of the weaning-to-oestrus interval when sows were housed adjacent to anoestrus sows may be due to aggression during daily contact as Turner and Tilbrook (2006) found that hormones related to stress (ACTH and corticosteriods) delay ovarian events leading to ovulation.

Expression of oestrus

Effects of group housing on oestrus detection rate and expression of oestrus are generally small. For example, Langendijk et al. (2000) found no effect of group housing (groups of four) versus individual housing after weaning on oestrus detection rate or duration of oestrus. Factors that may affect expression of oestrus in groups are social status of the sows, group size and space allowance. Pedersen et al. (1998a) showed that sows receiving the highest amount of aggression in group housing with two other sows after weaning showed significant less social behaviour and a shorter duration of oestrus. Moreover, Pedersen et al. (2003) showed that subordinate sows in pair housing showed less proceptive and receptive behaviour towards a boar as compared to dominant sows. Subordinate sows showed fear-related behaviour in response to boar stimulation even when they were in oestrus. Thus, both oestrus detection and mating may be impaired in subordinate sows.

Not much is known about the effects of group size and space allowance on expression of oestrus in weaned sows. Generally, it is thought that in group housing, space allowance and pen design should be sufficient to allow adequate escape opportunities in order to alleviate the social stress experienced by subordinates in a group (Pedersen et al. 2003).


It can be concluded that effects of group housing of sows after weaning on onset of oestrus and expression of oestrus are variable. Positive effects may be found if oestrous sows are in the group and in cases of extended weaning-to-oestrus intervals. Negative effects can be found if severe aggression takes place in newly formed groups.

Concluding Remarks

In general, it seems that more welfare-friendly housing systems can have at least similar reproductive results as conventional housing systems. However, several phases of the reproductive cycle need specific management attention to avoid low litter size, piglet crushing and lactational ovulation. In general, in outdoor pig production systems, the management control may be more of a challenge and is obviously also more affected by variable environmental conditions (e.g. sun burn, cold stress).

Another crucial element is the fact that after a switch to a new housing system, not only the farmers, but also the sows have to get used to the new systems and associated changes in management. This may temporary result in lower performance of the sows after such a switch.

Conflict of interest

None of the authors have any conflict of interest to declare.