Schriftenreihe des Instituts für Tierzucht und Tierhaltung der Christian-Albrechts-Universität zu Kiel, Heft 222, 2018
©2018 Selbstverlag des Instituts für Tierzucht und Tierhaltung
der Christian-Albrechts-Universität zu Kiel
Olshausenstraße 40, 24098 Kiel
Schriftleitung: Prof. Dr. J. Krieter
ISSN: 0720-4272
Gedruckt mit Genehmigung des Dekans der Agrar- und Ernährungswissen-
schaftlichen Fakultät der Christian-Albrechts-Universität zu Kiel
Aus dem Institut für Tierzucht und Tierhaltung
der Agrar- und Ernährungswissenschaftlichen Fakultät
der Christian-Albrechts-Universität zu Kiel
Evaluation of group-housing systems of lactating sows –
Impact on behavioural, health and performance parameters
Dissertation
zur Erlangung des Doktorgrades
der Agrar- und Ernährungswissenschaftlichen Fakultät
der Christian-Albrechts-Universität zu Kiel
vorgelegt von
Master of Science
Charlotte Gertrud Elisabeth Grimberg-Henrici
aus Köln
Dekan: Prof. Dr. Joachim Krieter
Erster Berichterstatter: Prof. Dr. Joachim Krieter
Zweiter Berichterstatter: Prof. Dr. Imke Traulsen
Tag der mündlichen Prüfung: 24. Januar 2018
Die Förderung erfolgte dankenswerter Weise aus Mitteln des Zweckvermögens des
Bundes bei der Landwirtschaftlichen Rentenbank.
Meiner Familie
TABLE OF CONTENTS
GENERAL INTRODUCTION
…………………………………………………………………………………… 001
CHAPTER ONE
Does housing influence maternal behaviour in sows?
…………………………………………………………………………………… 007
CHAPTER TWO
What do maternal tests actually test?
…………………………………………………………………………………… 031
CHAPTER THREE
The effect of different housing systems on reproductive traits
and the behaviour of low-risk and high-risk crushing sows
…………………………………………………………………………………… 047
CHAPTER FOUR
Cortisol levels and health indicators of sows and their piglets
living in a group-housing and a single-housing system
…………………………………………………………………………………… 0 =73
GENERAL DISCUSSION
…………………………………………………………………………………… 097
GENERAL SUMMARY
…………………………………………………………………………………… 113
ZUSAMMENFASSUNG
…………………………………………………………………………………… 117
1
GENERAL INTRODUCTION
Since the permanent fixation of pregnant sows is banned in Europe, it has become a matter of
discussion on whether the permanent fixation of lactating sows is still acceptable. The
dominant housing system for sows during farrowing and lactation is permanent fixation in
crates. Several studies have reported that crated sows showed higher stress levels compared to
loose-housed sows (Lawrence et al., 1994; Jarvis et al., 2002; Oliviero et al., 2008). In
addition, Arellano et al. (1992) reported that sows in crates performed significantly more
frequent stereotypies compared to loose-housed sows. The main reason for permanent fixation
in crates is the prevention of piglet losses due to crushing. In the first 24 hours post partum the
piglets are most vulnerable to crushing and 50 % of the crushed piglets are documented in the
first two days after birth (Marchant et al., 2001; Kilbride et al., 2012). However,
Lambertz et al. (2015) showed that fixation of the sows for a short period is sufficient to
reduce piglet losses. Systems without any fixation offer the sows even more space,
self-determination during farrowing and better contact with their piglets.
Bohnenkamp et al. (2013b) investigated a group-housing system in which the sows were kept
in pens with fixation in farrowing crates, however, the sows had the opportunity to leave the
pens one day after birth to move freely and to meet other sows. Five days post partum, all
litters were also able to run together in the group-housing system. Here, no differences in
piglet mortality rates were found between group-housed sows and crated sows. Further
positive effects of loose-housed sows were fewer skin lesions (Boyle et al., 2002) and better
locomotion due to higher muscle weights and bone strength (Marchant and Broom, 1996). For
group-housed sows during lactation, other studies have observed a stronger maternal bonding
with their piglets (Arey and Sancha, 1996), improved social skills of their weaned piglets
(Bohnenkamp et al., 2013a) and more playing behaviour of the piglets during lactation
(Arey and Sancha, 1996; van Nieuwamerongen et al., 2015). Playful behaviour has been
described as a positive welfare indicator (Fagen, 1981).
Maternal behaviour has been found to be unchanged throughout the progress of
domestication, (Jensen, 1986; Stolba and Wood-Gush, 1989). This is in line with behavioural
observations of domestic sows and crossbred sows (wild x domestic) with regard to maternal
behaviour (Špinka et al., 2000). Thus, a group-housing system imitates best the structures
seen in the nature of wild boars. This means living in small groups, farrowing individually,
returning to the group with the litter and contact with other sows and piglets during lactation
(Jensen, 1986). However, the pre-weaning piglet mortality in these alternative systems, such
as group-housing systems and free-farrowing systems, remains an economical and ethical
2
issue and has a multifactorial origin. Piglet mortality during lactation is associated with the
health of the piglets (Pedersen et al., 2011), the litter size and lower birth weights related
thereto (Milligan et al., 2002), the maternal performance (Marchant et al., 2001;
Andersen et al., 2005; Burri et al., 2009; Wischner et al., 2009), the condition and age of the
sow (Weary et al., 1998), the pen design (Weary et al., 1996; Herskin et al., 1998;
Damm et al., 2006; Baxter et al., 2015), the expertise of the sow with alternative farrowing
systems (Marchant et al., 2000), the carefulness of lying down and rolling movements of the
sow as reviewed by Damm et al. (2005) and the expertise of the stockpersons (Li et al., 2010).
The aim of the present thesis
The aim of this thesis was to investigate the performances of lactating sows and their piglets
in group-housing systems with different housing designs with the focus especially on
maternal behaviour, health indicators and reproductive traits. Different GH systems were
evaluated with regard to their feasibility for pig husbandry. Chapters One and Two describe a
group-housing system experiment in which six sows were fixed in farrowing crates in their
pens with the opportunity to leave the pens one day after birth, whereas the piglets had to stay
in the pens until five days post partum. Chapters Three and Four present the findings of an
investigation into a group-housing system in which ten sows were housed with their litter in
free-farrowing pens without any fixation until six days post partum. The sows and their
piglets were then given the opportunity to leave the pens. In both group-housing systems the
sows had a running area to meet other sows and piglets. The performances of the
group-housed sows were compared with sows housed in conventional single-housing systems
with permanent fixation.
The first Chapter investigates differences in maternal behaviour of group-housed and
single-housed sows. The sows were tested in the second and fourth week of lactation in six
successive maternal tests conducted in their home pens and in a test arena used to create a test
situation for the sows apart from their housing environment. The reproductive traits and the
condition of the sows were also analysed.
The second Chapter deals with the question of what maternal tests actually test for since
most studies only use a selection of tests to investigate maternal behaviour. It has not been
proven which types of maternal behaviour these tests evaluate and, thus, which maternal tests
are advisable. Therefore, the behavioural data collected in Chapter One with regard to the
3
reaction of the sows to piglet stress signals, separation from and reunion with their piglets
were analysed using a factor analysis to identify redundancies in behavioural parameters.
The third Chapter compares the reproductive performances especially regarding the piglet
losses of group-housed sows living in a smaller and in a larger system and single-housed
sows. In addition, sows were classified based on the number of crushed piglets in high-risk
and low-risk crushing sows. These sows were observed in the first 72 hours post partum to
find differences in their lying down and rolling behaviour and the behaviour of their piglets in
order to obtain information on critical situations in which piglets are crushed.
The fourth Chapter provides insights into the health status of group-housed and
single-housed sows and their piglets. All sows were scored one week ante partum and four
weeks post partum concerning lesions on their udder, body, shoulder, tail and vulva.
Furthermore, the locomotion, the dirtiness and the condition of the sows were documented. In
addition, saliva cortisol samples were collected from the sows one week ante partum, two
weeks post partum and four weeks post partum to obtain information on differences in stress
levels. Moreover, all piglets were evaluated in their first and fourth weeks of lactation with
regard to skin lesions on their face, body, carpus and tail.
The demand for loose-housing systems for sows during lactation is increasing. Therefore,
more insights are important when developing and advising on new systems. The aim of this
thesis was to evaluate group-housing systems with regard to their feasibility for pig
husbandry.
4
REFERENCES
Andersen, I.L., Berg, S., Bøe, K.E., 2005. Crushing of piglets by the mother sow (Sus scrofa)
- purely accidental or a poor mother? Applied Animal Behaviour Science 93, 229-243.
Arellano, P.E., Pijoan, C., Jacobson, L.D., Algers, B., 1992. Stereotyped behaviour, social
interactions and suckling pattern of pigs housed in groups or in single crates. Applied
Animal Behaviour Science 35, 157-166.
Arey, D.S., Sancha, E.S., 1996. Behaviour and productivity of sows and piglets in a family
system and in farrowing crates. Applied Animal Behaviour Science 50, 135-145.
Baxter, E.M., Adeleye, O.O., Jack, M.C., Farish, M., Ison, S.H., Edwards, S.A., 2015.
Achieving optimum performance in a loose-housed farrowing system for sows: the
effects of space and temperature. Applied Animal Behaviour Science 169, 9-16.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013a. Comparison of
growth performance and agonistic interaction in weaned piglets of different weight
classes from farrowing systems with group or single housing. animal 7, 309-315.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013b. Group housing for
lactating sows with electronically controlled crates: 1. Reproductive traits, body
condition, and feed intake. Journal of animal science 91, 3413-3419.
Boyle, L.A., Leonard, F.C., Lynch, P.B., Brophy, P., 2002. Effect of gestation housing on
behaviour and skin lesions of sows in farrowing crates. Applied Animal Behaviour
Science 76, 119-134.
Burri, M., Wechsler, B., Gygax, L., Weber, R., 2009. Influence of straw length, sow
behaviour and room temperature on the incidence of dangerous situations for piglets in
a loose farrowing system. Applied Animal Behaviour Science 117, 181-189.
Damm, B.I., Forkman, B., Pedersen, L.J., 2005. Lying down and rolling behaviour in sows in
relation to piglet crushing. Applied Animal Behaviour Science 90, 3-20.
Damm, B.I., Moustsen, V., Jørgensen, E., Pedersen, L.J., Heiskanen, T., Forkman, B., 2006.
Sow preferences for walls to lean against when lying down. Applied Animal
Behaviour Science 99, 53-63.
Fagen, R., 1981. Animal play behavior.
Herskin, M.S., Jensen, K.H., Thodberg, K., 1998. Influence of environmental stimuli on
maternal behaviour related to bonding, reactivity and crushing of piglets in domestic
sows. Applied Animal Behaviour Science 58, 241-254.
5
Jarvis, S., Calvert, S.K., Stevenson, J., VanLeeuwen, N., Lawrence, A.B., 2002. Pituitary-
adrenal activation in pre-parturient pigs (Sus scrofa) is associated with behavioural
restriction due to lack of space rather than nesting substrate. Animal Welfare 11, 371-
384.
Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs.
Applied Animal Behaviour Science 16, 131-142.
Kilbride, A.L., Mendl, M., Statham, P., Held, S., Harris, M., Cooper, S., Green, L.E., 2012. A
cohort study of preweaning piglet mortality and farrowing accommodation on 112
commercial pig farms in England. Preventive Veterinary Medicine 104, 281-291.
Lambertz, C., Petig, M., Elkmann, A., Gauly, M., 2015. Confinement of sows for different
periods during lactation: effects on behaviour and lesions of sows and performance of
piglets. animal 9, 1373-1378.
Lawrence, A.B., Petherick, J.C., McLean, K.A., Deans, L.A., Chirnside, J., Gaughan, A.,
Clutton, E., Terlouw, E.M.C., 1994. The effect of environment on behaviour, plasma
cortisol and prolactin in parturient sows. Applied Animal Behaviour Science 39, 313-
330.
Li, Y.Z., Johnston, L., Hilbrands, A., 2010. Pre-weaning mortality of piglets in a bedded
group-farrowing system. Journal of Swine Health and Production 18.
Marchant, J.N., Broom, D.M., 1996. Effects of dry sow housing conditions on muscle weight
and bone strength. Animal science 62, 105-113.
Marchant, J.N., Broom, D.M., Corning, S., 2001. The influence of sow behaviour on piglet
mortality due to crushing in an open farrowing system. Animal science 72, 19-28.
Marchant, J.N., Rudd, A.R., Mendl, M.T., Broom, D.M., Meredith, M.J., Corning, S.,
Simmins, P.H., 2000. alternative and conventional farrowing systems. The Veterinary
Record 147, 214.
Milligan, B.N., Fraser, D., Kramer, D.L., 2002. Within-litter birth weight variation in the
domestic pig and its relation to pre-weaning survival, weight gain, and variation in
weaning weights. Livestock Production Science 76, 181-191.
Oliviero, C., Heinonen, M., Valros, A., Hälli, O., Peltoniemi, O.A.T., 2008. Effect of the
environment on the physiology of the sow during late pregnancy, farrowing and early
lactation. Animal reproduction science 105, 365-377.
Pedersen, L.J., Berg, P., Jørgensen, G., Andersen, I.L., 2011. Neonatal piglet traits of
importance for survival in crates and indoor pens. J. Anim. Sci. 89, 1207-1218.
6
Špinka, M., Illmann, G., de Jonge, F., Andersson, M., Schuurman, T., Jensen, P., 2000.
Dimensions of maternal behaviour characteristics in domestic and wildxdomestic
crossbred sows. Applied Animal Behaviour Science 70, 99-114.
Stolba, A., Wood-Gush, D.G.M., 1989. The behaviour of pigs in a semi-natural environment.
Animal Production 48, 419-425.
van Nieuwamerongen, S.E., Soede, N.M., van der Peet-Schwering, C.M.C., Kemp, B.,
Bolhuis, J.E., 2015. Development of piglets raised in a new multi-litter housing
system vs. conventional single-litter housing until 9 weeks of age. Journal of animal
science 93, 5442-5454.
Weary, D.M., Pajor, E.A., Fraser, D., Honkanen, A.-M., 1996. Sow body movements that
crush piglets: a comparison between two types of farrowing accommodation. Applied
Animal Behaviour Science 49, 149-158.
Weary, D.M., Phillips, P.A., Pajor, E.A., Fraser, D., Thompson, B.K., 1998. Crushing of
piglets by sows: effects of litter features, pen features and sow behaviour. Applied
Animal Behaviour Science 61, 103-111.
Wischner, D., Kemper, N., Krieter, J., 2009. Nest-building behaviour in sows and
consequences for pig husbandry. Livestock Science 124, 1-8.
7
CHAPTER ONE
Does housing influence maternal behaviour in sows?
Charlotte G.E. Grimberg-Henricia, Kathrin Büttnera, Christian Meyerb, Joachim Krietera
a Institute of Animal Breeding and Husbandry, Christian-Albrechts-University,
Olshausenstr. 40, D-24098 Kiel, Germany
b Chamber of Agriculture of Schleswig-Holstein,
Gutshof 1, D-24327 Blekendorf, Germany
Published in Applied Animal Behaviour Science, 2016, 180: 26-34
8
ABSTRACT
Farrowing crates prevent sows during lactation from moving freely and interacting
unrestrictedly with their piglets. The aim of this study was to compare sows during lactation
in a group-housing system (GH; n=23) and sows in a conventional, single-housing system
(SH; n=24) with regard to their maternal behaviour. GH sows were only fixed in their pens
three days ante partum until one day post partum. For the remaining amount of time they were
able to choose between their home pen and a shared running area. Piglets were able to leave
their pens on day five post partum. Data were collected in four batches with six sows in each
housing system. All sows were observed in week 2 and week 4 of lactation in six successive
tests concerning their maternal behaviour. The sows’ reaction to piglet distress calls,
separation from and reunion with their piglets was tested both in their home pens and in a test
arena for a maximum of five minutes. The test arena (3.9 m x 3.7 m) provided a piglet nest in
a corner. The sows were only able to hear and smell their piglets. In the piglet scream test in
the home pen, GH sows were more responsive to piglet screaming than SH sows. GH sows
showed more body movements towards their screaming piglets and aggressiveness towards
the experimenter (p<0.05) as well as stronger postural reactions at the end of the test; i.e.
standing (p<0.05). However, in the piglet scream test in the test arena, SH sows remained
near their handled piglet more frequently (p<0.05) and vocalised more frequently (p<0.05).
Whereas, GH sows tended to explore the test arena more (p<0.10). During the separation test
in the home pen, no behavioural differences between GH and SH sows could be obtained.
During the separation test in the test arena, all sows remained near the piglet nest with their
piglets. Furthermore, SH sows walked more (p<0.05), while GH sows explored the test arena
more frequently (p<0.05). In the reunion test in the home pen, GH sows tended to vocalise
more frequently (p<0.10). No behavioural differences could be found between GH and
SH sows in the reunion test in the test arena. Regarding total piglet losses (e.g. crushing,
underweight, runting, spay legs), GH sows had lower total losses compared to SH sows
(p<0.05). Furthermore, GH sows crushed fewer piglets than SH sows (p<0.05). To conclude,
GH sows showed stronger behavioural reactions in the home pen and SH sows in the test
arena. Thus, the housing system has an effect on maternal behaviour. Further research is
needed to obtain more information, if the significantly lower piglet losses of GH sows are
related to the stronger maternal reactions in the home pens of these sows and to the housing
conditions ante partum.
9
Keywords
Maternal behaviour; Group-housing system; Behavioural testing; Lactating sows
INTRODUCTION
Several studies under semi-natural conditions have shown that maternal behaviour in pigs has
not changed in the process of domestication, meaning the animals need to express maternal
behaviour for example in terms of nest building behaviour (Stolba and Wood-Gush, 1989;
Špinka et al., 2000; Damm et al., 2002). Furthermore, maternal behaviour is very important
for piglet survival and growth, due to the fact that it reduces starvation and crushing. These
are the most common reasons for piglet mortality, which are related to asphyxia and
hypothermia (Pedersen et al., 2011). However, the expression of maternal behaviour can vary
between sows (Špinka et al., 2000; Andersen et al., 2005). Several studies have shown that
sows with lower piglet mortality rates expressed more nest building behaviour
(Andersen et al., 2005; Wischner et al., 2009), were calmer during farrowing
(Andersen et al., 2005) and were more careful during lying down movements
(Burri et al., 2009). Furthermore, sows with fewer piglet losses performed more nose-to-nose
contacts with their piglets when they changed posture (Andersen et al., 2005). Another
important maternal characteristic of sows is their willingness to protect their piglets.
Aggressive sows showed a stronger behavioural response to separation from their piglets and
had also fewer piglet losses than less aggressive sows (Hellbrügge et al., 2008).
Previous studies have shown that maternal characteristics are measurable. In the piglet scream
test, the reaction of sows was tested when their piglets were handled and stimulated to scream
(Grandinson et al., 2003; Hellbrügge et al., 2008; Melišová et al., 2014). In the separation and
reunion test, sows were observed when they were separated from their piglets and again
reunited with their piglets (Pitts et al., 2002; Andersen et al., 2005; Hellbrügge et al., 2008).
Key variables in the maternal behaviour of sows recorded in these tests were for example the
responsiveness to piglet distress calls, restlessness and the amount of aggression
(Hellbrügge et al., 2008). Additionally, in the separation test, a note was made of whether
sows searched and called for their piglets (Pitts et al., 2002). The time that the sows spent
with nosing, the latency until nursing and the length of nursing was recorded during the
reunion of sows with their piglets (Pitts et al., 2002; Andersen et al., 2005).
10
In present pig husbandry, the most common housing system of sows at farrowing and during
lactation is the single-housing system with farrowing crates. In this system, sows are housed
individually and are confined in crates during lactation. Single-housing systems limit the
possibility of sows to move freely, to interact naturally with their piglets and to express
unhindered maternal behaviour. Furthermore, no contact between other sows and their litters
is possible. Solutions for more welfare for lactating sows are group-housing systems as
‘get-away’ systems, multi-suckling systems or a combination of these two systems, such as
reviewed by van Nieuwamerongen et al. (2014). The group-housing system described by
Bohnenkamp et al. (2013) is inspired by the natural rhythm of sows under semi-natural
conditions (e.g. separation from the group ante partum, building a nest, leaving the nest post
partum while piglets stay in the nest, returning to the group with the litter, social contacts with
other sows and litters) (Jensen, 1986).
To our knowledge, a combined analysis of the effect of housing conditions and maternal
behaviour is underrepresented. Therefore, the aim of this study was to analyse the effect of
group-housed and single-housed sows with regard to their maternal behaviour. Sows were
tested in week 2 and 4 of lactation in their home pens and in a test arena with regard to their
maternal reaction to piglet distress calls, separation from their piglets and reunion with their
piglets.
MATERIAL AND METHODS
Animals and housing
The study was conducted on the Futterkamp agricultural research farm of the Chamber of
Agriculture of Schleswig-Holstein over a period from May 2014 until September 2014. A
total of 47 cross-bred (Large White x Landrace) multiparous sows were tested during lactation
in a group-housing system (GH) and in a conventional, single-housing system (SH) with
regard to their maternal behaviour. All sows assigned to the GH system were normally housed
in the SH system. All litters were standardised to 13 piglets for each sow until two days
post partum. The GH system was designed as described by Bohnenkamp et al. (2013). In the
GH system, six sows were housed together (Figure 1). Each sow had a single pen
(1.8 m x 2.6 m) provided with a farrowing crate and electronically controlled gates. In
addition, all sows shared a running area (2.4 m x 5.4 m). The sows were only fixed in the
farrowing crates three days ante partum until one day post partum. A flat barrier prevented the
11
piglets from leaving the single pen, whereas the sow could continue to leave the pen. The
barrier was removed five days post partum and all litters of the six sows had the possibility to
mix in the running area. GH sows were fed by hand in their pens. Sows in the SH system were
fixed in farrowing crates (2.0 m x 2.6 m). SH sows were fed electronically. Sows in both
housing systems were fed a commercial lactating meal which increased constantly during
lactation to a maximum of 7.5 kg per day (divided into three to four meals). The piglets
received a commercial creep diet. Both housing systems had water-heated piglet resting areas
(0.6 m2) and manipulable material (e.g. plastic balls) was available for the sows and the
piglets. The temperature varied in the two housing systems between 19 and 21°C. Lights
(80 lx) were switched on at 6 am and off at 8 pm. In both housing systems, sows and piglets
shared a drinking bowl and had free access to water. Boars were not castrated and the piglets
were weaned on average 27.8 ± 0.15 days post partum.
Data were collected in four batches with six sows per batch in each housing system. During
gestation all sows were housed in a dynamic group with electronic feeding stations and were
randomly moved to the GH or SH system, respectively, one week before the calculated
farrowing date. When the sows did not farrow on the calculated date, the birth was initiated
with an injection of Prostaglandin F2α. The birth of 37 sows out of 47 sows was initiated. The
sows farrowed on average 0.8 ± 0.15 days later as the before calculated date. The sows did
not differ significantly concerning their number of parities between the two housing systems
(GH: 5.3 ± 0.47 vs. SH: 4.9 ± 0.40; n.s.) and within the batches (batch 1: GH 3.8 ± 0.91 vs.
SH 3.5 ± 0.56; n.s.; batch 2: GH 6.0 ± 0.68 vs. SH 6.0 ± 0.68; n.s.; batch 3: GH 6.8 ± 1.16 vs.
SH 5.5 ± 0.92; n.s.; batch 4: GH 4.8 ± 0.79 vs. SH 4.6 ± 0.80; n.s.).
Recorded traits
Before the sows were moved to the GH and SH systems, all sows were weighed, their back fat
(BF) was measured and their Body Condition Score (BCS) was recorded (7.8 ± 0.15 days
ante partum). The BF thickness (emaciated: <10 mm; thin: 11-15 mm; ideal: 16-18 mm;
fat: 19-22 mm; overly fat: >22 mm) was measured with an ultrasound scanner
(Agroscan, Hauptner & Heberholz, Solingen, Germany) 5 cm from the median line behind the
shoulder, in the middle of the back and before the hip (Bohnenkamp et al., 2013). The mean
of these three measurement points formed the BF value for each sow. The BCS ranged from
1 to 5 (1: emaciated (hips, spine prominent to the eye); 2: thin (hips, spine easily felt without
pressure); 3: ideal (hips, spine felt only with firm pressure); 4: fat (hips, spine cannot be felt);
5: overly fat (hips, spine heavily covered)). After weaning (27.8 ± 0.15 days post partum), the
12
same parameters were measured again for each sow. Furthermore, reproductive traits were
documented as the number of piglets born alive, stillborn piglets, weaned piglets, piglet losses
and individual weights of the piglets one day after birth and at weaning. Piglet losses were
recorded during the whole lactation period regarding cause, date, weight of the piglet and
location of the dead piglet for the GH system (home pen or shared running area). Total piglet
losses included piglets that were crushed and piglets that died due to other reasons
(e.g. underweight, runting, splay legs).
Figure 1
Schematic view of the two housing systems. The group-housing (GH) system with six
individual pens (4.7 m2) and running area (13 m2). The single-housing (SH) system with
individual pens (5.2 m2).
13
Figure 2
Schematic view of time line for maternal testing in the home pen (HP) and in the test arena
(TA).
Behavioural testing
All sows from both housing systems were tested in six different successive tests to investigate
their maternal behaviour. The sows were observed in behavioural tests in week 2
(12.8 ± 0.15 days post partum) and in week 4 (25.8 ± 0.15 days post partum) of lactation
(Figure 2). The tests were performed in the home pens (HP) of the sows and in a test arena
(TA). The TA was built to create a test situation for the sows, which was uncoupled from their
housing environment. Thus, the TA was a similar and new environment for all sows. The tests
started at 8 am and all sows were tested in a random order in each housing system. It was
blocked by treatment with a random order for the sows. The TA was in a separate room.
Because of logistical issues each housing system had its own TA. These TAs had the same
design and dimensions. The TA was 3.9 m x 3.7 m large and the non-transparent side walls
were 1.1 m high (Figure 3). The entrance for the sow was 1.2 m wide and there was a piglet
nest equipped with a heat lamp in one corner. The nest was 1.5 m x 1.5 m x 2.12 m large and
the non-transparent side walls were 1.1 m high. The front wall of the nest had a door
(0.8 m wide and 0.4 m high) and was connected with the inside of the TA. The TA was
divided into four parts (painted on the floor) with four possible localisations of the sows.
14
Piglet scream test in home pen
In the first test, the reaction of the sow to distress calls from her whole litter by catching them
was observed. The test was modified according to the experiment by Held et al. (2006). The
test showed the willingness of the sow to protect her piglets. Furthermore, the test
investigated the sow’s reactivity and responsiveness to piglet distress calls. The piglet distress
call was an intensive sound with a high frequency and was used to simulate a piglet in great
need (Andersen et al., 2005). Before the test started, GH sows and their piglets were confined
in their own HP. Two experimenters caught the piglets and put them in a transport carriage
while one of these experimenters documented the posture of the sow before the test started
and the posture of the sow when the last piglet was put in the transport carriage. The posture
of the sow could have two possible dimensions, standing or touching the floor with her body
(sitting, kneeling and lying). Standing was scored as strongest postural reaction of the sow.
Furthermore, the sow’s vocalisation (yes/no) was also noted and the maximum behavioural
response of the sow was evaluated in form of a behavioural grade at the end of the test
(0– no reaction; 1– weak reaction (head movement towards screaming piglets);
2– medium reaction (body movement towards screaming piglets); 3– strong reaction
(attack towards experimenter). One observation per sow and test was documented with regard
to posture and behavioural grade. During this test, one piglet was separated for the piglet
scream test in the TA, which was performed later in the experimental setup.
Separation test in test arena
After the piglet scream test, the sow was brought to the TA. The sow had ten minutes to
habituate to the new environment. After the habituation, the piglets were transported to the
TA and were put in the piglet nest in the corner. When the first piglet was put in the piglet
nest, the separation test started. The sow was able to hear and smell her piglets. The sow was
observed every ten seconds for a whole observation period of five minutes (31 observations
per sow and test) concerning localisation in the TA (Figure 3), posture, behaviour and
vocalisation (Table 1). The test design was modified according to previous experiments by
Pitts et al. (2002) and Andersen et al. (2005). The test showed how focused the sow was
towards her piglets during separation and investigated the responsiveness of the sow to her
piglets. Behaviour was preferred which was directed towards the piglets such as vocalisation,
occupation with the piglet nest, staying near the piglet nest and stress signals such as
restlessness.
15
Figure 3
Schematic view of the test arena (TA).
Reunion test in test arena
After the separation test, the reunion test was performed. For this, the door of the piglet nest
was opened to reunite the sow with her piglets. The test started when the door was opened.
The piglets were carefully pushed with a broom to leave the nest. When all piglets had left the
nest, the door was closed. The reaction of the sow was recorded again every ten seconds for
an observation period of five minutes (31 observations per sow and test) with regard to
localisation (Figure 3), posture, behaviour and vocalisation (Table 1). This test was a
modified version of the test by Pitts et al. (2002) and Andersen et al. (2005). The reaction of
the sow to her piglets was tested when the piglets were suddenly present. Behaviour which
focused on the piglets such as nose-to-nose contact with the piglets, running after the piglets,
nursing and vocalisation were evaluated as good maternal behaviour.
16
Piglet scream test in test arena
For the piglet scream test, the previously separated piglet was now taken to the TA where the
sow with the rest of her piglets was walking around. One experimenter stood with the piglet
outside the TA next to the piglet nest (localisation 1) and motivated the piglet to scream by
placing it in a back position. The reaction of the sow was observed again every ten seconds
for 30 seconds (four observations per sows and test) with regard to localisation (Figure 3),
posture, behaviour and vocalisation (Table 1). This test was similarly constructed compared to
the experiments of Grandinson et al. (2003). According to the piglet scream test in the HP,
this test investigated the willingness of the sow to protect her piglets and the reactivity and
responsiveness of the sow to a piglet distress call. The behaviour was scored as good maternal
behaviour when the sow turned towards her piglet using vocalisation, by staying near the
screaming piglet and by getting in contact with her other piglets.
Separation test in home pen
After the tests in the TA, the sow returned to her empty HP, i.e. without her piglets. The
observation started immediately when the sow had reached the empty HP (Pitts et al., 2002).
The behaviour of the sow, posture and vocalisation was recorded every ten seconds for five
minutes (31 observations per sow and test) (Table 1). The aim of this test was to investigate
whether the sow had searched for her piglets. Searching and stressed behaviour was
restlessness such as standing postures, exploring the floor for piglets and vocalisation.
Reunion test in home pen
The piglets were then transported from the TA to their HP and were placed with the sow. The
observation started when the first piglet was placed in the pen (Pitts et al., 2002). The
behaviour of the sow, posture and vocalisation were recorded every ten seconds for five
minutes (31 observations per sow and test) (Table 1). Similar to the reunion test in the TA, the
reaction of the sow was observed when her piglets were suddenly placed in the HP. Caring
behaviour as nursing, nose-to-nose contacts and vocalisation of the sow was evaluated as
good maternal behaviour.
17
Table 1
Ethogram for behavioural observation in the tests in the home pen (HP) and in the test arena
(TA) (modified according to Pitts et al. (2002) and Andersen et al. (2005)).
Variable Behaviour Location Description
Posture Active HP Sow stands or kneels down.
Inactive HP Sow sits or lies.
Active TA Sow walks or kneels down.
Inactive TA Sow stands, sits or lies.
Behaviour Exploring floor HP / TA Sow sniffs or roots with the nose within 0.2 m of
the floor.
Exploring nest TA Sow explores with her nose or scratches with her
front legs the front wall of the piglet nest in the
TA.
Exploring TA TA Sow sniffs or roots with the nose within 0.2 m of
the floor and explores with her nose or scratches
with her front legs the side walls of the TA.
Contact HP / TA Sow has contact with her nose with a piglet or
searches for contact by running after a piglet.
Nursing HP / TA Sow lies laterally. Suckling starts with udder
massage of 70 % of the piglets and is followed
by the milk flow. The suckling ends if less than
30 % of the piglets are present at the udder or if
the sow stopped lying on her side.
Vocalisation Vocalisation HP / TA Any vocalisation by the sow.
18
Statistical analysis
All data were analysed with the statistical software SAS® 9.2 (SAS Institute Inc., 2008). The
MIXED procedure was used for data with normal distribution. The GLIMMIX procedure was
applied for count data (poisson-distribution) and binomial data. Fixed effects were added in a
stepwise manner to the model. The Akaike’s information criteria corrected (AICC) and the
Baysian information criteria (BIC) were used to compare the different models. The model
with the smallest AICC and BIC values was chosen.
The data of the reproductive traits (birth and weaning weight of the piglets, number of piglets
born alive, stillborn, weaned, crushed piglets and total piglet losses) and the BW, BCS and BF
thickness of the sows were analysed with the MIXED procedure. The mixed model included
the fixed effects housing (GH, SH), batch (1, 2, 3, 4) and parity class (class 1: 2, 3;
class 2: 4-6; class 3: >6). The interaction between the fixed effects group and batch was tested
and was removed because of no significant effects and no improved goodness of the fit
criteria. The sow was added as random effect nested in group and batch to the model of birth
weights and weaning weights. In addition, the model for the weaning weights was added with
the birth weights and the lactation length as linear continuous variables. The model of BW,
BCS and BF thickness post partum were completed with the BW, BCS and BF thickness ante
partum as linear continuous variables. The significance of differences in the least square
means was adjusted with the Bonferroni-correction.
The count data of the behavioural tests (separation test in TA, reunion test in TA, piglet
scream test in TA, separation test in HP, reunion test in HP) were analysed using the
GLIMMIX procedure with a poisson-distribution with the log link function. For localisation
2, 3, 4 in the TA only few observation were found and were combined for analysis purposes
as one localisation 234. The model included the fixed effects housing (GH, SH), batch
(1, 2, 3, 4), parity class (class 1: 2, 3; class 2: 4-6; class 3: >6) and week (2, 4). The interaction
between the fixed effects group and batch was tested and was removed because of no
significant effects and no improved goodness of the fit criteria. The sow was added to the
model as random effect nested in group and batch. The significance of differences in the least
square means was adjusted with the Bonferroni-correction.
The binomial data of the piglet scream test in the HP were analysed using the GLIMMIX
procedure with a binomial distribution with the logit link function. For the behavioural grades
0 and 3 only few observations were documented and for analysis purposes the behavioural
grades of 0 and 1 were combined to grade 1 (no to weak reaction) and the grades 2 and 3 were
combined to grade 2 (medium to strong reaction). The model included the fixed effects
19
housing (GH, SH), batch (1, 2, 3, 4), parity class (class 1: 2, 3; class 2: 4-6; class 3: >6) and
week (2, 4). The interaction between the fixed effects group and batch was tested and was
removed because of no significant effects and no improved goodness of the fit criteria. The
sow was added to the model as random effect nested in group and batch. The significance of
differences in the least square means was adjusted with the Bonferroni-correction.
RESULTS
Reproductive traits
Table 2 shows the results for the reproductive traits of GH and SH sows. The traits total piglet
losses and crushed piglets differed between the two housing systems. The total percentage of
the piglet losses during lactation was 10.7 % for GH sows and 18.3 % for SH sows. 69.7 % of
all piglet losses from both housing systems occurred at birth and the first day post partum
(GH: 67.5 % vs. SH: 70.6 %). The sows from the GH system had significantly lower total
piglet losses at birth and the first day post partum (GH: 1.2 ± 0.30 vs. SH: 2.1 ± 0.30; p<0.05)
and during the whole lactation (GH: 1.8 ± 0.32 vs. SH: 2.9 ± 0.32; p<0.05). 52.8 % of all
piglets died due to crushing. Thereof, 66.6 % of these piglets were crushed within the day at
birth and the first day post partum. Sows from the GH system crushed at birth and the first
day post partum (GH: 0.2 ± 0.22 vs. SH: 1.1 ± 0.22; p<0.05) and during the whole lactation
(GH: 0.6 ± 0.27 vs. SH: 1.5 ± 0.26; p<0.05) significantly lower piglets compared to sows
from the SH system. The number of crushed piglets was marginal influenced by the parity
class. Sows from parity class 3 crushed marginally more piglets compared to sows from the
parity class 1 (class 1: 0.5 ± 0.35 vs. class 2: 1.0 ± 0.30 vs. class 3: 1.7 ± 0.40; p=0.0960).
Other causes that led to piglet mortality such as being underweight, runting, splay legs etc.
were greater for GH sows than for SH sows (GH: 65.9 % vs. SH: 50.7 %; n.s.). Furthermore,
the parity class 2 and 3 had more stillborn piglets compared to parity class 1
(class 1: 0.3 ± 0.38 vs. class 2: 1.5 ± 0.32 vs. class 3: 2.3 ± 0.41; p<0.05). Sows from the
parity class 1 weaned heavier piglets compared to sows from the parity class 2 and 3
(class 1: 8.1 kg ± 0.19 vs. class 2: 7.4 kg ± 0.17 vs. class 3: 7.76 kg ± 0.22; p<0.05).
20
Table 2
Least square means (LSM) and standard error (SE) of the reproductive traits of group-housed
(GH) and single-housed (SH) sows.
GH
(n=23)
LSM ± SE
SH
(n=24)
LSM ± SE
Piglets born alive / sow 16.6 ± 0.74 15.7 ± 0.73
Stillborn piglets / sow 0 1.1 ± 0.28 0 1.7 ± 0.28
Individual birth weight (kg) 0 1.3 ± 0.04 0 1.3 ± 0.04
Total piglet losses / sow 1.8a ± 0.32 2.9b ± 0.32
Crushed piglets / sow 0.6a ± 0.27 1.5b ± 0.26
Piglets weaned / sow 12.5 ± 0.12 12.2 ± 0.12
Individual weaning weight (kg) 0 7.6 ± 0.15 0 7.8 ± 0.15
a-b Significant differences between GH and SH sows (p<0.05).
Table 3
Least square means (LSM) and standard error (SE) of body weight (BW), Body Condition
Score (BCS) and back fat (BF) thickness of group-housed (GH) and single-housed (SH) sows.
GH
(n=23)
SH
(n=24)
Week 1
ante partum
LSM ± SE
Week 4
post partum
LSM ± SE
Week 1
ante partum
LSM ± SE
Week 4
post partum
LSM ± SE
Body weight (kg) 268.3 ± 4.44 231.0 ± 2.75 272.3 ± 4.38 236.0 ± 2.72
Body Condition Score 003.4 ± 0.05 002.9 ± 0.08 003.4 ± 0.05 002.9 ± 0.06
Back fat thickness (mm) 014.9 ± 0.56 012.1 ± 0.37 015.0 ± 0.55 012.9 ± 0.28
Body condition
The BW, the BCS and BF thickness of the sows were compared ante partum (7.8 ± 0.15 days
before farrowing) and post partum (27.8 ± 0.15 days after farrowing) between the two
housing systems. GH sows and SH sows did not differ statistically concerning these body
condition parameters (Table 3). On average, SH sows were slightly heavier ante partum
21
(GH: 268.3 kg ± 4.44 vs. SH: 272.3 kg ± 4.38; n.s.) and post partum
(GH: 231.0 kg ± 2.75 vs. SH: 236.3 kg ± 2.36; n.s.) compared to GH sows. The BW differed
significantly between the three parity classes ante partum (class 1: 250.9 kg ± 5.79 vs.
class 2: 269.3 kg ± 4.95 vs. class 3: 290.7 kg ± 6.23; p<0.05) and post partum
(class 1: 214.4 kg ± 3.95 vs. class 2: 226.9 kg ± 3.06 vs. class 3: 249.3 kg ± 4.31; p<0.05).
The BCS of sows from the parity class 1 was lower than the BCS of sows from the parity
class 2 and 3 post partum (class 1: 2.7 ± 0.08 vs. class 2: 2.9 ± 0.07 vs.
class 3: 3.0 ± 0.09; p<0.05). Furthermore, the BF thickness of sows from the parity class 3
was thicker compared to sows from the parity class 1 and 2 post partum
(class 1: 11.8 mm ± 0.41 vs. class 2: 12.3 mm ± 0.36 vs. class 3: 13.5 mm ± 0.45; p<0.05).
Behavioural testing
Table 4 shows the results of count data of the different behavioural maternal tests. The results
of the piglet scream test in in the HP are shown in the text. All tests are explained in more
detail below.
Piglet scream test in home pen
GH sows were obviously more responsive to the piglet scream test in the HP. 80 % of
GH sows received higher behavioural grades in the categories medium to strong reaction
(body movements towards their piglets and aggressiveness towards the experimenter)
compared to 40 % of SH sows when catching their piglets (GH: 0.8 ± 0.08 vs.
SH : 0.4 ± 0.10; p<0.05). In addition, at the end of the piglet scream test, 80 % of GH sows
showed stronger postural reactions, i.e. standing postures (GH: 0.8 ± 0.08 vs.
SH : 0.4 ± 0.10; p<0.05).
Separation test in test arena
SH sows were more active, i.e. walking, during separation from their piglets in the TA
(GH: 5.5 ± 0.61 vs. SH: 8.0 ± 0.82; p<0.05), whereas GH sows explored more the TA
(GH: 7.0 ± 1.32 vs. SH: 3.4 ± 0.72; p<0.05). Sows from both housing systems were near their
piglets more frequently than in the rest of the TA (localisation 234) during testing
(GH: 20.2 ± 1.37 vs. 7.8 ± 1.63; SH: 21.9 ± 1.37 vs. 5.5 ± 1.17). Furthermore, they vocalised
more (week 2: 21.0 ± 1.73 vs. week 4: 18.2 ± 1.52; p<0.05) and explored the TA less
frequently in week 2 than in week 4 of lactation (week 2: 4.5 ± 0.64 vs.
week 4: 5.7 ± 0.80; p<0.05).
22
Table 4
Least square means (LSM) of the frequencies of the count data with standard error (SEM) and
the percentage (%) of behavioural observations of group-housed (GH) and single-housed (SH)
sows tested in week 2 and 4 of lactation and behavioural observations of all sows for week 2
and week 4 tested in the separation test in the test arena (TA), reunion test in the TA, piglet
scream test in the TA, separation test in the home pen (HP) and reunion test in the HP.
Test Variable
GH
(n=23)
SH
(n=24)
Week 2
(n=47)
Week 4
(n=47)
LSM ± SEM % LSM ± SEM % % %
Separation test
in TA
5 min observation
(31 observations per sow)
Localisation 1 20.2 ± 1.37 66.9 21.9 ± 1.46 72.8 70.7 69.1
Localisation 234 07.8 ± 1.63 33.1 05.5 ± 1.17 27.2 29.3 30.9
Active 5.5a± 0.61 18.5 8.0b± 0.82 27.3 23.1 22.9
Inactive 25.1a± 0.74 81.5 22.4b± 0.69 72.7 76.9 77.1
Vocalisation 18.9 ± 2.14 66.9 20.2 ± 2.25 70.7 73.6 64.0
Exploring TA 7.0a± 1.32 28.4 3.4b± 0.72 17.1 20.0 25.3
Exploring nest 6.9 ± 0.70 23.3 08.6 ± 0.83 29.0 26.5 25.9
Reunion test
in TA
5 min observation
(31 observations per sow)
Localisation 1 09.5a± 0.84 29.7 12.3b± 0.78 39.7 34.9 34.7
Localisation 234 21.7a± 0.73 70.3 18.6b± 0.66 60.4 65.1 65.3
Active 00.9 ± 0.72 31.5 10.3 ± 0.79 35.7 39.1 28.2
Inactive 21.2 ± 0.78 68.5 20.0 ± 0.74 64.3 61.0 71.8
Vocalisation 06.0 ± 1.12 30.4 09.0 ± 1.63 43.6 52.5 21.8
Exploring TA 13.3 ± 0.84 44.5 11.9 ± 0.76 40.9 34.2 51.1
Contact 11.6 ± 0.69 39.8 10.8 ± 0.64 36.8 48.2 28.3
Piglet scream test
in TA
30 sec observation
(4 observations per sow)
Localisation 1 2.0a± 0.21 50.5 2.8b± 0.25 71.4 61.7 60.6
Localisation 234 1.9a± 0.27 49.5 1.1b±0.18 28.7 38.3 39.4
Active 01.3 ± 0.17 32.6 01.1 ± 0.16 29.2 34.6 27.1
Inactive 02.7 ± 0.24 67.4 02.8 ± 0.25 70.8 65.4 72.9
Vocalisation 1.8a± 0.23 47.3 2.6b± 0.29 67.2 65.4 49.5
Exploring TA 1.0c± 0.22 29.4 0.6d± 0.14 17.2 16.0 30.3
Contact 00.5 ± 0.11 15.8 00.5 ± 0.10 15.1 21.3 09.6
Separation test
in HP
5 min observation
(31 observations per sow)
Active 20.9 ± 1.89 71.0 20.0 ± 1.79 69.2 73.9 66.3
Inactive 04.8 ± 1.83 29.0 02.9 ± 1.17 30.8 26.1 33.7
Vocalisation 09.8 ± 2.89 46.7 05.4 ± 1.61 35.6 48.7 33.4
Exploring floor 08.4 ± 0.98 29.9 07.3 ± 0.84 25.7 31.4 24.2
Reunion test
in HP
5 min observation
(31 observations per sow)
Active 10.0 ± 2.14 39.2 06.9 ± 1.49 34.0 42.1 31.0
Inactive 16.9 ± 2.27 60.8 17.9 ± 2.38 66.0 57.9 69.1
Vocalisation 15.3c± 1.78 55.6 11.3d± 1.32 40.9 56.7 39.7
Contact 06.8 ± 0.57 22.6 05.8 ± 0.50 19.6 23.5 18.7
Nursing 02.6 ± 1.20 24.3 01.5 ± 0.71 17.4 27.9 13.7
a-b Significant differences between GH and SH sows (p<0.05). c-d Marginal differences between GH and SH sows (p<0.10).
23
Reunion test in test arena
SH sows were more frequent in localisation 1 than GH sows during the reunion test in the TA
(GH: 9.5 ± 0.84 vs. SH: 12.3 ± 0.78; p<0.05). Sows from both housing systems were more
active (week 2: 11.4 ± 0.70 vs. week 4: 8.3 ± 0.55; p<0.05), vocalised more
(week 2: 11.4 ± 1.50 vs. week 4: 4.7 ± 0.65; p<0.05) and had more frequent contact with their
piglets in week 2 compared to week 4 of lactation (week 2: 14.6 ± 0.70 vs.
week 4: 8.6 ± 0.49; p<0.05). Furthermore, both groups explored the TA more in week 4
compared to week 2 (week 2: 10.3 ± 0.58 vs. week 4: 15.4 ± 0.78; p<0.05). Regarding the
different parity classes, sows from the parity class 3 were less active (class 1: 10.9 ± 1.09 vs.
class 2: 10.8 ± 0.92 vs. class 3: 7.8 ± 0.89; p<0.05) and vocalised more compared to the parity
classes 1 and 2 (class 1: 5.1 ± 1.25 vs. class 2: 6.2 ± 1.31 vs. class 3: 12.4 ± 3.15; p<0.05).
Piglet scream test in test arena
SH sows remained near their piglets (localisation 1) significantly more frequent
(GH: 2.0 ± 0.21 vs. SH: 2.8 ± 0.25; p<0.05) and vocalised more compared to GH sows during
the piglet scream test in the TA (GH: 1.8 ± 0.23 vs. SH: 2.6 ± 0.29; p<0.05), whereas, GH
sows tended to explore the TA more (GH: 1.0 ± 0.22 vs. SH: 0.6 ± 0.14; p=0.0679).
Furthermore, sows of both housing systems vocalised more (week 2: 2.5 ± 0.26 vs.
week 4: 1.9 ± 0.22; p<0.05), explored the TA less (week 2: 0.5 ± 0.12 vs.
week 4: 1.0 ± 0.19; p<0.05) and had more contact with their piglets in week 2 compared to
week 4 of lactation (week 2: 0.8 ± 0.13 vs. week 4: 0.4 ± 0.09; p<0.05).
Separation test in home pen
No behavioural differences could be observed between GH and SH sows during reunion with
their piglets in their HP. Sows from both housing systems were more active
(week 2: 21.6 ± 1.45 vs. week 4: 19.4 ± 1.27 1.31; p<0.05), vocalised more
(week 2: 8.8 ± 1.86 vs. week 4: 6.0 ± 1.28; p<0.05) and explored the floor more frequently in
week 2 compared to week 4 of lactation (week 2: 8.9 ± 0.78 vs. week 4: 6.8 ± 0.63; p<0.05).
Reunion test in home pen
GH sows tended to call more frequently their piglets during reunion in the HP compared to
SH sows (GH: 15.3 ± 1.78 vs SH: 11.3 ± 1.32; p=0.0727). Sows from both housing systems
were more active (week 2: 9.7 ± 1.50 vs. week 4: 7.2 ± 1.12; p<0.05) and vocalised more
frequently in week 2 compared to week 4 of lactation (week 2: 15.8 ± 1.35 vs.
24
week 4: 11.0 ± 0.97; p<0.05). Furthermore, both groups had more contacts with their piglets
(week 2: 7.1 ± 0.49 vs. week 4: 5.6 ± 0.42; p<0.05) and nursed their piglets more frequently
in week 2 compared to week 4 of lactation (week 2: 3.0 ± 0.94 vs. week 4: 1.4 ± 0.47;
p<0.05).
DISCUSSION
Reproductive traits
In the present study sows housed in the GH system had significantly lower total piglet losses
and crushed fewer piglets than SH sows. Two thirds of the total piglet losses of both housing
systems occurred at birth and at the first day post partum. Previous research conducted in
these two housing systems found no significant differences with regard to total piglet losses of
GH and SH sows and they reported on more crushed piglets of GH sows, however, this was
not statistically confirmed (Bohnenkamp et al., 2013). The design and the management of the
GH system were not altered between the study of Bohnenkamp et al. (2013) and the present
study. In the study of Bohnenkamp et al. (2013) an explanation for more crushed piglets in the
GH system was that the pens of GH sows were smaller (0.52 m2) and, thus, the piglets had
less space to react to posture changes of the sow. Li et al. (2010) reported lower piglet
mortality rates in a GH system over several years. Here, an explanation was that the
stockpersons had improved their skills in the course of time with regard to better observation
and handling of the sows and, thus, had acquired a better approach to detecting problems.
When the present study started, the GH system had already been implemented on the research
farm for five years. The experience of the stockpersons could have had an effect on the piglet
mortality rates in the present study. Furthermore, the sows in the GH systems were only fixed
for three days ante partum and one day post partum, whereas, SH sows were fixed the whole
time ante partum and post partum. GH sows had four days of free movement before
farrowing. This time of free movement could probably have an influence on the physically
and psychologically balance of the GH sows around farrowing. Lower piglet mortality rates
are associated with sows that are calmer during farrowing (Andersen et al., 2005) and sows
that are more careful during lying down movements (Burri et al., 2009). Further research
could investigate if results were comparable when the sows of both housing systems were
confined directly when they moved to the farrowing stable.
25
Furthermore, in the present study a tendency was found that older sows crushed more piglets
than younger sows. A reason for more piglet losses due to crushing by older sows could be
the bigger size of the sows. An high incidence of more crushed piglets of sows from higher
parity classes were also found in a study from Weary et al. (1998).
No differences were found between GH and SH sows with regard to other reproductive traits.
These results are in accordance with previous research, except for the individual weaning
weight of the piglets, which was significant lower for the piglets in the GH system in the
study of Bohnenkamp et al. (2013). Here, the lower weaning weight of piglets from GH sows
was explained by the greater amount of missed nursings and playful behaviour of these
piglets. In the present study no differences concerning the weaning weight of the piglets were
found. Creep feed was offered in both housing systems directly three days after birth. Perhaps
the early creep feed offering could counteract the weight losses of the piglets caused by the
enhanced activity in the GH system.
Behavioural testing
In the series of tests to investigate the maternal behaviour, differences in maternal reactions
could be obtained between GH and SH sows. These findings allow to conclude that housing
conditions do influence maternal behaviour in sows. GH sows showed significantly stronger
maternal reactions when they were tested in the HP, compared to SH sows when they were
observed in the TA.
However, all GH sows were usually housed in a SH system, an effect of the GH system on
their maternal behaviour could already be observed in their first litter in the GH system. All
sows were tested for their first time in the described maternal tests and, thus, it was for all
sows a new situation. Additionally, the TA was used to create a test situation that was
uncoupled from their housing environment. Nevertheless, the GH sows experienced the GH
system for the first time, whereas, SH sows were usually housed in the SH system. A
possibility to correct for that effect in further studies could be to use nulliparous sows which
can then be observed over several lactations. However, this also increases the risk of
habituation to the test procedure and the diversity of sows with different number of parities is
lost. Furthermore, the experimenter was not blind to the treatment groups in the present study.
However, a good structured and short ethogram was used with behaviours that were clearly
defined to avoid misinterpretations.
In the piglet scream test in the HP, GH sows were significantly more responsive to the piglet
distress calls of their piglets than SH sows. In the study of Arey and Sancha (1996), sows
26
housed in a family system showed also stronger postural reactions by standing up during a
piglet scream test compared to SH sows. With regard to piglet crushing, the sensibility and
responsiveness of the sows play a central role for piglet survival. For instance,
Weary et al. (1996) showed that piglet survival increased immensely when the sow reacted to
a trapped piglet within one minute. It was found that the scream of a piglet is the most
important stimulus for a sow to react (Hutson et al., 1991). Sows that showed a strong
response to a screaming piglet were also calmer and showed fewer risky posture changes after
birth (Thodberg et al., 2002). The positive correlation between high responsiveness and low
piglet mortality caused by crushing are in accordance to the results of the present study.
Although most piglet losses due to crushing occurred in the first days post partum, several
behavioural studies conducted maternal tests later during lactation. For example Andersen et
al. (2005) divided first two groups of sows into crushers and non-crushers regarding their
piglet losses due to crushing. At day seven and eight post partum, these sows were tested in a
piglet distress call test to investigate their maternal reaction. Also Held et al. (2006)
performed the maternal tests at day five and six post partum regarding piglet losses which
occurred in the first days post partum. Pitts et al. (2002) tested sows also seven days and one
day before weaning concerning their maternal performances in several maternal tests.
GH sows were highly responsive to the piglet distress calls in the test situation in the HP and
crushed also significant lower piglets in the course of this study. A reason for the high
responsiveness of GH sows could be that this is a necessity when housed in a GH system. In
GH systems, different sows and piglets live together and, thus, it is important for sows to
identify their natural piglets to allocate their maternal investment, whereas SH sows have no
doubt that the piglets around them are their natural piglets. This could also explain the more
frequent vocalisation of GH sows in the reunion test in the HP. A stronger and more constant
communication between sows and their piglets were also observed in sows housed in
‘get-away’ and family systems compared to confined sows (Arey and Sancha, 1996;
Pitts et al., 2002).
SH sows remained near their screaming piglet more frequently and vocalised more compared
to GH sows in the piglet scream test in the TA. Furthermore, they seemed to be more strained,
by walking around during the separation test in the TA, whereas GH sows explored the TA
more frequently. An explanation that SH sows were more active could be that they were not
used to leave their home pen and move freely compared to GH sows. A reason that SH sows
were more stressed compared to GH sows could be that SH sows were not used to separation
from their piglets. However, GH sows were more experienced in being separated from their
27
piglets as long as they could hear their piglets. In GH systems, it is common for sows and
piglets not to be together the whole time. While the sow uses the running area, her piglets
could rest in the pen for example. However, in the SH system the sows are used to always be
surrounded by their piglets.
Furthermore, it was observed in this study that the behavioural reactivity of the sows of the
two housing systems changed in the course of time (between week 2 and 4). In all tests, the
behavioural tension of the sows decreased. In week 4 the sows vocalised less frequently, were
less active, had less contact with their piglets, nursed less frequently and explored the testing
environment more often. A possible explanation could be that the sows became habituated to
the test procedure. However, another reason for the decrease in behavioural tension and
maternal care could be the beginning of the weaning process (Jensen et al., 1991; Bøe, 1993).
Bøe (1993) observed a clear decline in the nursing frequency of sows and the time that sows
spent with their piglets between the second and fourth week of lactation. However, the
behavioural tension in the piglet scream test in the HP increased between week 2 and 4 of
testing. The piglet distress call seems to remain the most important stimulus for sows to react
(Hutson et al., 1991) .
Regarding the management in the housing systems, GH sows had probably more contact to
the stockperson than SH sows, especially because of the architecture of the GH system and
the feeding procedure. However, there was no more interaction between feeder and sow
regarding physical or verbal attention for the sows to minimize the differences between the
two housing systems. Feeding is a positive association for pigs with a stockperson
(Hemsworth et al., 1996). However, if an interaction between sows and stockperson during
feeding would have an influence on the maternal performance in the tests, GH sows would
react less tense in the piglet scream test in the home pen. In the present study, GH sows
reacted significantly stronger in the piglet scream test in the home pen compared to SH sows.
28
CONCLUSION
It could be concluded that housing conditions influence maternal behaviour in sows. In the
present study, sows of both housing systems showed good maternal behaviour, depending on
the tests carried out. GH sows showed good maternal behaviour in the HP and SH sows
showed good maternal behaviour in the TA. However, GH sows had significantly fewer piglet
losses. One explanation for the lower mortality rates in the GH systems could be that the
stockpersons themselves were more familiar with the GH system. Furthermore, the possibility
of free movement before farrowing could support more physical and psychological balance of
GH sows, which could influence piglet mortality. Further research is needed to obtain more
information, if the significantly lower piglet losses of GH sows are related to the stronger
maternal reactions in the home pens of these sows and to the housing conditions ante partum.
29
REFERENCES
Andersen, I.L., Berg, S., Bøe, K.E., 2005. Crushing of piglets by the mother sow (Sus scrofa)
- purely accidental or a poor mother? Applied Animal Behaviour Science 93, 229-243.
Arey, D.S., Sancha, E.S., 1996. Behaviour and productivity of sows and piglets in a family
system and in farrowing crates. Applied Animal Behaviour Science 50, 135-145.
Bøe, K., 1993. Maternal behaviour of lactating sows in a loosehousing system. Applied
Animal Behaviour Science 35, 327-338.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013. Group housing for
lactating sows with electronically controlled crates: 1. Reproductive traits, body
condition, and feed intake. Journal of animal science 91, 3413-3419.
Burri, M., Wechsler, B., Gygax, L., Weber, R., 2009. Influence of straw length, sow
behaviour and room temperature on the incidence of dangerous situations for piglets in
a loose farrowing system. Applied Animal Behaviour Science 117, 181-189.
Damm, B.I., Bildsøe, M., Gilbert, C., Ladewig, J., Vestergaard, K.S., 2002. The effects of
confinement on periparturient behaviour and circulating prolactin, prostaglandin F2α
and oxytocin in gilts with access to a variety of nest materials. Applied Animal
Behaviour Science 76, 135-156.
Grandinson, K., Rydhmer, L., Strandberg, E., Thodberg, K., 2003. Genetic analysis of on-
farm tests of maternal behaviour in sows. Livestock Production Science 83, 141-151.
Held, S., Mason, G., Mendl, M., 2006. Maternal responsiveness of outdoor sows from first to
fourth parities. Applied Animal Behaviour Science 98, 216-233.
Hellbrügge, B., Tölle, K.H., Bennewitz, J., Henze, C., Presuhn, U., Krieter, J., 2008. Genetic
aspects regarding piglet losses and the maternal behaviour of sows. Part 2. Genetic
relationship between maternal behaviour in sows and piglet mortality. animal 2, 1281-
1288.
Hemsworth, P.H., Verge, J., Coleman, G.J., 1996. Conditioned approach-avoidance responses
to humans: the ability of pigs to associate feeding and aversive social experiences in
the presence of humans with humans. Applied Animal Behaviour Science 50, 71-82.
Hutson, G.D., Wilkinson, J.L., Luxford, B.G., 1991. The response of lactating sows to tactile,
visual and auditory stimuli associated with a model piglet. Applied Animal Behaviour
Science 32, 129-137.
Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs.
Applied Animal Behaviour Science 16, 131-142.
30
Jensen, P., Stangel, G., Algers, B., 1991. Nursing and suckling behaviour of semi-naturally
kept pigs during the first 10 days postpartum. Applied Animal Behaviour Science 31,
195-209.
Li, Y.Z., Johnston, L., Hilbrands, A., 2010. Pre-weaning mortality of piglets in a bedded
group-farrowing system. Journal of Swine Health and Production 18.
Melišová, M., Illmann, G., Chaloupková, H., Bozděchová, B., 2014. Sow postural changes,
responsiveness to piglet screams, and their impact on piglet mortality in pens and
crates. Journal of animal science 92, 3064-3072.
Pedersen, L.J., Berg, P., Jørgensen, G., Andersen, I.L., 2011. Neonatal piglet traits of
importance for survival in crates and indoor pens. J. Anim. Sci. 89, 1207-1218.
Pitts, A.D., Weary, D.M., Fraser, D., Pajor, E.A., Kramer, D.L., 2002. Alternative housing for
sows and litters.: Part 5. Individual differences in the maternal behaviour of sows.
Applied Animal Behaviour Science 76, 291-306.
SAS Institute Inc., 2008. User's Guide (release 9.2). Cary, North Carolina, USA.
Špinka, M., Illmann, G., de Jonge, F., Andersson, M., Schuurman, T., Jensen, P., 2000.
Dimensions of maternal behaviour characteristics in domestic and wildxdomestic
crossbred sows. Applied Animal Behaviour Science 70, 99-114.
Stolba, A., Wood-Gush, D.G.M., 1989. The behaviour of pigs in a semi-natural environment.
Animal Production 48, 419-425.
Thodberg, K., Jensen, K.H., Herskin, M.S., 2002. Nursing behaviour, postpartum activity and
reactivity in sows: Effects of farrowing environment, previous experience and
temperament. Applied Animal Behaviour Science 77, 53-76.
van Nieuwamerongen, S.E., Bolhuis, J.E., van der Peet-Schwering, C.M.C., Soede, N.M.,
2014. A review of sow and piglet behaviour and performance in group housing
systems for lactating sows. animal 8, 448-460.
Weary, D.M., Pajor, E.A., Fraser, D., Honkanen, A.-M., 1996. Sow body movements that
crush piglets: a comparison between two types of farrowing accommodation. Applied
Animal Behaviour Science 49, 149-158.
Weary, D.M., Phillips, P.A., Pajor, E.A., Fraser, D., Thompson, B.K., 1998. Crushing of
piglets by sows: effects of litter features, pen features and sow behaviour. Applied
Animal Behaviour Science 61, 103-111.
Wischner, D., Kemper, N., Krieter, J., 2009. Nest-building behaviour in sows and
consequences for pig husbandry. Livestock Science 124, 1-8.
31
CHAPTER TWO
What do maternal tests actually test?
Charlotte G.E. Grimberg-Henricia, Irena Czycholla, Onno Burfeindb, Joachim Krietera
a Institute of Animal Breeding and Husbandry, Christian-Albrechts-University,
Olshausenstr. 40, D-24098 Kiel, Germany
b Chamber of Agriculture of Schleswig-Holstein,
Gutshof 1, D-24327 Blekendorf, Germany
Published in Applied Animal Behaviour Science, 2017, 189: 23-28
32
ABSTRACT
Several studies have used behavioural tests to characterise sows regarding their maternal
performance. These studies have always chosen a selection of behavioural tests and the
combination of tests varied between the studies. In the present study, 47 sows were tested in
week 2 and 4 of lactation in five successive maternal tests, which were conducted in their
home pens and in a test arena. The sows’ reaction to piglet stress signals, separation from and
reunion with their piglets were tested. The behavioural parameters observed in these maternal
tests were examined by means of a factor analysis to identify redundancies in behavioural
parameters. Five factors were extracted, explaining together 89.2 % of the total variance, of
which Factor 1 explains 33.5 %, Factor 2 18.4 %, Factor 3 14.5 %, Factor 4 12.6 % and
Factor 5 10.2 %. The interpretation of the factor loadings revealed four underlying maternal
factors: communication, care, contact and local attachment. Communication, thus
vocalisation, was extracted by the factor analysis as the most important maternal factor and is
used in sows to call their piglets in threatening situations as in the separation test and the
piglet scream test. Furthermore, communication plays a role for the maternal contact and care
factors. Sows used vocalisation to contact their piglets using nose-to-nose contacts, which
were observed in the reunion test and also in the piglet scream test. With regard to the care
factor, sows use vocalisation to call their piglets for nursing and to synchronise nursing with
other sows. The willingness to stay with their piglets, thus local attachment to their piglets,
was shown by sows in the separation and reunion test in the test arena and the piglet scream
test. Furthermore, the factor analysis proves that single maternal tests can also combine more
than one maternal factor and that the experimental environment in which the tests are placed
influences the significance of the tests. The present results demonstrate the complexity and
diversity of maternal behaviour.
Keywords
Factor analysis; Maternal behaviour; Behavioural testing; Sow
33
INTRODUCTION
Piglet performance such as survival and growth is greatly dependent on the quality of
maternal behaviour. For instance, sows with lower piglet mortality rates are known to show
more maternal behaviour such as more nest building (Andersen et al., 2005), expressing more
nose-to-nose contacts with their piglets during posture changes (Andersen et al., 2005) and
being more careful when lying down (Burri et al., 2009).
Maternal behaviour which leads to decreased piglet mortality and to increased piglet growth is
an important breeding goal. Different tests have been used to investigate maternal behaviour.
The piglet scream test is a commonly used maternal test in which the reaction of the sow
towards a screaming piglet (Grandinson et al., 2003; Løvendahl et al., 2005;
Melišová et al., 2014) or a sound (Grandinson et al., 2003) is tested. The piglet handling test
is a maternal test related to the piglet scream test, in which the reaction of the sow to the
handled piglet is observed (Grandinson et al., 2003; Held et al., 2006). Furthermore, sows are
tested in piglet-separation tests to investigate whether they search for their piglets, for
example by being more restless and by vocalising more frequently (Andersen et al., 2005;
Hellbrügge et al., 2008). A separation test is often followed by a reunion test in which the
reactions of the sows are tested when their piglets are returned to them (Pitts et al., 2002;
Andersen et al., 2005).
Most of the studies which have investigated maternal behaviour have used a selection of these
tests. To our knowledge, it has still not been proven which types of maternal behaviour these
tests evaluate and what these tests exactly test for. Furthermore, it is not known which
combinations of maternal tests and what types of tests are advisable.
The aim of this present study was therefore to give new insights into the structure of the
maternal behaviour of sows and to obtain more information on the significance of these tests.
To do this, sows were tested in week 2 and 4 of lactation in their home pens and in a test
arena in a piglet scream test, a separation test and a reunion test. As there are redundancies
between the tests, factor analysis was used as a helpful tool to investigate the actual
underlying structure of behaviour. This method has been used in animal welfare science, for
instance as a tool for lameness detection (Miekley et al., 2013) as well as for the evaluation of
emotional states in animals (Temple et al., 2011; Wemelsfelder et al., 2012).
34
MATERIAL AND METHODS
Animals and housing
The study was conducted on Futterkamp, the research farm of the Chamber of Agriculture of
Schleswig-Holstein from May 2014 until September 2014. A total of 47 cross-bred
(Large White x Landrace) multiparous sows were tested during lactation in a group-housing
system (GH) and in a conventional, single-housing system (SH) with regard to their maternal
behaviour (Grimberg-Henrici et al., 2016). The treatment groups were used in a previous
study (Grimberg-Henrici et al., 2016). In the present study, the different treatment groups
(SH, GH) were not considered for data analysis and thus were summarized as research
population. Data were collected in four batches with 12 sows per batch. It was considered that
the sows did not differ to a large extent concerning their number of parities within and
between the batches. All litters were standardised to 13 piglets for each sow until two days
post partum. In the GH system, six sows were housed together. Each sow had a single pen
(1.8 m × 2.6 m) provided with a farrowing crate and electronically controlled gates. In
addition, all sows shared a running area (2.4 m × 5.4 m). The sows were only fixed in the
farrowing crates from three days ante partum to one day post partum. A flat barrier prevented
the piglets from leaving the single pen, whereas the sow could leave the pen. The barrier was
removed five days post partum and all litters of the six sows had the possibility to mix in the
running area. Sows in the SH system were fixed in farrowing crates (2.0 m × 2.6 m). Boars
were not castrated and the piglets were weaned on average 27.8 ± 0.15 days post partum. The
farrowing stables had water-heated piglet resting areas (0.6 m2) and manipulable material
(e.g. plastic balls) was available for the sows and the piglets. The temperature in the stables
varied between 19 and 21 °C. Lights (80 lx) were switched on at 6 am and off at 8 pm. Sows
were fed a commercial lactating meal, the amount of which was increased constantly during
lactation to a maximum of 7.5 kg per day. The piglets received a commercial creep diet. Sows
and piglets shared a drinking bowl and had free access to water.
35
Figure 1
Schematic view of the test arena.
Behavioural testing
All sows were tested in five different successive tests to investigate their maternal behaviour.
For the sequence of testing, which is described below, was chosen for reasons of animal
welfare, feasibility and practical relevance. The sows were observed in behavioural tests in
week 2 (12.8 ± 0.15 days post partum) and in week 4 (25.8 ± 0.15 days post partum) of
lactation. The tests were performed in the home pens (HP) of the sows and in a test arena
(TA) (Figure 1). The TA was built to create a test situation for the sows, which was uncoupled
from their housing system and, thus, a new environment for all sows. The tests started at 8 am
and all sows were tested in a random order. The TA was in a separate room and was
3.9 m x 3.7 m large and the opaque side walls were 1.1 m high. The entrance for the sow was
1.2 m wide and there was a piglet nest in one corner. The nest was 1.5 m x 1.5 m x 2.12 m
large and the opaque side walls were 1.1 m high. The front wall of the nest had a door
(0.8 m wide and 0.4 m high) and was connected with the inside of the TA. The TA was
divided into four parts with four possible locations of the sows.
36
Separation test in test arena
The sow was brought to the TA and had ten minutes to habituate to the new environment.
After habituation, the piglets were transported to the TA and were put in the piglet nest.
Before transportation, one piglet was separated for the piglet scream test in the TA, which was
performed later in the experimental set-up. During the separation test the sow was able to hear
and smell her piglets. The separation test started when the first piglet was put in the piglet
nest. The sow was observed every ten seconds for a whole observation period of five minutes
(31 observations per sow and test) concerning location in the TA (Figure 1), posture,
behaviour and vocalisation (Table 1). The test design was modified according to previous
experiments by Pitts et al. (2002) and Andersen et al. (2005). The test was supposed to
showed how focused the sow was towards her piglets during separation and investigated the
responsiveness of the sow to her piglets. We concentrated on behaviour directed towards the
piglets such as vocalisation, occupation with the piglet nest, staying near the piglet nest and
stress signals.
Reunion test in test arena
After the separation test, the reunion test was performed: the door of the piglet nest was
opened to reunite the sow with her piglets. The test started when the door was opened. The
piglets were carefully pushed with a broom to leave the nest. When all piglets had left the
nest, the door was closed. The reaction of the sow was recorded again every ten seconds for
an observation period of five minutes (31 observations per sow and test) with regard to
location (Figure 1), posture, behaviour and vocalisation (Table 1). This test was a modified
version of the test by Pitts et al. (2002) and Andersen et al. (2005). In this test the reaction of
the sow to her piglets was observed when the piglets were suddenly reunited. Behaviour
which focused on the piglets such as nose-to-nose contact with the piglets, running after the
piglets, nursing and vocalisation were evaluated as good maternal behaviour.
Piglet scream test in test arena
For the piglet scream test, the previously separated piglet was now taken to the TA where the
sow with the rest of her piglets was walking around. One experimenter stood outside the TA
with the piglet next to the piglet nest (location 1) and motivated the piglet to scream by
placing it on its back. The reaction of the sow was observed again every ten seconds for 30
seconds (four observations per sow and test) with regard to location (Figure 1), posture,
behaviour and vocalisation (Table 1). This test was similarly constructed compared to the
37
experiments of Grandinson et al. (2003). It investigated the willingness of the sow to protect
her piglets and the reactivity and responsiveness of the sow to a piglet distress call. The
behaviour was scored as good maternal behaviour when the sow turned towards her piglet
using vocalisation, by staying near the screaming piglet and by coming into contact with her
other piglets.
Separation test in home pen
After the tests in the TA, the sow returned to her empty HP, i.e. without her piglets. The
observation started immediately when the sow had reached the empty HP (Pitts et al., 2002).
The behaviour of the sow, posture and vocalisation were recorded every ten seconds for five
minutes (31 observations per sow and test) (Table 1). The aim of this test was to investigate
whether the sow had searched for her piglets. Restlessness such as standing postures,
searching the ground for piglets and calling for the piglets was evaluated as searching and
stressed behaviour.
Reunion test in home pen
The piglets were then transported from the TA to their HP and were placed with the sow. The
observation started when the first piglet was placed in the pen (Pitts et al., 2002). The
behaviour of the sow, posture and vocalisation was recorded every ten seconds for five
minutes (31 observations per sow and test) (Table 1). Similar to the reunion test in the TA, the
reaction of the sow was observed when her piglets suddenly entered the HP. Caring behaviour
as nursing, nose-to-nose contacts and vocalisation of the sow was evaluated as good maternal
behaviour.
Statistical analysis
The count data of the behavioural tests was analysed with a factor analysis by using the
statistical software SAS® 9.4 (SAS Institute Inc., 2008). The PROC FACTOR procedure was
carried out, whereby a promax rotation was applied to facilitate the interpretation of results.
Each behavioural parameter achieved a certain loading on each factor, which is a number
between -1 and 1. Factor loadings were defined as significant if they were greater than or
equal to 0.4 (highly positive) and if they were less than or equal to -0.4 (highly negative). This
procedure and the interpretation was based on O'Rourke et al. (2013).
For location 2, 3, 4 (away from the piglet nest) in the TA only a few observations were
performed and were therefore combined for analysis purposes as one location 234.
38
Table 1
Ethogram for behavioural observation in the maternal tests in the home pen (HP) and in the
test arena (TA) (modified according to Pitts et al. (2002) and Andersen et al. (2005)).
Variable Behaviour Location Description
Posture Active HP Sow stands or kneels down.
Inactive HP Sow sits or lies.
Active TA Sow walks or kneels down.
Inactive TA Sow stands, sits or lies.
Behaviour Exploring floor HP Sow sniffs or roots with the nose within 0.2 m of
the floor.
Exploring nest TA Sow explores with her nose or scratches with her
front legs the front wall of the piglet nest in the
TA.
Exploring TA TA Sow sniffs or roots with the nose within 0.2 m of
the floor and explores with her nose or scratches
with her front legs the side walls of the TA.
Contact HP / TA Sow has contact with her nose with a piglet or
searches for contact by running after a piglet.
Nursing HP / TA Sow lies laterally. Suckling starts with udder
massage of 70 % of the piglets and is followed
by the milk flow. The suckling ends if less than
30 % of the piglets are present at the udder or if
the sow stopped lying on her side.
Vocalisation Vocalisation HP / TA Any vocalisation by the sow.
39
RESULTS
Table 2 demonstrates the median, minimum and maximum of the frequencies and the
percentage of the behavioural parameters in each maternal test. In the separation test 68.8 %
of the sows vocalised, 69.9 % were near the piglet nest in location 1, 26.2 % explored the
piglet nest and 22.6 % explored the TA. In the reunion test in the TA 37.1 % of the sows
vocalised, 34.8 % were near the piglet nest in location 1, 33.6 % were active, 38.2 % had
contact with their piglets and 42.6 % explored the TA. In the third test, the piglet scream test
in the TA, 57.5 % of the sows vocalised, 61.2 % were near their screaming piglet in
location 1, 30.9 % were active, 15.4 % had contact with their other piglets in the TA and
23.1 % explored the TA. In the separation test in the HP 41.0 % of the sows vocalised, 70.1 %
were active and 27.8 % explored the floor. In the last test, the reunion test in the HP, 48.2 %
of the sows vocalised, 36.5 % were active, 21.1 % had contact with their piglets, 20.8 %
nursed their piglets and 1.3 % explored the floor.
The different factor loadings of the behavioural parameters (greater than or equal to 0.4 and
less than or equal to -0.4) of the five maternal tests are shown in Table 3. Five factors are
retained by the Kaiser-Guttman criterion (Guttman, 1954; Kaiser, 1960). These five factors
explain 89.2 % of the total variance of which Factor 1 explains 33.5 %, Factor 2 18.4 %,
Factor 3 14.5 %, Factor 4 12.6 % and Factor 5 10.2 %.
The parameter ‘vocalisation’ loads positively on Factor 1 in the separation and the piglet
scream test in the TA and in the separation test in the HP (Table 1 and Table 3). Furthermore,
the sow’s presence ‘near the piglet nest’ loads on Factor 1 positively, while ‘exploring TA’
loads negatively in the separation and piglet scream test in the TA. On Factor 2 the
behavioural parameters ‘vocalisation’, ‘active’ and ‘exploring floor’ in the separation test in
the HP are positively related to the parameters ‘active’ and ‘contact with piglets’ of the sows
in the reunion test in the HP. In the reunion test in the TA ‘vocalisation’, ‘active’ and ‘contact
with piglets’ loads positively on Factor 3, while ‘exploring TA’ loads negatively. On Factor 4
‘vocalisation’ and ‘nursing’ are positively related. In the piglet scream test in the TA ‘active’
and ‘contact with piglets’ of the sows load positively together on Factor 5, while the sow’s
presence ‘near the piglet nest’ in the separation test and ‘near the screaming piglet’ in the
piglet scream test in the TA loads negatively.
40
Table 2
Median, minimum (min) and maximum (max) of the frequencies and the percentage (%) of
the behavioural parameters of the sows (n=47) tested in week 2 and 4 of lactation in the
separation test, reunion test and piglet scream test in the test arena (TA) and in the separation
and reunion test in the home pen (HP).
Test Parameter median min max %
Separation test in TA
5min observation
(31 observations per sow)
Vocalisation 24.0 0.0 31.0 68.8
Near piglet nest (location 1) 21.5 6.0 31.0 69.9
Exploring nest 09.0 0.0 26.0 26.2
Exploring TA
05.0 0.0 27.0 22.6
Reunion test in TA
5min observation
(31 observations per sow)
Vocalisation 08.0 0.0 31.0 37.1
Near piglet nest (location 1) 11.0 2.0 25.0 34.8
Active 10.0 1.0 22.0 33.6
Contact with piglets 11.0 2.0 28.0 38.2
Exploring TA
13.0 1.0 28.0 42.6
Piglet scream test in TA
30s observation
(4 observations per sow)
Vocalisation 03.0 0.0 04.0 57.5
Near screaming piglet (location 1) 03.0 0.0 04.0 61.2
Active 01.0 0.0 03.0 30.9
Contact with piglets 00.0 0.0 03.0 15.4
Exploring TA
00.0 0.0 04.0 23.1
Separations test in HP
5min observation
(31 observations per sow)
Vocalisation 11.0 0.0 31.0 41.0
Active 23.5 4.0 31.0 70.1
Exploring floor
08.0 0.0 24.0 27.8
Reunion test in HP
5min observation
(31 observations per sow)
Vocalisation 13.0 0.0 31.0 48.2
Active 10.0 0.0 31.0 36.5
Contact with piglets 07.0 0.0 21.0 21.1
Nursing 00.0 0.0 29.0 20.8
Exploring floor
00.0 0.0 07.0 01.3
41
Table 3
Factor loadings on behavioural parameters on the five maternal tests are shown. For clarity of
the table factor loadings are illustrated greater than or equal to 40 and less than or
equal to -40.
Test Parameter F1 F2 F3 F4 F5
Separation test
in test arena
Vocalisation -76
Near piglet nest (location 1) -74
Exploring TA
-67
Reunion test
in test arena
Vocalisation -48
Near piglet nest (location 1) -48
Active -64
Contact with piglets -77
Exploring TA
-68
Piglet scream test
in test arena
Vocalisation -60
Near screaming piglet (location 1) -56
Active -72
Contact with piglets -65
Exploring TA
-57
Separations test
in home pen
Vocalisation -50 -43
Active -69
Exploring floor
-67
Reunion test
in home pen
Vocalisation -87
Active -72
Contact with piglets -64
Nursing
-82
42
DISCUSSION
The five extracted factors of the behavioural parameters in the five maternal behavioural tests
present different forms of maternal motivation and make the complexity and diversity of
maternal behaviour visible.
Factor 1 explains for the largest part of variance and indicates the most important maternal
factor, the reaction of the sows when their piglets are in danger. Špinka et al. (2000) evaluated
the protective reactions of the sow to piglets in threating situations also as important maternal
factor by using a factor analysis. In the present study, ‘vocalisation’ loads in three maternal
tests together on the first factor and this indicates that communication is an important
instrument for sows with piglets in threatening situations. Hutson et al. (1991) stated that the
scream of a piglet is the most important stimulus for a sow to react to. The responsiveness of
the sows to it plays a central role in piglet survival with regard to piglet crushing. For
instance, Weary et al. (1996) demonstrates that piglet survival increased immensely when the
sow reacted to a trapped piglet within one minute. Grimberg-Henrici et al. (2016) found that
group-housed sows reacted more strongly in a piglet scream test compared to conventional
single-housed sows and showed fewer piglet losses in the course of their study. Another
maternal factor belonging to Factor 1 is the local attachment of the sows to their piglets,
which is demonstrated by the positive loading of ‘near the piglet nest’. The study conducted
by Špinka et al. (2000) found comparable results, where more than half of the experimental
sows did not leave the nest and thus their piglets for food. The negative loadings on
‘exploring TA’ for Factor 1 showed the behaviour of the sows that was not directed towards
their piglets. The sow explored the TA during threating test situations for their piglets and
expressed no behaviour that demonstrated interest in the piglets and thus indicated disinterest
towards their piglets.
Factor 2 illustrates the reaction of the sows while searching for their piglets and recognising
their piglets. Vocalisation of the sows is also used here to obtain information from their
piglets. However, another situational type of vocalisation is detected here compared to
Factor 1. If vocalisation of Factor 1 and Factor 2 would be the same type they would load
only on one collective factor. Marchant et al. (2001) identified in their study three categories
of calls: single grunts, single squeals and rapidly repeated grunts. For the single grunts two
types were found which differed in their duration. Additionally, the practical experiences in
the stables and during testing approve different forms of vocalisation. Furthermore, the tactile
contact between sow and piglets is an important maternal factor. Andersen et al. (2005)
discovered that higher sow contact with piglets is related to lower mortality in piglets.
43
The same behavioural parameters as on Factor 2, i.e. ‘vocalisation’, ‘active’, ‘contact with
piglets’ load positively on Factor 3. However, these are the behavioural parameters measured
in another test. Thus the tests do not load together on one factor. The experimental
environment in which the test is placed influences the significance of the test and thus the
maternal response of the sows. Furthermore, the negative loading of ‘exploring TA’
demonstrates the disinterest of the sows with regard to their piglets as explained for Factor 1.
‘Vocalisation’ and ‘nursing’ load together on Factor 4 and this indicates a care factor of
maternal behaviour. Here, vocalisation is used to call the piglets for nursing and thus to
synchronise nursing with other sows. This synchronisation of nursing bouts with other sows
has an evolutional background and is observed in domestic free-range pigs
(Newberry and Wood-Gush, 1985) as well as in group-housed sows (Wechsler and
Brodmann, 1996). Wechsler and Brodmann (1996) concluded that the synchronisation of the
nursing bouts of sows coordinates their daily rhythm and minimises the cross-suckling of
piglets between sows.
Factor 5 indicates two different maternal factors of sows. On the one hand, Factor 5
demonstrates the local attachment of the sows to their screaming piglet. On the other hand,
Factor 5 shows the motivation of the sows to get into contact with their remaining piglets in
the TA. The piglet scream test combines two different but equivalent maternal factors in one
test. The sows which decide to be attached to the screaming piglet in danger and the other
sows which decide to get into contact with their remaining piglets to ensure that they are
unharmed.
Summarising the different factor loadings, four underlying maternal factors can be
extracted: communication, care, contact and local attachment (Table 4). The present results
illustrate that more than one maternal test is needed to prove maternal factors. Furthermore,
the results validate that different maternal tests can include one or more maternal factors in
one test. Hence, maternal tests can provide more than one statement about maternal
characteristics and need to be interpreted very carefully. For example, sows that do not focus
on a screaming piglet in a piglet scream test but try to get in contact with their remaining
piglets show a contact factor of maternal behaviour instead of a communication or local
attachment factor. Pitts et al. (2002) tested sows in five comparable maternal tests which were
used in the present study and found large differences between the sows concerning their
performance in the maternal tests. However, on an individual level the sows were highly
consistent in their maternal performance. This indicates how important it is to test sows in a
number of maternal tests to cover the different factors of maternal behaviour. The present
44
study reveals that a single test cannot be sufficient to fully detect maternal behaviour.
Furthermore, the results prove that different maternal tests explain different proportions of the
total variance. This information could be useful in choosing the most suitable test. Hence, it is
an important contribution to future research to rely on different maternal tests. Moreover,
further research is needed to gain knowledge on whether there are additional maternal factors
which should also be taken into consideration.
Table 4
Schematic view of the maternal domains in the different behavioural tests.
Test Communication Contact Care Local attachment
Separation test in test arena Factor 1 Factor 1
Reunion test in test arena Factor 3 Factor 5
Piglet scream test in test arena Factor 1 Factor 5 Factor 5
Separation test in home pen Factor 1 Factor 2
Reunion test in home pen Factor 2 Factor 4
CONCLUSION
Five factors were extracted by the factor analysis. The interpretation of these factors revealed
four underlying maternal factors: communication, care, contact and local attachment. The
results indicate that the reaction of the sow to their piglets in threating situations has a high
significance with regard to the maternal motivation of a sow. Furthermore, it was proven that
one test can cover more than one maternal factor. Moreover, a comprehensive and careful
interpretation of the results is required. It was shown that the experimental environment has
an influence on the significance of the tests. This is an important contribution to knowledge
when testing and describing maternal behaviour in sows as the retained factors make the
complexity and diversity of maternal behaviour visible. Maternal behaviour is
multidimensional, thus we recommend the evaluation of sows in a number of tests concerning
their maternal performances.
45
REFERENCES
Andersen, I.L., Berg, S., Bøe, K.E., 2005. Crushing of piglets by the mother sow (Sus scrofa)
- purely accidental or a poor mother? Applied Animal Behaviour Science 93, 229-243.
Burri, M., Wechsler, B., Gygax, L., Weber, R., 2009. Influence of straw length, sow
behaviour and room temperature on the incidence of dangerous situations for piglets in
a loose farrowing system. Applied Animal Behaviour Science 117, 181-189.
Grandinson, K., Rydhmer, L., Strandberg, E., Thodberg, K., 2003. Genetic analysis of on-
farm tests of maternal behaviour in sows. Livestock Production Science 83, 141-151.
Grimberg-Henrici, C.G.E., Büttner, K., Meyer, C., Krieter, J., 2016. Does housing influence
maternal behaviour in sows? Applied Animal Behaviour Science 180, 26-34.
Guttman, L., 1954. Some necessary conditions for common-factor analysis. Psychometrika
19, 149-161.
Held, S., Mason, G., Mendl, M., 2006. Maternal responsiveness of outdoor sows from first to
fourth parities. Applied Animal Behaviour Science 98, 216-233.
Hellbrügge, B., Tölle, K.H., Bennewitz, J., Henze, C., Presuhn, U., Krieter, J., 2008. Genetic
aspects regarding piglet losses and the maternal behaviour of sows. Part 1. Genetic
analysis of piglet mortality and fertility traits in pigs.
Hutson, G.D., Wilkinson, J.L., Luxford, B.G., 1991. The response of lactating sows to tactile,
visual and auditory stimuli associated with a model piglet. Applied Animal Behaviour
Science 32, 129-137.
Kaiser, H.F., 1960. The application of electronic computers to factor analysis. Educational
and psychological measurement.
Løvendahl, P., Damgaard, L.H., Nielsen, B.L.m., Thodberg, K., Su, G., Rydhmer, L., 2005.
Aggressive behaviour of sows at mixing and maternal behaviour are heritable and
genetically correlated traits. Livestock Production Science 93, 73-85.
Marchant, J.N., Whittaker, X., Broom, D.M., 2001. Vocalisations of the adult female
domestic pig during a standard human approach test and their relationships with
behavioural and heart rate measures. Applied Animal Behaviour Science 72, 23-39.
Melišová, M., Illmann, G., Chaloupková, H., Bozděchová, B., 2014. Sow postural changes,
responsiveness to piglet screams, and their impact on piglet mortality in pens and
crates. Journal of animal science 92, 3064-3072.
Miekley, B., Traulsen, I., Krieter, J., 2013. Principal component analysis for the early
detection of mastitis and lameness in dairy cows. Journal of Dairy Research 80, 335-
343.
46
Newberry, R.C., Wood-Gush, D.G.M., 1985. The suckling behaviour of domestic pigs in a
semi-natural environment. Behaviour 95, 11-25.
O'Rourke, N., Psych, R., Hatcher, L., 2013. A step-by-step approach to using SAS for factor
analysis and structural equation modeling. Sas Institute.
Pitts, A.D., Weary, D.M., Fraser, D., Pajor, E.A., Kramer, D.L., 2002. Alternative housing for
sows and litters.: Part 5. Individual differences in the maternal behaviour of sows.
Applied Animal Behaviour Science 76, 291-306.
SAS Institute Inc., 2008. User's Guide (release 9.4). Cary, North Carolina, USA.
Špinka, M., Illmann, G., de Jonge, F., Andersson, M., Schuurman, T., Jensen, P., 2000.
Dimensions of maternal behaviour characteristics in domestic and wildxdomestic
crossbred sows. Applied Animal Behaviour Science 70, 99-114.
Temple, D., Manteca, X., Velarde, A., Dalmau, A., 2011. Assessment of animal welfare
through behavioural parameters in Iberian pigs in intensive and extensive conditions.
Applied Animal Behaviour Science 131, 29-39.
Weary, D.M., Pajor, E.A., Fraser, D., Honkanen, A.-M., 1996. Sow body movements that
crush piglets: a comparison between two types of farrowing accommodation. Applied
Animal Behaviour Science 49, 149-158.
Wechsler, B., Brodmann, N., 1996. The synchronization of nursing bouts in group-housed
sows. Applied Animal Behaviour Science 47, 191-199.
Wemelsfelder, F., Hunter, A.E., Paul, E.S., Lawrence, A.B., 2012. Assessing pig body
language: Agreement and consistency between pig farmers, veterinarians, and animal
activists. Journal of animal science 90, 3652-3665.
47
CHAPTER THREE
The effect of different housing systems on reproductive traits
and the behaviour of low-risk and high-risk crushing sows
Charlotte G.E. Grimberg-Henricia, Kathrin Büttnera, Reikja Y. Ladewiga,
Christian Meyerb, Joachim Krietera
a Institute of Animal Breeding and Husbandry, Christian-Albrechts-University,
Olshausenstr. 40, D-24098 Kiel, Germany
b Chamber of Agriculture of Schleswig-Holstein,
Gutshof 1, D-24327 Blekendorf, Germany
48
ABSTRACT
Free-farrowing systems and group-housing systems for lactating sows are sensitive systems
and require an optimal interaction of different environmental factors to be successful. The aim
of the present study was to compare sows’ reproductive traits during lactation in two
group-housing systems (GH; GHbig: n=40; GHsmall: n=40) and in a single-housing system
(SH; n=63). The two GH systems differed in size (GHbig: 8.3 m2 / sow; GHsmall: 7.1 m2 / sow).
GH sows were separated into free-farrowing pens from three days ante partum until six days
post partum. For the remaining time, the sows and piglets were allowed run together in the
whole GH system. Data were collected in four batches with 20 GH sows (GHbig: n=10;
GHsmall: n=10) and 16 SH sows in each housing system. Regarding the reproductive traits,
sows of both GH systems had significantly more total piglet losses (e.g. crushing,
underweight, runt) and crushed more piglets during lactation compared to SH sows (p<0.05).
In addition, GHsmall sows had higher total piglet losses compared to GHbig sows (p<0.05).
Besides the reproductive traits, the lying down and rolling behaviour of both high-risk
crushing sows (HRC; ≥35 % crushed piglets; n=10) and low-risk crushing sows
(LRC; ≤20 % crushed piglets; n=10) in the GH system were investigated in the first 72 hours
post partum to obtain more information about critical situations of piglets being crushed.
HRC and LRC sows did not differ in their frequency of lying down movements. However,
LRC sows performed significantly more lying down movements by using the pen walls,
which has been described as the safest way to prevent crushing (p<0.05). In addition, piglets
of LRC sows were more active during lying down, which was negatively correlated with the
number of crushed piglets during the first 72 hours post partum (p<0.05). Furthermore,
HRC sows rolled significantly more frequently and especially rolling movements from one
side to the other side of the sow were highly correlated with the number of crushed piglets
during the first 72 hours post partum (p<0.05). In conclusion, the safety of the piglets was
reduced in the GH systems related to higher pre-weaning mortality. In addition, the pen size
of the free-farrowing pens of the GH systems had an influence on the total piglet losses. More
piglet losses were documented for sows in smaller free-farrowing pens. Additionally, the
detailed observation of HRC and LRC sows within the same housing system showed high
variation in their maternal behaviour and in their postural changes in free-farrowing pens.
HRC sows performed more lying down movements without using the pen walls and rolled
more frequently compared to LRC sows.
49
Keywords
Group housing system; Free-farrowing pens; Piglet mortality; Low-risk and high-risk
crushing sows
INTRODUCTION
A group-housing system for lactating sows offers the sows and piglets a more natural rhythm
ante partum and post partum (Jensen, 1986), a stronger sow-piglet relation
(Arey and Sancha, 1996; Grimberg-Henrici et al., 2016) and a more gentle weaning procedure
due to fewer and less intensive fights between the piglets (Bohnenkamp et al., 2013b).
Besides these positive effects of a group-housing system during lactation, the high
pre-weaning piglet mortality remains a critical economical and ethical issue. Piglet mortality
rates range from 11-34 % (Pedersen et al., 1998; Weber, 2000; Marchant et al., 2001;
Andersen et al., 2007; Baxter et al., 2015) in free-farrowing pens and group-housing systems.
Piglets are most vulnerable to crushing in the first 24 h post partum and 50 % of the crushed
piglets are documented in the first two days after birth (Marchant et al., 2001;
Kilbride et al., 2012).
The pre-weaning mortality of piglets due to crushing is a multifactorial problem. The
probability of a piglet of being crushed is associated with the health of the piglet
(Marchant et al., 2001; Pedersen et al., 2011), the litter size relating to lower birth weights
(Milligan et al., 2002), the maternal performance (Marchant et al., 2001;
Andersen et al., 2005; Burri et al., 2009; Wischner et al., 2009) and the condition and age of
the sow (Weary et al., 1998). In addition, piglet mortality is influenced by pen design
(Weary et al., 1996a; Herskin et al., 1998; Weary et al., 1998), the expertise of the sow with
alternative farrowing systems (Marchant et al., 2000), the expertise of the stockpersons
(Li et al., 2010) and the lying down and rolling behaviour of the sow as reviewed by
Damm et al. (2005).
Especially the lying down and rolling behaviour of the sows has a great influence on the
number of crushed piglets. Some studies have reported high numbers of crushed piglets
during lying down movements (Weary et al., 1998; Marchant et al., 2001). In turn, other
studies have documented more crushed piglets during rolling movements
(Weary et al., 1996a; Weary et al., 1998; Danholt et al., 2011). Several studies have described
pre-lying behaviour, which has a positive effect on the number of crushed piglets
(Schmid, 1991; Marchant et al., 2001). However, no studies have found pre-rolling behaviour
50
which makes rolling movements extremely dangerous for piglets, as reviewed by
Damm et al. (2005). In addition, pen design has an influence on these behaviours. For
example, fewer piglets are crushed if sows use the pen walls during lying down
(Marchant et al., 2001), and related thereto, sloped walls are shown to be very useful in
reducing crushing (Marchant et al., 2001). Furthermore, sows roll less on concrete floors
compared to plastic floors (Weary et al., 1998). Softer floor types such as sand and straw are
said to reduce the number of crushed piglets, however, the frequency of rolling movements is
not reduced (Herskin et al., 1998).
In light of this, the objective of the present study was to investigate differences in
pre-weaning piglet mortality between group-housed and single-housed sows. Two different
sizes of free-farrowing pens within a group-housing system were tested with regard to
differences in piglet mortality rates. Reproductive traits and piglet losses were documented in
detail during lactation. In addition, sows with few crushed piglets (LRC) and sows with many
crushed piglets (HRC) were observed with regard to their lying down and rolling behaviour
72 hours post partum to obtain more information about critical situations of piglets being
crushed.
MATERIAL AND METHODS
Animals and housing
The study was conducted on the Futterkamp agricultural research farm of the Chamber of
Agriculture of Schleswig-Holstein over a period from April 2016 until January 2017. A total
of 143 cross-bred (Large White × Landrace) nulliparous and multiparous sows and their
piglets (Pietrain x (Large White x Landrace)) were observed during lactation in a
group-housing system (GH; GHbig: n=40; GHsmall: n=40) and in a conventional
single-housing system (SH; n = 63). Data were collected in four batches with 20 sows in the
GH system and 16 sows in the SH system per batch.
The GH systems differed in size but had an identical design (Figure 1). Ten sows were housed
together in both GH systems. Each sow had a free-farrowing pen (GHbig: 2.2 m × 2.6 m;
GHsmall: 2.2 m x 2.4 m) provided with an embedded and covered piglet nest (1.7 m2) with a
rubber floor and a heating lamp. The pens had a straw rack and anti-crush rails at the pen
walls. Furthermore, in the free-farrowing pen, half of the floor was a plastic, slatted floor and
the other half a concrete, slatted floor. Manipulable material (plastic balls) was also available
51
to the sows and the piglets. All pens had an entrance for the sows and a separated entrance for
the piglets to a shared running area (GHbig: 2.5 m × 12.0 m; GHsmall: 2.5 m x 11.0 m), which
was also equipped with anti-crush rails at the walls. The floor of the running area was a
concrete, slatted floor. Each sow from the GHbig system had in total 8.3 m2 and each sow from
the GHsmall system 7.1 m2 including the individual space in the free-farrowing pen and the
running area. The sows with their litters were separated into the free-farrowing pens from
three days ante partum until six days post partum. After six days post partum, the gates of the
pens were opened, and all sows and piglets were allowed to mix. Sows in the GH systems
were fed electronically in the pens. To prevent aggression over feed, all sows were fed ad
libitum with a commercial meal with a maximum of 10 kg per day and all sows were able to
feed in any available pen. Drinking nipples for the piglets and the sows were available in each
pen. A separate feeding area for the creep diet for the piglets was provided in the running
area.
Sows in the SH system were fixed permanently in farrowing crates (2.0 m × 2.6 m) and were
fed electronically with a commercial meal which increased constantly during lactation to a
maximum of 7.5 kg per day. SH pens had water-heated piglet resting areas (0.6 m2) with
heating lamps and, in addition, manipulable material (plastic balls and wood) was made
available. The floor in these pens was a plastic slatted floor with an iron slatted floor part in
the lying area of the sows. Sows and piglets shared a drinking trough. Creep feed was
provided in feeding bowls for the piglets.
All litters were standardised to 14 piglets for each sow until two days post partum. Boars were
not castrated and all piglets of batches 2 and 4 were not tail-docked. The piglets were weaned
on average 26 ± 1 days post partum. During gestation, all sows were housed in a dynamic
group with electronic feeding stations and were randomly moved to the GH or SH system,
respectively, one week before the calculated farrowing date. When the sows did not farrow on
the calculated date, the birth was initiated with an injection of Prostaglandin F2α. The
temperature varied in the two housing systems between 19 and 21°C. Lights were switched on
at 6 am and off at 8 pm.
52
Figure 1
A schematic view of the group-housing system and single-housing system (2.0 m × 2.6 m).
Two different variants of a group-housing system were tested with ten individual pens
numbered from one to ten (GHbig: 2.2 m × 2.6 m; GHsmall: 2.2 m x 2.4 m) and a shared
running area (GHbig: 2.5 m × 12.0 m; GHsmall: 2.5 m x 11.0 m).
Reproductive traits
The reproductive traits of the sows were documented as the number of live-born piglets,
stillborn piglets, weaned piglets, total piglet losses and individual weights of the piglets one
day after birth and at weaning. Piglet losses were recorded during the whole lactation period
regarding cause, date, weight of the piglet and location of the dead piglet. Total piglet losses
included piglets that were crushed and piglets that had died due to other causes
(e.g. underweight, runt, splay legs).
Video analysis
All sows were videotaped during their time in the farrowing stables. Video cameras
(Axis M3024LVE, 5 frames/s) were positioned above each pen to obtain a complete
overview. In the free-farrowing pens, high numbers of crushed piglets were documented in
the first days after birth. Sows with low and sows with high piglet motility rates due to
53
crushing were analysed to obtain information on dangerous situations for piglets of being
crushed and to find differences in sows’ and piglets’ behaviour. The sows were observed
continuously in the first 72 h post partum based on Philipps et al. (2014). The borders for
high-risk and low-risk crushing sows were calculated with UNIVARIATE in the statistical
software SAS® 9.4 (SAS Institute Inc., 2008) to determinate the quantiles of the 25 %, with
low and high numbers of crushed piglets respectively. On the basis of this calculation and
with regard to almost equal distribution of the number of parities of the sows, ten GH sows
were selected with piglet crushing mortality rates equal or less than 20 % – which corresponds
to 1 to 3 crushed piglets (low-risk crushing (LRC) sows) – and ten GH sows with piglet
crushing mortality rates equal or greater than 35 % – which corresponds to 7 to 10 crushed
piglets (high-risk crushing (HRC) sows). The video recordings were analysed with the
Behavioural Observation Research Interactive Software (BORIS version 4.1.4) and the
frequencies of the lying down and rolling movements of the sows, the position of the piglets
and the number of crushed piglets were recorded (Table 1).
Statistical analysis
All data were analysed with the statistical software SAS® 9.4 (SAS Institute Inc., 2008). Fixed
effects were added in a stepwise manner to the model. The Akaike’s information criteria
corrected (AICC) and the Bayesian information criteria (BIC) were used to compare the
different models. The model with the smallest AICC and BIC values was chosen. The
residuals were tested for normal distribution.
Reproductive traits
The normally distributed data of the reproductive traits (number of piglets born alive,
total piglet losses, birth and weaning weight of the piglets, weight of deceased piglets
(except stillborn piglets)) were analysed with the MIXED procedure. The mixed model
included the fixed effects housing system (GHbig, GHsmall, SH), batch (B1-B4), parity class
(class 1: 1; class 2: 2-4; class 3: ≥5), the interaction between housing system and batch and
the interaction between housing system and parity class. The interactions were removed if no
significant effects were found. The sow was added to the model of birth weights and weaning
weights as a random effect nested in housing system and batch. The birth weights and the
lactation length were added to the model for the weaning weights as linear continuous
variables. The significance of differences in the least square means was adjusted by the
Bonferroni-correction.
54
The non-normally distributed data of the reproductive traits (number of stillborn, crushed
piglets and piglets that died due to other causes) were analysed with the GLIMMIX procedure
with a poisson-distribution using the log-link function. The model included the fixed effects
housing system (GHbig, GHsmall, SH), batch (B1-B4), parity class (class 1: 1; class 2: 2-4;
class 3: ≥5), the interaction between housing system and batch and the interaction between
housing system and parity class. In addition, the fixed effect day of death
(birth to day 2: Birth-D2; day 3 to day 5: D3-D5; equal or greater than day 6: >D6) with the
interaction between housing system and day of death was added to the model of the number
of crushed piglets. The interactions were removed if no significant effects were found. The
significance of differences in the least square means was adjusted by the
Bonferroni-correction.
Video analysis
The normally distributed data of the reproductive traits of the sows from the video analysis
(number of piglets born alive, number of parities) were analysed with the MIXED procedure.
The model included the fixed effects group (LRC, HRC), parity class (class 1: 1; class 2: 2-4;
class 3: ≥5) and the interaction between group and parity class. The interaction was removed
if no significant effect was found.The significance of differences in the least square means
was adjusted by the Bonferroni-correction.
The non-normally distributed data of the reproductive traits of the sows from the video
analysis (number of crushed piglets during lactation, number of crushed piglets during
72 hours post partum) and the count data of the video analysis (sows’ and piglets’ behaviour)
were analysed with the GLIMMIX procedure with a poisson-distribution using the log-link
function. The model included the fixed effects group (LRC, HRC), parity class (class 1: 1;
class 2: 2-4; class 3: ≥5) and the interaction between group and parity class. The interaction
was removed if no significant effect was found. The significance of differences in the least
square means was adjusted by the Bonferroni-correction. Moreover, the relationships between
sows’ and piglets’ behaviour and the number of the documented crushed piglets during the
first 72 hours post partum were tested with the Spearman Correlation Coefficient.
55
Table 1
Ethogram for behavioural observations of lying down and rolling movements of the sows and
piglets’ positions in the first 72 hours post partum (modified according to
Wischner et al. (2010)).
Parameter Description
Sows
Lying down From a standing or sitting posture in a ventral, sternal
or lateral recumbency.
Lying down using pen walls From a standing or sitting posture in a ventral, sternal
or lateral recumbency by leaning on a pen wall to lie
down.
Lying down without using pen walls From a standing or sitting posture in a ventral, sternal
or lateral recumbency by lying down freely without
using a pen wall.
Rolling Postural changes in a lying posture from one lateral
position to another lateral position (side-to-side) or
from a ventral or sternal position to a lateral position
(belly-to-side).
180˚ rolling (side-side) Postural changes in a lying posture from one lateral
position to another lateral position (side-to-side).
90˚ rolling (belly-side) Postural changes in a lying posture from a ventral or
sternal position to a lateral position (belly-to-side).
90˚ rolling (side-belly) Postural changes in a lying posture from lateral
position to a ventral or sternal position (side-to-belly).
Piglets
Position ‘nest’ At least 70 % of the piglets were resting in the piglet
nest.
Position ‘near sow’ At least 70 % of the piglets were active near the sow or
resting near the sow.
Position ‘active’ At least 70 % of the piglets were walking or running
around in the pen.
Position ‘non-synchronous’ At least 70 % of the piglets did not perform the same
behaviour.
56
RESULTS
Reproductive traits
Housing system
In the present study, 17.5 ± 0.28 piglets were born alive per sow and the birth weight of the
piglets was on average 1.22 kg ± 0.01 for all housing systems. With regard to piglet losses of
live-born piglets, sows from the GH systems had significantly higher total piglet losses during
lactation compared to sows from the SH system (GHbig: 4.64 ± 0.47 vs. GHsmall: 6.19 ± 0.47
vs. SH: 3.17 ± 0.38; p<0.05). Furthermore, sows from GHsmall had significantly more total
piglet losses compared to sows from GHbig (p<0.05). In addition, sows from both GH systems
crushed significantly more piglets over all batches compared to SH sows (p<0.05)
(Figure 2a). Figure 2b demonstrates the course of the crushed piglets of the different housing
systems during lactation. 65.3 % of the piglets were crushed during birth and the first two
days post partum. 23.1 % of the crushed piglets were documented between day 3 and 5
post partum. The running area was opened for the GH sows and their piglets at day six of
lactation. After opening the pens, 11.6 % of the crushed piglets were found. Sows from both
GH systems crushed significantly more piglets during birth and at day 1 and 2 post partum
compared to the SH sows (p<0.05). However, the sows did not differ statistically between the
housing systems regarding their number of crushed piglets after day 2 post partum.
Furthermore, the deceased piglets of sows from both GH systems were significantly heavier
compared to the deceased piglets of the SH sows (GHbig: 1.36 kg ± 0.05 vs.
GHsmall: 1.29 kg ± 0.05 vs. SH: 1.03 kg ± 0.05; p<0.05). Other causes that led to piglet
mortality such as underweight, runt, splay legs etc. did not differ between the housing systems
(Figure 2c). The number of stillborn piglets fluctuated over the batches (Figure 2d) and did
not differ statistically between GH and SH sows.
Figure 3 shows that the weaning weight of the GH piglets fluctuated between the batches,
however, the weaning weight of the SH piglets remained stable. No differences were found
between both GH systems. SH sows weaned in three out of four batches heavier piglets
compared to sows from both GH systems (p<0.05).
57
Figure 2
LSMeans with standard error of the reproductive traits (number of crushed piglets (a), piglets
that died due to other causes (c), stillborn piglets (d)) for the interaction between housing
system (group-housing system (GHbig, GHsmall); single-housing system (SH)) and batch
(B1-B4), and LSMeans with standard error of the number of crushed piglets (b) for the
interaction between housing system (group-housing system (GHbig, GHsmall); single-housing
system (SH)) and day of death (Birth-D2, D3-D5, <D6).
Parity classes
Sows from the parity class 3 born significantly fewer piglets compared to sows from the
parity class 1 and 2 (class 1: 18.1 ± 0.63 vs. class 2: 17.9 ± 0.41 vs. class 3: 16.4 ± 0.40;
p<0.05). In addition, sows from the parity class 3 had significantly more stillborn piglets
compared to sows from the parity class 1 and 2 (class 1: 0.88 ± 0.19 vs. class 2: 1.38 ± 0.16
vs. class 3: 2.00 ± 0.19; p<0.05). Furthermore, sows from all parity classes differed
58
significantly from each other regarding the birth weight of their piglets
(class 1: 1.10 kg ± 0.04 vs. class 2: 1.25 kg ± 0.03 vs. class 3: 1.29 kg ± 0.02; p<0.05). Sows
from the parity class 2 weaned significantly heavier piglets compared to the other parity
classes (class 1: 6.90 kg ± 0.13 vs. class 2: 7.44 kg ± 0.09 vs. class 3: 7.00 kg ± 0.08; p<0.05).
The three parity classes did not differ regarding the total piglet losses and the number of
crushed piglets.
Figure 3
LSMeans with standard error of the weaning weight of the piglets for the interaction between
housing system (group-housing system (GHbig, GHsmall); single-housing system (SH)) and
batch (B1-B4).
59
Video analysis
Low-risk and high-risk crushing sows
Table 2 shows the reproductive and behavioural differences between LRC and HRC sows and
their piglets. LRC and HRC sows did not differ significantly regarding their number of
parities. However, HRC sows born significantly more live-born piglets compared to
LRC sows (p<0.05). Furthermore, HRC sows crushed significantly more piglets during the
first 72 hours post partum and during lactation (p<0.05). The 72 hours post partum video
analysis detected 72.8 % of the total number of crushed piglets during lactation of these sows.
Furthermore, HRC sows crushed significantly more piglets during lying down and rolling
movements compared to LRC sows (p<0.05). However, an equal number of crushed piglets
were documented for lying down and rolling movements, respectively, for LRC and
HRC sows.
The two groups did not differ in the frequency of lying down movements. Moreover, no
differences were found between the sows with regard to their usage of the pen walls when
lying down. However, HRC sows performed significantly more lying down movements
without using the pen walls (p<0.05). No correlations were found between lying down
movements and the number of crushed piglets in the first 72 hours post partum. With regard
to rolling movements, HRC sows rolled significantly more compared to LRC sows during the
first 72 hours after birth (p<0.05). In addition, HRC sows performed significantly more
90˚ rolling, i.e. from the side to the belly and from the belly to the side. Moreover,
180˚ rolling, from one side to the other side, was more frequently observed in HRC sows
(p<0.05). Especially rolling movements from one side to the other side showed a highly
positive correlation of 0.81 with the number of crushed piglets in the first 72 hours
post partum (p<0.05). In general, a moderate correlation of 0.65 was found (p<0.05) for
rolling movements and the number of crushed piglets in the first 72 hours post partum.
With regard to the piglets’ positions, piglets were located near the sow more frequently during
rolling movements compared to lying down movements. Moreover, piglets of HRC sows
were significantly less active and were found in the nest significantly more frequently during
lying down and rolling movements (p<0.05). Furthermore, piglets from HCR sows were
significantly less synchronous in their behaviour during the rolling movements of the sows
(p<0.05). Regarding lying down, a moderately negative correlation of -0.56 was found
between the activity of the piglets and the number of crushed piglets in the first 72 hours
post partum. A moderate positive correlation of 0.50 was also found between the
60
piglets’ position in the nest and the number of crushed piglets in the first 72 hours
post partum. However, no correlation for rolling could be detected between the piglets’
positions and the number of crushed piglets in the first 72 hours post partum.
Parity classes
Table 3 shows the reproductive and behavioural performances of the sows from the three
parity classes. Sows from the parity class 1 gave birth to significantly more piglets compared
to sows from parity class 3 (p<0.05). The sows of the three parity classes did not differ
regarding the number of crushed piglets in the first 72 hours post partum, the number of
crushed piglets during lactation and number of crushed piglets during lying down and rolling
movements.
The sows showed no differences in lying down behaviour and the use or non-use of the pen
walls when lying down. With regard to rolling behaviour, sows from the parity classes 1 and 2
rolled significantly more compared to sows from the parity class 3 (p<0.05). In addition, sows
from the parity class 1 and 2 performed significantly more 90˚ rolling from the belly to the
side (p<0.05) and from the side to the belly compared to sows from the parity class 3
(p<0.05), whereas, 180˚ rolling from one side to the other side did not differ significantly
between the parity classes.
With regard to the piglets’ position during lying down movements, the piglets of sows from
the parity class 1 were in the nest significantly less frequently, were more frequently near the
sows and more active compared to piglets of older sows (p<0.05). During rolling movements,
piglets of sows from the parity classes 1 and 2 were in the nest significantly less frequently
(p<0.05). Moreover, piglets of sows from the parity class 2 were less synchronous in their
behaviour during rolling movements compared to piglets of sows from the parity classes 1
and 3 (p<0.05).
61
Table 2
LSMeans with standard error of the reproductive traits and of lying down and rolling
frequencies in the first 72 hours post partum of low-risk crushing (LRC) and high-risk
crushing (HRC) sows housed in the group-housing systems (GHbig,GHsmall), and LSMeans
with standard error of the percentages of piglets’ positions (%) during lying down and rolling
behaviour.
Parameter
Low-risk crushing
(LRC) sows
(n=10)
High-risk crushing
(HRC) sows
(n=10)
Piglets born alive 14.0a ± 0.73 16.2b ± 0.68
Crushed piglets 72 hours post partum 1.21a ± 0.36 5.78b ± 0.77
Crushed piglets during lactation 1.81a ± 0.44 8.84b ± 0.95
Number of parities
4.20a ± 0.86 3.70a ± 0.86
Lying down 35.6a ± 2.04 36.7a ± 1.93
Crushed piglets 72 hours post partum 0.65a ± 0.26 2.64b ± 0.52
Lying down using pen walls 28.3a ± 1.83 24.7a ± 1.58
Lying down without using pen walls 7.29a ± 0.90 12.0b ± 1.10
Piglet position ‘nest’ 28.9a ± 1.82 36.1b ± 1.92
Piglet position ‘near sow’ 11.8a ± 1.15 14.1a ± 0.82
Piglet position ‘active’ 45.4a ± 2.34 33.4b ± 1.84
Piglet position ‘non-synchronous’
12.0a ± 1.18 13.4a ± 1.17
Rolling 24.6a ± 1.65 47.6b ± 2.21
Crushed piglets 72 hours post partum 0.52a ± 0.23 2.97b ± 0.56
180˚ rolling (side-side) 1.18a ± 0.35 8.41b ± 0.92
90˚ rolling (belly-side) 13.9a ± 1.27 18.9b ± 1.39
90˚ rolling (side-belly) 9.45a ± 1.02 20.0b ± 1.45
Piglet position ‘nest’ 13.9a ± 1.24 19.2b ± 1.39
Piglet position ‘near sow’ 58.7a ± 2.62 57.5a ± 2.41
Piglet position ‘active’ 18.1a ± 1.50 10.0b ± 1.00
Piglet position ‘non-synchronous’ 8.67a ± 1.01 12.1b ± 1.12
a-b Significant differences between the LRC and HCR sows (p<0.05).
62
Table 3
LSMeans with standard error of the reproductive traits and of lying down and rolling
frequencies in the first 72 hours post partum of the sows from three parity classes
(class 1: gilts; class 2: sows with 2-4 litters; class 3: ≥5) housed in the group-housing systems
(GHbig,GHsmall), and LSMeans with standard error of the percentage of the piglets’ positions
(%) during lying down and rolling behaviour.
Parameter Parity class 1
(n=7)
Parity class 2
(n=5)
Parity class 3
(n=8)
Piglets born alive 17.0a ± 0.81 14.7ab ± 1.00 13.7b ± 0.76
Crushed piglets 72 hours post partum 2.62a ± 0.61 2.72a ± 0.68 2.60a ± 0.58
Crushed piglets during lactation
3.68a ± 0.72 4.20a ± 0.84 4.16a ± 0.74
Lying down 36.5a ± 2.29 36.4a ± 2.82 35.6a ± 2.14
Crushed piglets 72 hours post partum 1.00a ± 0.37 1.29a ± 0.46 1.76a ± 0.48
Lying down using pen walls 25.4a ± 1.91 26.8a ± 2.46 27.0a ± 1.86
Lying down without using pen walls 10.8a ± 1.25 9.14a ± 1.35 8.26a ± 1.03
Piglet position ‘nest’ 23.5a ± 1.84 36.5b ± 2.80 39.3b ± 2.25
Piglet position ‘near sow’ 21.5a ± 1.76 6.44b ± 1.15 15.4c ± 1.41
Piglet position ‘active’ 41.9a ± 2.45 45.0b ± 3.26 31.4b ± 1.99
Piglet position ‘non-synchronous’
12.2a ± 1.33 12.3a ± 1.62 13.7a ± 1.33
Rolling 45.5a ± 2.57 38.0a ± 2.74 23.0b ± 1.72
Crushed piglets 72 hours post partum 1.58a ± 0.49 1.40a ± 0.50 0.86a ± 0.34
180˚ rolling (side-side) 3.58a ± 0.73 2.99a ± 0.68 2.94a ± 0.62
90˚ rolling (belly-side) 20.4a ± 1.71 18.8a ± 2.00 11.2b ± 1.20
90˚ rolling (side-belly) 20.5a ± 1.73 15.7a ± 1.75 8.08b ± 1.02
Piglet position ‘nest’ 14.2a ± 1.43 14.0a ± 1.68 22.0b ± 1.68
Piglet position ‘near sow’ 64.5a ± 3.05 54.4a ± 3.46 55.9a ± 2.68
Piglet position ‘active’ 12.8a ± 1.35 16.4a ± 1.99 11.7 a ± 1.21
Piglet position ‘non-synchronous’ 7.55a ± 1.04 14.6b ± 1.79 9.79a ± 1.12
a-c Significant differences between the sows from the three parity classes (p<0.05).
63
DISCUSSION
Reproductive traits
In the present study, 65 % of the crushed piglets were documented at birth and the first two
days of lactation. This is in line with results of Kilbride et al. (2012) and
Marchant et al. (2001), who observed 50 % of the crushed piglets in the first two days after
birth. In addition, the course of the number of the crushed piglets during lactation is in
accordance to the results of Marchant et al. (2000). Furthermore, GH sows had significantly
higher total piglet losses compared to SH sows. These results are comparable with other
studies which investigated sows in free-farrowing systems compared to sows kept in crates
(Marchant et al., 2001; Hales et al., 2014; van Nieuwamerongen et al., 2015). In
free-farrowing systems sows are not fixed in crates ante partum, during farrowing and post
partum. This situation was applicable to the GH sows in this study when the sows were
enclosed in their pens until six days post partum. The high mortality rates of live-born piglets
for GH sows in the present study correspond with results of other studies on free-farrowing
systems which have reported mortality rates between 22-34 % (Pedersen et al., 1998;
Marchant et al., 2001). However, other studies have reported equal or lower piglet mortality
rates around 11-18 % of sows in free-farrowing systems (Cronin et al., 2000; Weber, 2000;
Baxter et al., 2015). In these studies, the free-farrowing pens were divided into a lying area
for the sow near the piglet nest, a dunging and a feeding area (Cronin et al., 2000; Weber,
2000; Baxter et al., 2015). These functional areas in the free-farrowing pens were possibly
missing in the pens of the present study. In addition, in studies which found low piglet
mortality rates in the free-farrowing pens, the number of live-born piglets was on average
11-12 piglets, whereas, sows in the present study gave birth to 17 piglets on average. Several
studies has found that the number of live-born piglets and piglet mortality is positively
correlated (Roehe and Kalm, 2000; Weber et al., 2009; Pedersen et al., 2011). Moreover, the
individual birth weights decrease if the numbers of live-born piglets increase (Roehe and
Kalm, 2000; Wolf et al., 2008). However, an optimal birth weight is important for piglet
survival and vitality (Baxter et al., 2008). More vital piglets and, thus, more reactive piglets
would be beneficial to possibly reduce piglet mortality especially in free-farrowing systems.
In the study of van Nieuwamgerongen et al. (2015), the total piglet losses during lactation of
GH sows were 3.22 piglets per sows and for SH sows 1.52 piglets per sows. In this study, the
litters were also standardised to 14 piglets per sows. However, the number of live-born piglets
was 15 piglets per sow on average compared to 17 piglets per sows in the present study.
64
Furthermore, in the present study, piglet mortality was twice that high compared to the results
of van Nieuwamgerongen et al. (2015). Thus, lower numbers of live-born piglets seemed to
have an effect on the piglet mortality despite litter equalisation.
Besides, sows from both GH systems and SH sows did not differ statistically with regard to
mortality rates of piglets that died for other causes such as underweight and runt. However,
Weber et al. (2007), Verhovsek et al. (2007) and van Nieuwamgerongen et al. (2015) all
reported significantly higher numbers of piglets of crated sows having died due to other
causes compared to sows from free-farrowing systems. One possible explanation for this
observation was that the piglets scored as underweight and as runts in a SH system, died in a
GH system due to crushing and thus crushing was documented as the cause of death instead
of underweight and runt.
Furthermore, in the present study, the weight of the deceased piglets was significantly higher
for piglets of GH sows compared to SH sows. In addition to the results that sows from both
GH systems crushed a higher number of piglets compared to SH sows, Weary et al. (1998)
also observed a greater number of smaller, crushed piglets at birth, however, a greater number
of larger, crushed piglets were documented on days 2 and 3 post partum. These piglets were
possibly more active at the udder and thus more frequently present during postural changes of
the sow which for the piglets increases the risk of being crushed.
Moreover, the pen design can have an influence on the pre-weaning piglet mortality and on
risky postural movements of the sows. Environmental stimuli (sand and straw) which
increased nest-building behaviour and the comfort of the sow showed a positive effect on
piglet mortality and maternal performance (Herskin et al., 1998). Furthermore,
Damm et al. (2006) found that sows preferred to use walls for lying down with anti-crush rails
less and to use plain or sloped walls more often. In the present study, GHsmall sows lived in
free-farrowing pens of a size of 5.2 m2 and had significantly more total piglet losses compared
to GHbig sows, which lived in pens of a size of 6.2 m2. Weber et al. (2009) did not report
significant effects of the pen size on piglet losses. However, they detected a slight tendency
that sows in smaller pens had more total piglet losses and crushed more piglets. They suggest
a minimum pen size of 5 m2. However, Baxter et al. (2015) found that the piglet mortality
increased significantly if sows were housed in free-farrowing pens larger than 9.7 m2
compared to sows in smaller free-farrowing pens of 7.9 m2. An explanation for the higher
piglet mortality in larger pens was that the sows had more floor space where they could lie
down and roll without coming into contact with supportive structures such as anti-crush rails.
65
With regard to weaning weights, GH piglets were lighter in three out of four batches
compared to SH piglets at weaning. Bohnenkamp et al. (2013c) also described that GH piglets
were lighter compared to SH piglets. Here, the lower weaning weight of GH piglets was
explained by the greater amount of missed nursing and more playful behaviour. However,
Van Nieuwamerongen et al. (2015) did not find differences concerning weaning weights
between piglets from crated sows compared to piglets from sows housed in a GH system.
Video analysis
Low-risk and high-risk crushing sows
In the first 72 hours post partum, 73 % of the total number of crushed piglets during lactation
were detected by observing the behaviour of the sow continuously. It can be evaluated that
this period of time gives a valid overview of the behavioural differences of these sows
regarding high-risks and low-risks for crushing.
In the present study, HRC sows had significantly more live-born piglets compared to
LRC sows. This is in line with results of Phillipps et al. (2014). They also observed HRC and
LRC sows in free-farrowing pens and found larger litters with lighter piglets for HRC sows.
Andersen et al (2005) also found larger litters for HRC sows. Several studies have indicated
that larger litters are correlated with higher piglet mortality (Roehe and Kalm, 2000;
Weber et al., 2009; Pedersen et al., 2011; Phillips et al., 2014). Moreover, the number of
crushed piglets did not differ largely between lying down and rolling movements for LRC and
HRC sows respectively. Thus, in the present study, for the piglets there was an equal risk of
being crushed during lying down and rolling movements. Several studies have differed in
their documentation of the incidence of crushed piglets during postural changes with some
having reported a high incidence of crushed piglets caused by rolling movements of the sows
(Weary et al., 1996a; Weary et al., 1998; Danholt et al., 2011). Others in contrast have
described a higher number of crushed piglets during lying down movements
(Weary et al., 1998; Marchant et al., 2001).
With regard to lying down movements, it can be assumed that LRC sows performed these
movements more carefully compared to HRC sows. No differences were found in the
frequency of lying down movements between these two groups and also no differences in the
piglets’ positions near or away from the sow. Burri et al. (2009) also observed lower piglet
mortality rates for sows which were more carefully during lying down movements. The
pre-lying behaviour of the sows was described by Schmid (1991) as rooting and pawing to
wake up the piglets, moving around, gathering of all piglets on one side and lying down on
66
the opposite side of the piglets. It has also been said that pre-lying behaviour has a positive
effect on the number of crushed piglets (Schmid, 1991; Marchant et al., 2001). In the present
study, the piglets of LRC sows were more active during lying down compared to piglets of
HRC sows due to possibly better performed, pre-lying behaviour. In addition, the activity of
the piglets was negatively correlated with the risk to being crushed during lying down
movements. Furthermore, Burri et al. (2009) stated there is a significant higher risk of piglets
of being crushed if the piglets were not gathered on one side of the sow’s body. Moreover,
a positive finding in the present study was that the sows used the pen walls more frequently
while lying down, which is the safest way to prevent crushing according to
Marchant et al. (2001). Whereas, HRC sows laid down more frequently without using the pen
walls.
An important difference in the present study between HRC and LRC sows was the frequency
of rolling movements. HRC sows rolled significantly more and performed more 90° rolling
from the belly to the side or from the side to the belly and rolled more often from one side to
the other side (180° rolling). All rolling movements were positively correlated with the risk of
piglets being crushed, whereas, rolling from one side to the other side of the sows was the
most dangerous for the piglets. One main problem as reviewed by Damm et al. (2005) is that
sows have no behavioural patterns to introduce rolling movements as the pre-lying behaviour
to wake up the piglets to reduce the risk of being crushed. In the present study, no differences
in the percentage of the piglets’ position to be near the sow during rolling movements were
found between LRC and HRC sows, which probably shows that LRC sows rolled more
carefully. Weary et al. (1996a) showed that rolling movements performed more slowly
resulted in lower numbers of crushed piglets. In addition, piglets of LRC sows were more
active during rolling movements, which possibly indicates more interactions between the
LRC sows and their piglets during rolling compared to HRC sows and their piglets. However,
the restriction of rolling movements could also be contrary to the natural behaviour of the
sows since rolling from the side onto the belly is part of the gradual weaning process to
terminate nursing (Jensen, 1988). Studies have found that environmental influences can
decrease the risk of being crushed during rolling movements. Weary et al. (1998) reported
that sows rolled less frequently on concrete floors compared to sows on plastic floors.
Herskin et al. (1998) found that straw and sand bedding can reduce the risk of crushing
compared to a concrete floor, although no differences in the frequency of rolling movements
were described. Thus, softer types of floors seem to be less dangerous.
67
Another approach to reduce crushing could be a higher acceptance of the piglets to rest in the
piglet nest. Marchant et al. (2001) concluded that high mortality rates due to crushing are
caused because piglets rest near the sow most of their time and have a reduced agility after
birth. However, the motivation of the piglets to use the piglet nest and thus to be away from
the sows is found to start around two days after birth (Hrupka et al., 1998; Berg et al., 2006).
But the first days are the most critical time for crushing as shown in the present study.
Hrupka et al. (2000) showed that physical contact in a cold environment is more attractive to
the piglets than loneliness in a warm environment. Furthermore, Vasdal et al. (2010) tested
different types of piglet creep areas to increase the time of the piglets in the nest. However,
soft bedding material and improved thermal comfort did not attract the piglets to the nest
more. In a semi-natural environment, piglets will spend 90% of their time in the first two days
post partum together with the sow in the nest to develop the sow-piglet relationship
(Stanged and Jensen, 1991) and to obtain warmth, milk and protection against predators
(Weary et al., 1996b). However, this natural need of the piglets to be near the sow is a high-
risk factor in being crushed in pig husbandry. However, in the present study, piglets of
HRC sows were observed to stay more frequently in the nest compared to piglets of
LRC sows. This is contrary to our hypothesis that the acceptance of the nest of the piglets of
LRC sows would be higher. However, Berg et al. (2006) did not find a relation between the
piglets’ use of the nest and piglet mortality. Thus, more carefulness of the sows in lying down
and rolling movements is more important to prevent crushing than the acceptance of the nest
of the piglets.
Parity classes
The sows from the three parity classes did not differ with regard to the number of crushed
piglets during the first 72 hours post partum and during lactation. However,
Marchant et al. (2000) found that the piglet mortality of live-born piglets was significantly
associated with the number of parities and the body length of the sows. The comparison of the
three parity classes of the sows resulted in some evidence that older sows are possibly more
experienced regarding their maternal abilities compared to younger sows during lying down
movements. Although sows from the different parity classes did not differ in their frequency
of lying down movements, piglets of older sows were in the nest significantly more frequently
and away from the sow during lying down movements compared to piglets of gilts, which
probably indicates better pre-lying behaviour of the older sows (Schmid, 1991;
Burri et al., 2009). Furthermore, gilts performed rolling movements from the belly to the side
68
or from the side to the belly twice as often compared to older sows. These rolling movements
are used by the sows to terminate the suckling behaviour of the piglets (Jensen, 1988).
Bohnenkamp et al. (2013a) observed that gilts weaned their piglets earlier compared to older
sows. However, in the first week of lactation, no differences were found in suckling frequency
between gilts and older sows.
CONCLUSION
The present study analysed different environmental factors which provide more information
on pre-weaning mortality. Freedom of the sows reduced the safety of the piglets as shown by
the higher piglet mortality in the free-farrowing pens. Moreover, the pen size had an influence
on the piglet losses, which increased in smaller free-farrowing pens. Moreover, the detailed
observation of LRC and HRC sows within the same housing system showed high variation in
their maternal behaviour and in their postural changes in free-farrowing pens. HRC sows
performed more lying down movements without using the pen walls and rolled more
frequently compared to LRC sows.
ACKNOWLEDGMENTS
The project was supported by funds of the German Government’s Special Purpose Fund held
at Lantwirtschaftliche Rentenbank.
69
REFERENCES
Andersen, I.L., Berg, S., Bøe, K.E., 2005. Crushing of piglets by the mother sow (Sus scrofa)
- purely accidental or a poor mother? Applied Animal Behaviour Science 93, 229-243.
Andersen, I.L., Tajet, G.M., Haukvik, I.A., Kongsrud, S., Bøe, K.E., 2007. Relationship
between postnatal piglet mortality, environmental factors and management around
farrowing in herds with loose-housed, lactating sows. Acta Agriculturae Scand Section
A 57, 38-45.
Arey, D.S., Sancha, E.S., 1996. Behaviour and productivity of sows and piglets in a family
system and in farrowing crates. Applied Animal Behaviour Science 50, 135-145.
Baxter, E.M., Adeleye, O.O., Jack, M.C., Farish, M., Ison, S.H., Edwards, S.A., 2015.
Achieving optimum performance in a loose-housed farrowing system for sows: the
effects of space and temperature. Applied Animal Behaviour Science 169, 9-16.
Baxter, E.M., Jarvis, S., D’eath, R.B., Ross, D.W., Robson, S.K., Farish, M., Nevison, I.M.,
Lawrence, A.B., Edwards, S.A., 2008. Investigating the behavioural and physiological
indicators of neonatal survival in pigs. Theriogenology 69, 773-783.
Berg, S., Andersen, I.L., Tajet, G.M., Haukvik, I.A., Kongsrud, S., Bøe, K.E., 2006. Piglet
use of the creep area and piglet mortality-effects of closing the piglets inside the creep
area during sow feeding time in pens for individually loose-housed sows. Animal
science 82, 277-281.
Bohnenkamp, A.-L., Meyer, C., Müller, K., Krieter, J., 2013a. Group housing with
electronically controlled crates for lactating sows. Effect on farrowing, suckling and
activity behavior of sows and piglets. Applied Animal Behaviour Science 145, 37-43.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013b. Comparison of
growth performance and agonistic interaction in weaned piglets of different weight
classes from farrowing systems with group or single housing. animal 7, 309-315.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013c. Group housing for
lactating sows with electronically controlled crates: 1. Reproductive traits, body
condition, and feed intake. Journal of animal science 91, 3413-3419.
Burri, M., Wechsler, B., Gygax, L., Weber, R., 2009. Influence of straw length, sow
behaviour and room temperature on the incidence of dangerous situations for piglets in
a loose farrowing system. Applied Animal Behaviour Science 117, 181-189.
Cronin, G.M., Lefébure, B., McClintock, S., 2000. A comparison of piglet production and
survival in the Werribee Farrowing Pen and conventional farrowing crates at a
commercial farm. Australian Journal of Experimental Agriculture 40, 17-23.
70
Damm, B.I., Forkman, B., Pedersen, L.J., 2005. Lying down and rolling behaviour in sows in
relation to piglet crushing. Applied Animal Behaviour Science 90, 3-20.
Damm, B.I., Moustsen, V., Jørgensen, E., Pedersen, L.J., Heiskanen, T., Forkman, B., 2006.
Sow preferences for walls to lean against when lying down. Applied Animal
Behaviour Science 99, 53-63.
Danholt, L., Moustsen, V.A., Nielsen, M.B.F., Kristensen, A.R., 2011. Rolling behaviour of
sows in relation to piglet crushing on sloped versus level floor pens. Livestock Science
141, 59-68.
Grimberg-Henrici, C.G.E., Büttner, K., Meyer, C., Krieter, J., 2016. Does housing influence
maternal behaviour in sows? Applied Animal Behaviour Science 180, 26-34.
Hales, J., Moustsen, V.A., Nielsen, M.B.F., Hansen, C.F., 2014. Higher preweaning mortality
in free farrowing pens compared with farrowing crates in three commercial pig farms.
animal 8, 113-120.
Herskin, M.S., Jensen, K.H., Thodberg, K., 1998. Influence of environmental stimuli on
maternal behaviour related to bonding, reactivity and crushing of piglets in domestic
sows. Applied Animal Behaviour Science 58, 241-254.
Hrupka, B.J., Leibbrandt, V.D., Crenshaw, T.D., Benevenga, N.J., 1998. The effect of
farrowing crate heat lamp location on sow and pig patterns of lying and pig survival.
Journal of animal science 76, 2995-3002.
Hrupka, B.J., Leibbrandt, V.D., Crenshaw, T.D., Benevenga, N.J., 2000. The effect of thermal
environment and age on neonatal pig behavior. Journal of animal science 78, 583-591.
Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs.
Applied Animal Behaviour Science 16, 131-142.
Jensen, P., 1988. Maternal behaviour and mother-Young interactions during lactation in free-
ranging domestic pigs. Applied Animal Behaviour Science 20, 297-308.
Kilbride, A.L., Mendl, M., Statham, P., Held, S., Harris, M., Cooper, S., Green, L.E., 2012. A
cohort study of preweaning piglet mortality and farrowing accommodation on 112
commercial pig farms in England. Preventive Veterinary Medicine 104, 281-291.
Li, Y.Z., Johnston, L., Hilbrands, A., 2010. Pre-weaning mortality of piglets in a bedded
group-farrowing system. Journal of Swine Health and Production 18.
Marchant, J.N., Broom, D.M., Corning, S., 2001. The influence of sow behaviour on piglet
mortality due to crushing in an open farrowing system. Animal science 72, 19-28.
71
Marchant, J.N., Rudd, A.R., Mendl, M.T., Broom, D.M., Meredith, M.J., Corning, S.,
Simmins, P.H., 2000. alternative and conventional farrowing systems. The Veterinary
Record 147, 214.
Milligan, B.N., Fraser, D., Kramer, D.L., 2002. Within-litter birth weight variation in the
domestic pig and its relation to pre-weaning survival, weight gain, and variation in
weaning weights. Livestock Production Science 76, 181-191.
Pedersen, B.K., Jensen, K.H., Nielsen, N.P., 1998. Farrowing systems for the future. Agris.
Pedersen, L.J., Berg, P., Jørgensen, G., Andersen, I.L., 2011. Neonatal piglet traits of
importance for survival in crates and indoor pens. J. Anim. Sci. 89, 1207-1218.
Phillips, C.E., Farmer, C., Anderson, J.E., Johnston, L.J., Shurson, G.C., Deen, J., Keisler,
D.H., Conner, A.M., Li, Y.Z., 2014. Preweaning mortality in group-housed lactating
sows: Hormonal differences between high risk and low risk sows. Journal of animal
science 92, 2603-2611.
Roehe, R., Kalm, E., 2000. Estimation of genetic and environmental risk factors associated
with pre-weaning mortality in piglets using generalized linear mixed models. Animal
science 70, 227-240.
SAS Institute Inc., 2008. User's Guide (release 9.4). Cary, North Carolina, USA.
Schmid, H., 1991. Natürliche Verhaltenssicherungen der Hausschweine (Sus scrofa) gegen
das Erdrücken der Ferkel durch die Muttersau und die Auswirkungen
haltungsbedingter Störungen (Doctoral dissertation).
Stanged, G., Jensen, P., 1991. Behaviour of semi-naturally kept sows and piglets (except
suckling) during 10 days postpartum. Applied Animal Behaviour Science 31, 211-227.
van Nieuwamerongen, S.E., Soede, N.M., van der Peet-Schwering, C.M.C., Kemp, B.,
Bolhuis, J.E., 2015. Development of piglets raised in a new multi-litter housing
system vs. conventional single-litter housing until 9 weeks of age. Journal of animal
science 93, 5442-5454.
Vasdal, G., Glærum, M., Melišová, M., Bøe, K.E., Broom, D.M., Andersen, I.L., 2010.
Increasing the piglet's use of the creep area-A battle against biology? Applied Animal
Behaviour Science 125, 96-102.
Verhovsek, D., Troxler, J., Baumgartner, J., 2007. Peripartal behaviour and teat lesions of
sows in farrowing crates and in a loose-housing system. ANIMAL WELFARE-
POTTERS BAR THEN WHEATHAMPSTEAD- 16, 273.
72
Weary, D.M., Pajor, E.A., Fraser, D., Honkanen, A.-M., 1996a. Sow body movements that
crush piglets: a comparison between two types of farrowing accommodation. Applied
Animal Behaviour Science 49, 149-158.
Weary, D.M., Pajor, E.A., Thompson, B.K., Fraser, D., 1996b. Risky behaviour by piglets: a
trade off between feeding and risk of mortality by maternal crushing? Animal
Behaviour 51, 619-624.
Weary, D.M., Phillips, P.A., Pajor, E.A., Fraser, D., Thompson, B.K., 1998. Crushing of
piglets by sows: effects of litter features, pen features and sow behaviour. Applied
Animal Behaviour Science 61, 103-111.
Weber, R., 2000. Alternative housing systems for farrowing and lactating sows.
PUBLICATION-EUROPEAN ASSOCIATION FOR ANIMAL PRODUCTION 102,
109-116.
Weber, R., Keil, N.M., Fehr, M., Horat, R., 2007. Piglet mortality on farms using farrowing
systems with or without crates. Animal Welfare 16, 277-279.
Weber, R., Keil, N.M., Fehr, M., Horat, R., 2009. Factors affecting piglet mortality in loose
farrowing systems on commercial farms. Livestock Science 124, 216-222.
Wischner, D., Kemper, N., Krieter, J., 2009. Nest-building behaviour in sows and
consequences for pig husbandry. Livestock Science 124, 1-8.
Wischner, D., Kemper, N., Stamer, E., Hellbrügge, B., Presuhn, U., Krieter, J., 2010. Pre-
lying behaviour patterns in confined sows and their effects on crushing of piglets.
Applied Animal Behaviour Science 122, 21-27.
Wolf, J., Žáková, E., Groeneveld, E., 2008. Within-litter variation of birth weight in
hyperprolific Czech Large White sows and its relation to litter size traits, stillborn
piglets and losses until weaning. Livestock Science 115, 195-205.
73
CHAPTER FOUR
Cortisol levels and health indicators of sows and their piglets
living in a group-housing and a single-housing system
Charlotte G.E. Grimberg-Henricia, Kathrin Büttnera, Reikja Y. Ladewiga,
Christian Meyerb, Joachim Krietera
a Institute of Animal Breeding and Husbandry, Christian-Albrechts-University,
Olshausenstr. 40, D-24098 Kiel, Germany
b Chamber of Agriculture of Schleswig-Holstein,
Gutshof 1, D-24327 Blekendorf, Germany
Submitted for publication in Journal of Livestock Science
74
ABSTRACT
Since the permanent fixation of pregnant sows is banned in Europe, it has become a matter of
discussion as to whether the permanent fixation of lactating sows is still acceptable. A
group-housing system for lactating sows offers the sows and their piglets the possibility to
live in a more natural structure. The aim of this study was to compare sows from a
group-housing system (GH; n=80) with sows from a conventional single-housing system
(SH; n=64) with regard to health indicators and saliva cortisol levels. Data were collected in
four batches with 20 GH sows and 16 SH sows per batch. All sows were moved one week
before farrowing to the GH or the SH system, respectively. The GH sows were housed in a
group of ten sows and were separated into their individual free-farrowing pens from three
days ante partum until six days post partum. A running area was shared by all sows and
piglets. With regard to health indicators, the GH sows had significantly fewer skin lesions of
the udder and of the tail compared to the SH sows (p<0.05). Moreover, the GH sows had
more skin lesions on the body (p<0.05) due to ranking fights and were dirtier (p<0.05)
because they used the running area as a resting and dunging area. Furthermore, the GH sows
had higher body condition after weaning (p<0.05), which can be explained by ad libitum
feeding. With regard to health indicators of the piglets, the interaction between housing
system and batch showed that the GH piglets had fewer skin lesions of the face (p<0.05). In
addition, they had significantly fewer skin lesions of the face one week post partum and did
not differ from the SH piglets regarding their incidence of skin lesions of the face four weeks
post partum (p<0.05). Furthermore, the GH piglets showed more skin lesions of the body
compared to the SH piglets (p<0.05) across almost all batches. With regard to the saliva
cortisol levels, both the GH and SH sows did not differ in their basal levels two weeks ante
partum. However, the GH sows had significantly higher stress levels during lactation
compared to the SH sows (p<0.05). To conclude, no alarming health problems were
documented for the GH and SH sows or their piglets. Multi-suckling did not result in an
impaired sow condition, but reduced the incidence of skin lesions of the udder of the sows and
fewer skin lesions of the face were documented for GH piglets. However, higher numbers of
skin lesions of the body of the GH sows and GH piglets caused by social interactions were
described. Furthermore, higher cortisol levels were detected during lactation in the GH sows
compared to the SH sows. A possible explanation could be social stress and increased
physical activity. Thus, the GH system was more stressful for the sows compared to the SH
system.
75
Keywords
Group-housing system; Free-farrowing pens; Health indicators; Saliva cortisol levels;
Lactating sows; Piglets
INTRODUCTION
Since the permanent fixation of pregnant sows is banned in Europe, it raises the question of
whether the permanent fixation of lactating sows in crates is still acceptable. Free-farrowing
pens and group-housing systems for lactating sows have received more attention in recent
years (Bohnenkamp et al., 2013c; Phillips et al., 2014; van Nieuwamerongen et al., 2015).
The major concern with regard to alternative farrowing systems to free-farrowing pens and
group-housing systems is the piglet mortality rate due to crushing. To allow the sow to move
freely reduces the safety of the piglets (Phillips et al., 2014; van Nieuwamerongen
et al., 2015).
Besides the higher mortality rates, there are several positive effects of group-housing during
lactation. A group-housing system allows sows to live a more natural rhythm ante partum and
post partum in terms of living in small groups, separation ante partum into individual pens,
returning several days after birth with the litter to the group and living together with other
sows and their piglets during lactation (Jensen, 1986). A stronger maternal bonding of group-
housed sows with their piglets has been documented by Arey and Sancha (1996) and
Grimberg-Henrici et al. (2016). Here, sows showed stronger maternal reactions to their piglets
in threatening test situations and developed a more constant communication with their piglets.
Furthermore, the group-housing of piglets from different litters before weaning improved
their social skills. Bohnenkamp et al. (2013b) reported reduced agonistic behaviour of weaned
piglets from a group-housing system compared to weaned piglets from single-housing
systems. Group-housed piglets had significantly fewer fights, shorter fight durations and
fewer skin lesions. Furthermore, Arey and Sancha (1996) and van Nieuwamerongen et al.
(2015) found that piglets in group-housing systems showed significantly more play behaviour
during lactation compared to single-housed sows. Playful behaviour has been described as a
positive welfare indicator (Fagen, 1981; Held and Špinka, 2011).
Apart from the behavioural observations, the hormonal status especially the cortisol levels of
sows can provide information on acute and chronic stressors. Several studies have found that
housing conditions can influence the cortisol level of pigs. For example, Jarvis et al. (2006)
found that the fixation of sows in crates during lactation results in chronic stress indicated by
76
higher cortisol levels compared to loose-housed sows on straw. However, contrary results
were found by Geverink et al. (2003), who documented lower cortisol levels of individual
barren-housed gilts compared to enriched group-housed gilts.
In the present study, we concentrated on the sows’ and piglets’ health status during lactation.
Therefore, we compared sows and their piglets from a group-housing system with sows and
piglets from a single-housing system. Sows were scored concerning body condition,
locomotion and skin lesions one week ante partum and four weeks post partum before their
piglets were weaned. In addition, all piglets were evaluated in their first and fourth week of
lactation for skin lesions and injuries. Furthermore, saliva cortisol samples were collected
from the sows ante partum and during lactation to investigate differences in stress levels.
MATERIAL AND METHODS
Animals and housing
The study was conducted on the Futterkamp agricultural research farm of the Chamber of
Agriculture of Schleswig-Holstein over a period from April 2016 until January 2017. A total
of 144 cross-bred (Large White × Landrace) nulliparous and multiparous sows and 1,814
piglets (Pietrain x (Large White x Landrace)) were observed during lactation in a
group-housing system (GH; n=80) and in a conventional single-housing system (SH; n=64).
Data were collected in four batches with 20 sows in the GH system and 16 sows in the SH
system per batch.
The GH systems differed in size but had an identical design (Figure 1). In both GH systems,
ten sows were housed together. Each sow had a free-farrowing pen (GHbig: 2.2 m × 2.6 m;
GHsmall: 2.2 m x 2.4 m) provided with an embedded and covered piglet nest (1.7 m2) with a
rubber floor and a heating lamp. The pens had a straw rack and safety bars at the pen walls.
Furthermore, in the free-farrowing pen, half of the floor was a plastic, slatted floor and the
other half a concrete, slatted floor. Manipulable material (plastic balls) was also available to
the sows and the piglets. All pens had an entrance for the sows and a separated entrance for
the piglets to a shared running area (GHbig: 2.5 m × 12.0 m; GHsmall: 2.5 m x 11.0 m), which
was also equipped with safety bars at the walls. The floor of the running area was a concrete,
slatted floor. The sows with their litters were separated into the free-farrowing pens from
three days ante partum until six days post partum. After six days post partum, the gates of the
pens were opened and all sows and piglets were allowed to mix. Sows in the GH system were
77
fed electronically in the pens. To prevent aggression regarding the feed, all sows were fed ad
libitum with a commercial meal with a maximum of 10 kg per day and all sows were able to
feed in every pen. Drinking nipples for the piglets and the sows were available in each pen. A
separate feeding area for the creep diet for the piglets was provided in the running area.
Sows in the SH system were fixed permanently in farrowing crates (2.0 m × 2.6 m) and were
fed electronically with a commercial meal which increased constantly during lactation to a
maximum of 7.5 kg per day. SH pens had water-heated piglet resting areas (0.6 m2) with
heating lamps, and manipulable material (plastic balls and wood) was available for the sows
and the piglets. The floor in these pens was a plastic, slatted floor with an iron, slatted floor
part in the lying area of the sows. Both sows and piglets shared a drinking trough. Creep feed
was provided in feeding bowls for the piglets.
Figure 1
A schematic view of the group-housing system and the single-housing system
(2.0 m × 2.6 m). Two different variants of a group-housing system were tested with ten
individual pens numbered from one to ten (GHbig: 2.2 m × 2.6 m; GHsmall: 2.2 m x 2.4 m) and
a shared running area (GHbig: 2.5 m × 12.0 m; GHsmall: 2.5 m x 11.0 m).
78
All litters were standardised to 14 piglets for each sow until two days post partum. Boars were
not castrated and all piglets of batches 2 and 4 were not tail-docked. The piglets were weaned
on average 26 ± 1 days post partum. During gestation, all sows were housed in a dynamic
group with electronic feeding stations and were randomly moved to the GH or SH system,
respectively, one week before the calculated farrowing date. When the sows did not farrow on
the calculated date, the birth was initiated with an injection of Prostaglandin F2α. The
temperature varied in the two housing systems between 19 and 21°C. Lights were switched on
at 6 am and off at 8 pm.
Individual health indicators of sows
All sows were scored before they were moved to the different housing systems (one week
ante partum) with regard to different health indicators, which are described below. All sows
were scored again for the same health indicators four weeks post partum. The indicators skin
lesions of the tail and dirtiness were only evaluated four weeks post partum.
Body condition
All sows were weighed, their back fat was measured and their Body Condition Score was
recorded. The back fat thickness was measured with an ultrasound scanner
(Agroscan, Hauptner & Heberholz, Solingen, Germany) 5 cm from the median line behind the
shoulder, in the middle of the back and before the hip. The mean of these three measurement
points formed the back fat thickness value for each sow. The Body Condition Score ranged
from 1 to 5 and was scaled in steps of 0.25 with score 1 for very thin sows and score 5 for fat
sows (Bohnenkamp et al., 2013c).
Locomotion
Sows were investigated for lameness using a four-point scale (Karlen et al., 2007). The sows
had to walk along a 20 m long concrete floor to assess their locomotion score. A score of 0
indicated that the sow was able to stand and walk with symmetrical movement. A scores of 1
was assigned if the sow was able to stand and walk but with impaired movement. Moderate
lameness and thus a score of 2 was given if the sows refused to put weight on any one leg
during standing and walking and the movement was impaired and with frequent weight shifts.
Sows with a locomotion score of 3 were severely restricted in standing and walking, and were
unable to put weight on one or more legs.
79
Skin lesions of the body
To assess skin lesions of the body of all sows, the scoring method was used according to
Welfare Quality® (2009). In the present study, skin lesions of four different body parts of both
body sides of the sows were scored (part 1- ears, part 2- shoulder, front;
part 3- back, abdomen; part 4- hindquarters). The following method was used to count the
different skin lesions of the body parts. A scratch longer than 2 cm, two parallel scratches
with a maximum distance of 0.5 cm or a wound smaller than 2 cm were counted as one lesion.
Wounds between 2 cm and 5 cm and healing wounds smaller than 5 cm were counted as five
skin lesions. Deep and bleeding wounds greater than 5 cm were counted as 16 skin lesions.
Based on this counting method of the skin lesions, a score of 0 was given for each body part,
respectively, if none or a maximum of four skin lesions were visible. A score of 1 was given
if five to ten skin lesions were found. A score of 2 was given if 11 or more skin lesions were
visible. At the end, scores of all body parts were combined into one total score for each sow.
A total score of 0 was given if all body parts had an individual score of 0. A total score of 1
was given if any body part had an individual score of 1 or if only one body part had an
individual score of 2. A total score of 2 was given if two or more body parts had an individual
score of 2.
Developmental stages of shoulder ulcer
The developmental stages of a shoulder ulcer of any body sides were assessed as described by
Lundgren et al. (2012). A scale from 0 to 4 was used. A score of 0 indicated no ulcers. A
score of 1 was given for ulcers in the epidermal layer of the skin with possible crust
formation. Ulcers in the epidermal and dermal layers of the skin with crusts and possible scar
tissue were given a score of 2. A score of 3 was assigned for ulcers in the subcutaneous layer
of the skin with crust formations. The highest score of 4 was given for ulcers that reached
deep into the muscles with possibly visible shoulder bone. Both sides of the shoulder were
scored and a total score for each sow was given regarding the higher score of one shoulder.
Skin lesions of the udder
All skin lesions of the udder of the sows were documented by using the scoring method
developed by Hansson and Lundeheim (2012). The scores between 0 and 3 described the
extent of skin lesions at the udder region. A score of 0 indicated that no skin lesions at the
udder were visible. A score of 1 described that less than 50 % of the udder showed skin
80
lesions. A score of 2 was given if more than 50 % skin lesions were found. And at least a
score of 3 was scored if skin lesions were visible over the whole udder.
Skin lesions of the vulva
Skin lesions of the vulva were investigated by using the scoring method from Welfare
Quality® (2009). Skin lesions of the vulva were accessed with scores between 0 and 2. No
damaged vulvae or vulvae with small skin lesions of less than 2 cm or with scarred tissue
were awarded a score of 0. A score of 1 was given for vulvae with injuries larger than 2 cm.
Vulvae with injuries larger than 2 cm which were bleeding received a score of 2.
Skin lesions of the tail
Skin lesions of the tail were assessed by using the scoring method from Veit et al. (2016) and
Abriel and Jais (2013). If no skin lesions were visible, a score of 0 was given. A score of 1
was given for scratches and light bite marks. Moderate damage to the tail was given a score of
2 and severe damage a score of 3.
Dirtiness
The visually presence of manure on the body was scored by using the method from Welfare
Quality® (2009). Sows received a score of 0 if 10 % of the body was soiled. A score of 1 was
given if 10- 30 % of the body was soiled and a score of 2 was given if more than 30 % of the
body was soiled. Both body sides were taken into account to assess dirtiness and the score of
the dirtier body side was used.
Piglet individual health indicators
All piglets of both housing systems were scored in the first week and fourth week post partum
with regard to different health indicators, which are described below. The health indicator for
skin lesions of the body was only evaluated in the fourth week post partum.
Skin lesions of the face
Skin lesions of the face were scored by using a modified scoring method from
Brown et al. (1996). The scoring scale covered four different scores. A score of 0 was given if
piglets had no visible skin lesions. Superficial skin lesions received a score of 1. Scratches,
deeper cuts and dark scabs were documented as score 2. A score of 3 was given if the piglets
had lacerations and deep or infected wounds.
81
Skin lesions of the carpus
For skin lesions of the carpus the scoring scale from Zoric et al. (2008) was used. The scale
included four different scores. No skin lesions of the carpus were scored as 0. A score of 1
was given for loss of hair or hairless locations of the carpus and hornification of the carpus. A
score of 2 documented skin abrasions of the carpus. Piglets with skin wounds or scabs of the
carpus received a score of 3.
Skin lesions of the body
To investigate skin lesions of the body, the piglets were divided into three body parts
(part 1- shoulder, front; part 2- back, abdomen; part 3- hindquarters). Only the left side of the
body was scored. A modified scoring method from Stukenborg et al. (2012) and
Bohnenkamp et al. (2013b) was used for this evaluation. If no skin lesions or sporadically
minor skin lesions were visible, a score of 0 was given. Several skin lesions on a body part
without accumulation were evaluated as medium skin injuries, which received a score of 1.
A score of 2 was given for severe skin lesions in a body part with accumulation. At the end,
the scores of all body parts were combined to one total score for each piglet. A total score of 0
was given if all body parts had an individual score of 0. A total score of 1 was given if any
body part had an individual score of 1. A total score of 2 was given if any body part had an
individual score of 2.
Saliva collection and cortisol analysis
Saliva was collected between 8.00 h and 9.00 h in the morning on two successive days. The
sows were tested two weeks ante partum to obtain their basal cortisol levels. Furthermore,
saliva was collected from the sows two weeks post partum and four weeks post partum to test
the effect of the different housing systems during lactation. The sows were allowed to chew
on a Salivette® (Salivette® Cortisol, Sarstedt, Germany) until the swab was fully moist
(about 30 - 60 s). Swabs were placed in tubes and stored on ice. In the laboratory, at 11:00 h,
the saliva samples were centrifuged for 2 min at 1500×g and stored at -18 °C until analysis.
Salivary cortisol was determined using an enzyme immunoassay kit (Cortisol Speichel
ELISA, ACoat-a-Count Cortisol TKCO, IBL International, Hamburg, Germany). To avoid
inter-assay differences, samples were assayed on the same day in duplicate.
82
Statistical analysis
All data were analysed with the statistical software SAS® 9.4 (SAS Institute Inc., 2008). Fixed
effects were added to the model in a stepwise manner. The Akaike’s information criteria
corrected (AICC) and the Bayesian information criteria (BIC) were used to compare the
different models. The model with the smallest AICC and BIC values was chosen. The health
indicators of the sows and the piglets were documented as multinomial data during data
collection. Due to unevenly distributed data in the different categories (Table 1), the health
indicators were summarised to binomial categories (0=no alteration; 1=alteration) for analysis
purposes. In addition, the data for the health indicators for the sows and piglets and the saliva
cortisol levels for the sows of the two GH systems (GHbig and GHsmall) were combined
because no notable statistical differences were found.
The data of the health indicators of the sows were analysed using the GLIMMIX procedure
with a binomial distribution (link-function=logit), respectively for the ante partum and post
partum scoring. The model for the data of the health indicators of the sows included the fixed
effects housing system (GH, SH), batch (B1-B4), parity class (class 1: 1; class 2: 2-4;
class 3: ≥5), the interaction between housing system and batch, and between housing system
and parity class. The interactions were removed if no significant effects were found. The
significance of differences in the least square means was adjusted by the
Bonferroni-correction.
The data for the health indicators for the piglets were analysed using the GLIMMIX
procedure with a binomial distribution (link-function=logit). The model for the data of the
health indicators of the piglets included the fixed effects housing system (GH, SH), batch
(B1-B4), lactation week (W1, W4), the interaction between housing system and batch, and
between housing system and lactation week. The interactions were removed if no significant
effects were found. In addition, the piglet was added as a random effect nested in the housing
system and batch. The significance of differences in the least square means was adjusted by
the Bonferroni-correction.
The data of the condition of the sows (body weight, Body Condition Score, back fat
thickness) were analysed with the MIXED procedure. The mixed model included the fixed
effects housing system (GH, SH), batch (B1-B4), parity class (class 1: 1; class 2: 2-4;
class 3: ≥5), the interaction between housing system and batch, and between housing system
and parity class. The interactions were removed if no significant effects were found. The
model of body weight, Body Condition Score and back fat thickness post partum included the
body weight, Body Condition Score and back fat thickness ante partum as linear continuous
83
variables. The significance of differences in the least square means was adjusted by the
Bonferroni-correction.
The data of the saliva cortisol samples were not normally distributed and therefore the
logarithmic transformation was used. The logarithmic data were analysed with the MIXED
procedure including the fixed effects housing system (GH, SH), batch (B1-B4), parity class
(class 1: 1; class 2: 2-4; class 3: ≥5), cortisol sample period (basal level, lactation week 2,
lactation week 4) and cortisol sample day (sample 1, sample 2) nested in the cortisol sample
period. In addition, the interaction was tested between housing system and batch, between
housing system and sample period, and between housing system and parity class. The
interactions between housing system and batch, between housing system and parity class were
removed because of no significance. Furthermore, the sow was added as a random effect
nested in the housing system and batch. The significance of differences in the least square
means was adjusted by the Bonferroni-correction.
RESULTS
Health indicators of the sows
Housing system
All sows were evaluated regarding different health indicators one week ante partum and four
weeks post partum when the piglets were weaned (Table 2). Health scores that indicated
serious lesions and injuries were the exception in both housing systems. Furthermore, both the
GH sows and SH sows did not differ significantly concerning the different health indicators
one week ante partum. Whereas four weeks post partum the GH sows had significantly fewer
skin lesions of the udder and of the tail compared to the SH sows (p<0.05). Furthermore, the
GH sows showed a higher incidence of skin lesions of the body and were significantly dirtier
compared to the SH sows (p<0.05).
Skin lesions of the udder, vulva and body decreased during lactation for all housing systems.
However, skin lesions of the tail and the incidence of shoulder ulcers and problems with the
locomotion increased during lactation.
The GH sows and the SH sows did not differ significantly one week ante partum with regard
to body condition. After four weeks of lactation (Table 2), the GH sows were significantly
heavier and had a higher back fat thickness compared to the SH sows (p<0.05).
84
For the Body Condition Score, the Interaction was significant, therefore, in batch 4 the GH
sows had a significant higher Body Condition Score compared to the SH sows (GH:
B1: 2.93 ± 0.06; B2: 3.14 ± 0.06; B3: 3.25 ± 0.07; B4: 3.69 ± 0.07 vs. SH: B1: 2.84 ± 0.07;
B2: 2.96 ± 0.07; B3: 3.04 ± 0.07; B4: 2.97 ± 0.07; p<0.05).
Parity class
With regard to health indicators for the three parity classes, sows from parity class 2 had
significantly more skin lesions of the body one week ante partum compared to sows from
parity class 3 (class 1: 91.6 % ± 5.93 vs. class 2: 97.3 % ± 2.02 vs. class 3: 81.0 % ± 5.86;
p<0.05). However, four weeks post partum, sows from parity class 2 showed the lowest
incidence of skin lesions of the body and had significantly fewer lesions compared to sows
from parity class 1 (class 1: 18.5 % ± 8.96 vs. class 2: 3.50 % ± 2.21 vs. class 3:
12.2 % ± 4.95; p<0.05).
With regard to differences in the condition between the three parity classes one week ante
partum, all parity classes differed significantly from each other with regard to their weight
(class 1: 221 kg ± 4.87 vs. class 2: 256 kg ± 3.03 vs. class 3: 276 kg ± 3.03; p<0.05), sows
from parity class 2 had a significantly higher back fat thickness compared to sows from parity
class 3 (class 1: 16.5 mm ± 0.55 vs. class 2: 16.3 mm ± 0.35 vs. class 3: 15.0 mm ± 0.35;
p<0.05). Sows from parity class 1 had a significantly higher Body Condition Score compared
to sows from parity class 3 one week ante partum (class 1: 3.88 ± 0.08 vs. class 2: 3.77 ± 0.05
vs. class 3: 3.67 ± 0.05; p<0.05). Four weeks post partum, sows from parity class 3 were
significantly heavier compared to the other two parity classes (class 1: 210 kg ± 7.53 vs.
class 2: 229 kg ± 4.04 vs. class 3: 244 kg ± 4.36; p<0.05), had a significantly higher back fat
thickness (class 1: 12.9 mm ± 0.45 vs. class 2: 14.1 mm ± 0.29 vs. class 3: 14.6 mm ± 0.29;
p<0.05) and Body Condition Score compared to sows from parity class 1 (class 1: 2.98 ± 0.06
vs. class 2: 3.11 ± 0.04 vs. class 3: 3.22 ± 0.04; p<0.05).
85
Table 1
Frequencies of different categories of health indicators for the sows one week ante partum and
four weeks post partum and for the piglets one week and four weeks post partum. Skin lesions
of the tail and dirtiness of the sows and skin lesions of the body of the piglets were only
scored four weeks post partum.
Sows
Health indicator Category Week 1
ante partum
Week 4
post partum
(n=144) (n=140)
Skin lesions of the udder
0 0039 0064
1 0098 0068
2 007 0008
3
000 0000
Skin lesions of the body
0 0013 0115
1 0082 0020
2
0049 0005
Developmental stages of shoulder ulcer
0 0141 0122
1 0003 0010
2 0000 0005
3 0000 0003
4
0000 0000
Locomotion
0 0132 0107
1 0008 0026
2 0004 0007
3
0000 0000
Skin lesions of the tail*
0 000- 0040
1 000- 0022
2
000- 0008
Dirtiness
0 000- 0049
1 000- 0059
2 000- 0032
Piglets
Health indicator Category Week 1
post partum
(n=1814)
Week 4
post partum
(n=1696)
Skin lesions of the face
0 0877 1107
1 0498 0449
2 0355 0117
3
0084 0023
Skin lesions of the carpus
0 0750 0063
1 0361 1254
2 0552 0362
3
0151 0017
Skin lesions of the body
0 000- 1312
1 000- 0315
2 000- 0069
* Skin lesions of the tail were documented in batches 3 and 4 (n=70).
86
Table 2
LSMeans with standard error of the percentages of the health indicators and LSMeans with
standard error of the data of the condition of the sows from the group-housing system (GH)
and the single-housing system (SH) one week ante partum and four weeks post partum.
Dirtiness and skin lesions of the tail were only documented four weeks post partum.
Health indicator
Week 1
ante partum1)
Week 4
post partum
Sows
(n=144)
GH
(n=77)
SH
(n=63)
Skin lesions of the udder 72.1 ± 5.60 42.5a ± 6.10 71.1b ± 6.06
Skin lesions of the body 94.5 ± 3.53 28.4a ± 6.12 2.70b ± 1.97
Developmental stages of shoulder ulcer 0.00 7.85a ± 3.23 10.7a ± 4.17
Locomotion 0.00 21.1a ± 5.00 22.1a ± 5.55
Skin lesions of the vulva 34.9 ± 5.94 0.00 0.00
Dirtiness - 86.3a ± 4.26 39.4b ± 7.35
Skin lesions of the tail* - 29.7a ± 8.41 59.1b ± 9.37
Body weight (kg) 251 ± 2.90 234a ± 3.68 222b ± 4.07
Back fat thickness (mm) 15.9 ± 0.33 14.5a ± 0.25 13.2b ± 0.29
1) Pooled LSMeans with standard error of the health indicators ante partum for GH and SH sows because of no
significant differences (p<0.05). a-b Significant differences between the housing systems (p<0.05).
* Skin lesions of the tail were documented in batches 3 and 4 (GH: n=38; SH: n=32).
Health indicators of the piglets
All piglets were evaluated regarding different health indicators one week and four weeks post
partum. Skin lesions of the body were only documented four weeks post partum as an
exception (Figure 2).
The GH piglets had significantly fewer skin lesions of the face in batches 1, 2, 3 and 4
compared to the SH piglets (GH: B1: 20.7 % ± 1.89; B2: 56.2 % ± 2.36; B3: 28.7 % ± 2.28;
B4: 45.8 % ± 2.34 vs. SH: B1: 56.1 % ± 2.61; B2: 33.2 % ± 2.43; B3: 45.8 % ± 2.64;
B4: 59.5 % ± 2.57; p<0.05). Furthermore, the GH piglets had significantly more skin lesions
of the carpus in batches 1 and 2 compared to the SH piglets (GH: B1: 93.0 % ± 1.17;
B2: 87.9 % ± 1.78; B3: 85.9 % ± 2.07; B4: 92.9 % ± 1.17 vs. SH: B1: 78.6 % ± 2.26;
B2: 72.5 % ± 2.58; B3: 82.9 % ± 2.00; B4: 92.9 % ± 1.13; p<0.05). Skin lesions of the body
87
were documented significantly more frequent for the GH piglets in batches 2, 3 and 4
compared to the SH piglets (GH: B1: 32.5 % ± 3.10; B2: 21.3 % ± 2.74; B3: 37.3 % ± 3.42;
B4: 39.4 % ± 3.26 vs. SH: B1: 26.2 % ± 3.07; B2: 9.1 % ± 2.06; B3: 6.4 % ± 1.73;
B4: 6.7 % ± 1.74; p<0.05).
With regard to differences between the first and the fourth week of lactation, the GH piglets
had significantly fewer skin lesions of the face in week 1 of lactation compared to the SH
piglets and no differences were found in week 4 of lactation between the housing systems
(GH: W1: 39.6 % ± 1.68; W4: 34.0 % ± 1.68 vs. SH: W1: 64.3 % ± 1.70; W4: 33.1 % ± 1.70;
p<0.05). Furthermore, the GH piglets had significantly more skin lesions of the carpus in
week 4 of lactation compared to the SH piglets (GH: W1: 60.7 % ± 1.61; W4: 98.3 % ± 0.43
vs. SH: W1: 57.1 % ± 1.78; W4: 95.0 % ± 0.75; p<0.05).
Figure 2
LSMeans with standard error of the percentage
of the health indicators (skin lesions of the
face (a), body (b) and carpus (c)) of the piglets
for the interaction between housing system
(group-housing system (GH); single-housing
system (SH)) and batch (B1-B4).
88
Saliva cortisol levels
Table 3 shows differences in saliva cortisol levels between the GH sows and the SH sows
with regard to the three sample periods (basal level, lactation week 2, lactation week 4) and
the two successive sample days for each sample period. The basal cortisol levels which were
collected two weeks ante partum did not differ statistically between the housing systems.
Significantly higher saliva cortisol levels were detected for the GH sows compared to the SH
sows (p<0.05) two weeks post partum and four weeks post partum. Furthermore, the basal
cortisol levels of SH sows were significantly higher compared to their cortisol levels four
weeks post partum (p<0.05).
Moreover, two weeks ante partum the cortisol levels of the second sample day were
significantly higher compared to the first sample day (p<0.05). However, the two successive
sample days did not differ significantly from each other (p<0.05), respectively for the sample
period in the second and fourth week of lactation.
Table 3
LSMeans with standard error of saliva cortisol levels (nmol/l) of group-housed sows (GH)
and single-housed sows (SH).
Sample period
Saliva cortisol levels (nmol/l)1)
GH
(samples=398)
SH
(samples=306)
Sample day
(first / second)
Week 2 ante partum (basal) 4.71 a,A ± 1.06 4.44a,A ± 1.07 3.941 ± 1.06 / 5.312 ± 1.06
Lactation week 2 5.53a,A ± 1.06 3.67b,AB ± 1.06 4.62 ± 1.06 / 4.44 ± 1.06
Lactation week 4 4.71a,A ± 1.06 3.29b,B ± 1.07 3.67 ± 1.06 / 4.26 ± 1.06
1) Retransformed data of the logarithmic data. a-b Significant differences of the transformed data between the housing systems (p<0.05). A-B Significant differences of the transformed data between the sample periods within the housing systems
(p<0.05). 1-2 Significant differences of the transformed data between the sample days within the sample periods (p<0.05).
89
DISCUSSION
Health indicators of the sows
No alarming health problems were documented for the GH sows and the SH sows and their
piglets. Furthermore, health scores that indicated serious lesions and injuries were the
exception in both housing systems.
With regard to differences in the health indicators, the GH sows had fewer skin lesions of the
udder compared to the SH sows. Hultén et al. (1995) observed also in a study that GH sows
have less frequent teat and udder lesions during lactation compared to SH sows. In addition,
pre-weaning atrophy of all mammary glands was found for 6.6 % in GH sows. However, no
atrophy of the mammary gland was found for SH sows. Hultén et al. (1995) concluded that
the sow-piglet-interaction decreases during lactation for GH sows. Bohnenkamp et al. (2013a)
did not find differences in suckling frequency and duration between GH sows and SH sows,
however, detected that gilts have more injured teats during lactation possibly due to more
sensitive udders compared to older sows (Bohnenkamp et al., 2013c). The study by
Pajor et al. (2000) showed that sows reduce their time with the offspring during a five-week
lactation and that sows differ in their maternal motivation to spend time with the piglets. It
was also shown that sows voluntary choose a piglet-free area. Thus, a constant confinement of
sows together with their litter in SH pens is converse to their natural preference. In the present
study, GH sows had more freedom of movement and therefore also more freedom to regulate
suckling, which had a positive effect on udder health. In addition, GH sows had fewer skin
lesions of the tail compared to SH sows. The SH sows were permanently exposed to the oral,
manipulable behaviour of the piglets due to being fixed in a crate. The higher prevalence of
skin lesions of the tail of SH sows is possibly caused by the biting of the piglets.
Furthermore, GH sows had more skin lesions of the body compared to SH sows during
lactation. GH sows were composed of small groups during lactation, which again resulted in
ranking fights to organise a new group structure. Gilts were introduced to the older sows for
the first time. In addition, in the present study, sows from parity class 1 (gilts) and sows from
parity class 3 (5 and more parities) showed the highest incidence of skin lesions of the body.
It is obvious that these two parity classes had the most frequent aggressive interactions with
each other. Studies have also shown that pregnant sows housed in a group also have increased
skin lesions caused by ranking fights (Arey, 1999; Barney and Widowski, 2006). Thomson et
al. (2015) reported that gilts receive more attacks and initiate fewer attacks, however, older
sows initiate more attacks and receive fewer attacks.
90
At least living in a GH system results in dirtier sows due to a communal dunging and resting
area in the running area. No differences were found between GH sows and SH sows regarding
the prevalence of shoulder ulcer, lameness and other lesions. However, the study by Marchant
and Broom (1996) showed that more freedom of movement as present in the GH system
results in higher muscle weights and higher bone strength in sows and thus in better
locomotion.
In the present study, GH sows had a higher body condition resulting from ad libitum feeding
of the sows to prevent aggression over feed. The studies by Grimberg-Henrici et al. (2015)
and Bohnenkamp et al. (2013c) showed that if GH sows and SH sows are fed the same
commercial meal, then no differences in the sows’ condition are found. Furthermore, in the
present study, sows from parity class 1 (gilts) showed the highest reduction with regard to
their back fat thickness and Body Condition Score after four weeks of lactation.
Health indicators of the piglets
With regard to piglet health, GH piglets showed a lower prevalence for skin lesions of the
face compared to SH piglets. In addition, GH piglets had fewer skin lesions of the face in the
first week of lactation and showed an equivalent frequency for skin lesions of the face in the
fourth week of lactation compared to SH piglets. This indicates that the free-farrowing pens
had an influence on the prevalence of skin lesions of the face in the first week of lactation and
that multi-suckling in the fourth week of lactation had no negative effect on the piglet health
due to cross-suckling and possible fights for the sow’s teats. However,
van Nieuwamerongen et al. (2015) found a higher average snout damage score for piglets
from a GH system before weaning compared to piglets from a SH system probably due to
completion at the udder and cross-suckling. Although the litters were standardised to
14 piglets per sow and the piglets were mixed six days post partum as in the present study.
Furthermore, skin lesions of the carpus did not vary widely between GH and SH piglets,
although GH piglets were housed mainly on a concrete floor during lactation. Moreover,
GH piglets had significantly more skin lesions of the body compared to SH piglets. These
skin lesions are due to social interactions between piglets of different litters
(Bohnenkamp et al., 2013b). Mixing of different litters during lactation corresponds to their
natural behaviour (Jensen, 1986) and has advantages for their social development. The mixing
of litters within the first two weeks post partum resulted in less aggression and fewer skin
lesions compared to the mixing of litters at weaning, which resulted in intensive fights
between the piglets (Pitts et al., 2000; Kutzer et al., 2009; Bohnenkamp et al., 2013b).
91
However, van Nieuwamerongen et al. (2015) detected no differences in skin lesions due to
agonistic behaviour between piglets of a GH system and piglets of a SH system before
weaning. Also Kutzer et al. (2009) did not find any differences in skin lesions between litters
that had contact with other litters and unmixed litters. A reason for more skin lesions of the
body of the GH piglets in the present study could be due to the higher numbers of mixed
litters (ten litters) compared to the study by van Nieuwamerongen et al. (2015) (five litters).
An increased number of mixed litters possibly increases the number of social interactions
between the piglets.
Saliva cortisol levels
The measured saliva cortisol levels of the sows in the present study are comparable to saliva
cortisol levels of gilts in the study by Geverink et al. (2003). Here, saliva cortisol levels of
gilts ranged around 1.1 to 6.6 nmol/l in the morning samples between 8.00 h and 9.00 h. The
high range of the cortisol levels can be explained by the classification of the gilts as
high-resident and low-resistant during a backtest. In the study by Karlen et al. (2007), saliva
cortisol levels of sows in the first week of gestation were between 4.0 and 6.3 nmol/l.
The results of the present study suggest that the housing system has an influence on saliva
cortisol levels of the sows. GH and SH sows did not differ during gestation with regard to
their cortisol levels. However, when they lived in the different housing systems during
lactation, higher cortisol levels were measured for GH sows compared to SH sows.
Karlen et al. (2007) also found higher saliva cortisol levels during gestation in GH sows
compared to SH sows. In addition, Geverink et al. (2003) documented higher cortisol levels of
enriched group-housed gilts compared to individual barren-housed gilts. These studies explain
the higher cortisol levels of enriched group-housed sows due to more physical activity
because of more environmental stimuli and due to social interactions and ranking fights with
other sows. Studies related to the period around birth and several days after birth have
reported lower cortisol levels for sows housed in pens compared to sows fixed in crates
(Lawrence et al., 1994; Jarvis et al., 2002; Oliviero et al., 2008). They explain the higher
cortisol levels of crated sows due to fixation, longer birth durations, restricted contact to the
piglets and lack of nest material.
However in the present study, the cortisol levels were collected two weeks ante partum and in
the second and fourth week of lactation. The period around farrowing was not taken into
account. Thus, in this study, the higher saliva cortisol levels are more likely explained by
more physical activity and more social interactions. Krandendonk et al. (2007) investigated
92
GH and SH systems for pregnant sows and also stated higher cortisol levels for the GH sows
which were explained by higher physical activity of the sows. Furthermore in the present
study, the sows were allocated to new small groups for the GH system and gilts were
introduced into the group structure for the first time. Cauret et al. (2009) investigated repeated
regrouping of pregnant gilts and detected that instable groups (repeated regrouping) have
significantly higher cortisol levels compared to stable groups. In addition, the stage of
lactation seemed to influence the stress levels of sows as well. Thomsson et al. (2015)
grouped sows and their piglets with other sows together one week, two weeks and three
weeks after birth in a GH system. They concluded by analysing the cortisol levels of these
sows that mixing at three weeks post partum is less stressful for the sows compared to mixing
at one week post partum. They explained the higher cortisol levels of the sows when mixing
at one week post partum compared to three weeks post partum by the more stressful managing
of nursing for the sows in a group. A decrease in the cortisol levels of GH sows during
lactation was also observed in the present study, however, this decrease could not be
confirmed statistically.
Moreover, the basal levels of SH sows during gestation without permanent fixation in crates
are significantly higher compared to the cortisol levels in the fourth week of lactation with
permanent fixation in crates. Although, permanent fixation has been found to be stressful for
sows (Arellano et al., 1992), this cannot be confirmed in the present study. The higher levels
ante partum can be explained by social stress of the sows due to ranking fights when living in
a large group during gestation (Arey, 1999; Barney and Widowski, 2006).
With regard to differences between the days of sampling, the cortisol levels at the second day
in the sample period two weeks ante partum were significantly higher compared to the
cortisol levels of the first day of sampling. At the second day of sampling, the sows knew the
upcoming procedure and could possibly try to avoid the situation. This avoidance strategy
could be a possible explanation for the increased cortisol levels at the second day of sampling.
Especially in the gestation station the sows had more space and opportunities to shirk the
sample procedure.
93
CONCLUSION
The present study shows that more freedom of movement for GH sows results in a possible
better regulation of nursing and thus a lower prevalence of udder lesions compared to SH
sows. Furthermore, the lower risk of udder lesions in GH sows and the equal prevalence for
skin lesions of the face for GH piglets compared to SH piglets indicates that multi-suckling in
a group is not a health risk. In addition, multi-suckling did not result in an impaired condition
in GH sows. The higher number of body lesions of GH sows and piglets was caused by social
interactions. For piglets, contacts with other litters are fundamental to develop social skills in
later life. Furthermore, the analysis of the saliva cortisol samples of the sows detected clear
differences in stress levels of GH sows and SH sows during lactation. A possible explanation
for the higher cortisol levels for GH sows could be social stress and increased physical
activity. Thus, the sows in the group-housing system were subject to more stress compared to
the sows in the single-housing system.
ACKNOWLEDGMENTS
The project was supported by funds of the German Government’s Special Purpose Fund held
at Lantwirtschaftliche Rentenbank.
94
REFERENCES
Abriel, M., Jais, C., 2013. Influence of housing conditions on the appearance of cannibalism
in weaning piglets. Landtechnik 68, 389-393.
Arellano, P.E., Pijoan, C., Jacobson, L.D., Algers, B., 1992. Stereotyped behaviour, social
interactions and suckling pattern of pigs housed in groups or in single crates. Applied
Animal Behaviour Science 35, 157-166.
Arey, D.S., 1999. Time course for the formation and disruption of social organisation in
group-housed sows. Applied Animal Behaviour Science 62, 199-207.
Arey, D.S., Sancha, E.S., 1996. Behaviour and productivity of sows and piglets in a family
system and in farrowing crates. Applied Animal Behaviour Science 50, 135-145.
Barney, D., Widowski, T.M., 2006. Assessment of a group-housing system for gestating
sows: Effects of space allowance and pen size on the incidence of superficial skin
lesions, changes in body condition, and farrowing performance. Journal of Swine
Health and Production 14, 89-96.
Bohnenkamp, A.-L., Meyer, C., Müller, K., Krieter, J., 2013a. Group housing with
electronically controlled crates for lactating sows. Effect on farrowing, suckling and
activity behavior of sows and piglets. Applied Animal Behaviour Science 145, 37-43.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013b. Comparison of
growth performance and agonistic interaction in weaned piglets of different weight
classes from farrowing systems with group or single housing. animal 7, 309-315.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013c. Group housing for
lactating sows with electronically controlled crates: 1. Reproductive traits, body
condition, and feed intake. Journal of animal science 91, 3413-3419.
Brown, J.M.E., Edwards, S.A., Smith, W.J., Thompson, E., Duncan, J., 1996. Welfare and
production implications of teeth clipping and iron injection of piglets in outdoor
systems in Scotland. Preventive Veterinary Medicine 27, 95-105.
Couret, D., Otten, W., Puppe, B., Prunier, A., Merlot, E., 2009. Behavioural, endocrine and
immune responses to repeated social stress in pregnant gilts. animal 3, 118-127.
Fagen, R., 1981. Animal play behavior.
Geverink, N.A., Schouten, W.G.P., Gort, G., Wiegant, V.M., 2003. Individual differences in
behaviour, physiology and pathology in breeding gilts housed in groups or stalls.
Applied Animal Behaviour Science 81, 29-41.
Grimberg-Henrici, C.G.E., Büttner, K., Meyer, C., Krieter, J., 2016. Does housing influence
maternal behaviour in sows? Applied Animal Behaviour Science 180, 26-34.
95
Grimberg-Henrici, C.G.E., Vermaak, P., Bolhuis, J.E., Nordquist, R.E., van der Staay, F.J.,
2015. Effects of environmental enrichment on cognitive performance of pigs in a
spatial holeboard discrimination task. Animal cognition, 1-13.
Hansson, M., Lundeheim, N., 2012. Facial lesions in piglets with intact or grinded teeth. Acta
Veterinaria Scandinavica 54, 23.
Held, S.D.E., Špinka, M., 2011. Animal play and animal welfare. Animal Behaviour 81, 891-
899.
Hultén, F., Lundeheim, N., Dalin, A.M., Einarsson, S., 1995. A field study on group housing
of lactating sows with special reference to sow health at weaning. Acta Veterinaria
Scandinavica 36, 201-212.
Jarvis, S., Calvert, S.K., Stevenson, J., VanLeeuwen, N., Lawrence, A.B., 2002. Pituitary-
adrenal activation in pre-parturient pigs (Sus scrofa) is associated with behavioural
restriction due to lack of space rather than nesting substrate. Animal Welfare 11, 371-
384.
Jarvis, S., D'Eath, R.B., Robson, S.K., Lawrence, A.B., 2006. The effect of confinement
during lactation on the hypothalamic-pituitary-adrenal axis and behaviour of
primiparous sows. Physiology & behavior 87, 345-352.
Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs.
Applied Animal Behaviour Science 16, 131-142.
Karlen, G.A.M., Hemsworth, P.H., Gonyou, H.W., Fabrega, E., Strom, A.D., Smits, R.J.,
2007. The welfare of gestating sows in conventional stalls and large groups on deep
litter. Applied Animal Behaviour Science 105, 87-101.
Kranendonk, G., Van der Mheen, H., Fillerup, M., Hopster, H., 2007. Social rank of pregnant
sows affects their body weight gain and behavior and performance of the offspring.
Journal of animal science 85, 420-429.
Kutzer, T., Bünger, B., Kjaer, J.B., Schrader, L., 2009. Effects of early contact between non-
littermate piglets and of the complexity of farrowing conditions on social behaviour
and weight gain. Applied Animal Behaviour Science 121, 16-24.
Lawrence, A.B., Petherick, J.C., McLean, K.A., Deans, L.A., Chirnside, J., Gaughan, A.,
Clutton, E., Terlouw, E.M.C., 1994. The effect of environment on behaviour, plasma
cortisol and prolactin in parturient sows. Applied Animal Behaviour Science 39, 313-
330.
96
Lundgren, H., Zumbach, B., Lundeheim, N., Grandinson, K., Vangen, O., Olsen, D.,
Rydhmer, L., 2012. Heritability of shoulder ulcers and genetic correlations with mean
piglet weight and sow body condition. animal 6, 1-8.
Marchant, J.N., Broom, D.M., 1996. Effects of dry sow housing conditions on muscle weight
and bone strength. Animal science 62, 105-113.
Oliviero, C., Heinonen, M., Valros, A., Hälli, O., Peltoniemi, O.A.T., 2008. Effect of the
environment on the physiology of the sow during late pregnancy, farrowing and early
lactation. Animal reproduction science 105, 365-377.
Pajor, E.A., Kramer, D.L., Fraser, D., 2000. Regulation of contact with offspring by domestic
sows: temporal patterns and individual variation. Ethology 106, 37-51.
Phillips, C.E., Farmer, C., Anderson, J.E., Johnston, L.J., Shurson, G.C., Deen, J., Keisler,
D.H., Conner, A.M., Li, Y.Z., 2014. Preweaning mortality in group-housed lactating
sows: Hormonal differences between high risk and low risk sows. Journal of animal
science 92, 2603-2611.
Pitts, A.D., Weary, D.M., Pajor, E.A., Fraser, D., 2000. Mixing at young ages reduces
fighting in unacquainted domestic pigs. Applied Animal Behaviour Science 68, 191-
197.
SAS Institute Inc., 2008. User's Guide (release 9.4). Cary, North Carolina, USA.
Stukenborg, A., Traulsen, I., Stamer, E., Puppe, B., Krieter, J., 2012. The use of a lesion score
as an indicator for agonistic behaviour in pigs. Arch Tierz 55, 163-170.
Thomsson, O., Bergqvist, A.-S., Sjunnesson, Y., Eliasson-Selling, L., Lundeheim, N.,
Magnusson, U., 2015. Aggression and cortisol levels in three different group housing
routines for lactating sows. Acta Veterinaria Scandinavica 57, 9.
van Nieuwamerongen, S.E., Soede, N.M., van der Peet-Schwering, C.M.C., Kemp, B.,
Bolhuis, J.E., 2015. Development of piglets raised in a new multi-litter housing
system vs. conventional single-litter housing until 9 weeks of age. Journal of animal
science 93, 5442-5454.
Veit, C., Traulsen, I., Hasler, M., Tölle, K.-H., Burfeind, O., grosse Beilage, E., Krieter, J.,
2016. Influence of raw material on the occurrence of tail-biting in undocked pigs.
Livestock Science 191, 125-131.
Welfare Quality®, 2009. Assessment protocol for pigs. Welfare Quality®.
Zoric, M., Nilsson, E., Mattsson, S., Lundeheim, N., Wallgren, P., 2008. Abrasions and
lameness in piglets born in different farrowing systems with different types of floor.
Acta Veterinaria Scandinavica 50, 37.
97
GENERAL DISCUSSION
The aim of this thesis was to investigate the performance of lactating sows and their piglets in
group-housing systems (GH) and in a single-housing system (SH) with regard to their
feasibility for pig husbandry with the focus especially on reproductive traits, maternal
behaviour and health indicators.
The advantage of a GH system during lactation is that it can better fulfil natural behavioural
needs compared to conventional housing systems in which sows are housed individually.
Jensen et al. (1986) explored the behaviour of conventionally raised domestic sows in a
semi-natural environment ante and post partum. The observed behaviour of these sows was
very similar to documented behaviour of wild boars. The sows chose an appropriate place for
farrowing and built a nest which was comparable with nests of wild boars. Also, piglets
behaved naturally as hiders who waited in the nest. Around three days post partum the piglets
started to follow the sows for short trips. Nine days post partum the sows returned together
with their litters to the group and mixed with the other sows and litters. The sows weaned
their piglets until around 14 to 17 weeks of age. Thus, it can be concluded that our domestic
sows and piglets are able to cope with a GH system and have a genetical and natural need to
perform these behaviours.
Composition of groups and saliva cortisol levels
In nature, the most frequent group compositions are female groups with approximately five
sows (Dardaillon, 1988). These groups are formed by familiarity and relatedness between the
animals (Podgórski et al., 2014). Social contacts between pigs are an important welfare factor
because they naturally live in packs. In the present thesis, groups of six (Chapter One and
Two) and ten (Chapter Three and Four) sows were tested in GH systems. During gestation,
the sows lived in a large group with around 200 other sows. The sows were put into smaller
groups for the time during lactation in the GH systems. These new group compositions
resulted in more aggression between the sows due to ranking fights, which are shown in the
higher incidence of skin lesions of the body of the GH sows described in Chapter Four. In
addition, it was found that housing conditions had an influence on the saliva cortisol levels.
Two weeks ante partum when the sows were housed together in the gestation stable, no
differences between the salvia cortisol levels of the sows could be detected. However, during
lactation, GH sows had significantly higher cortisol levels compared to SH sows.
Higher cortisol levels of GH sows or enriched sows were also documented by
98
Karlen et al. (2007) and Geverink et al. (2003). An explanation for the higher cortisol levels
of sows in these studies was a higher physical activity because of more environmental stimuli,
more social interactions and ranking fights with other sows. Cauret et al. (2009) investigated
repeated regrouping of pregnant gilts and detected that instable groups (repeated regrouping)
had significantly higher cortisol levels compared to stable groups. Although
Arellano et al. (1992) stated that the permanent fixation of sows in crates is very stressful as
they showed significantly more stereotypies, and several studies have also found higher
cortisol levels for crated sows especially around farrowing (Lawrence et al., 1994;
Jarvis et al., 2002; Oliviero et al., 2008), social stress should not be underestimated in
GH systems. A sow’s aggression in a group and its restlessness are major disruptive factors,
thus, stable groups during gestation and lactation are recommended. In addition, smaller
groups (Hemsworth et al., 2013) and housing designs with pen divisions and barriers are also
found to decrease aggression (Edwards et al., 1993; Waran and Broom, 1993).
Another factor which influences aggression in a group is feeding management. Jealousy of
each other’s feed between sows is a common reason for aggression (Csermely and
Wood-Gush, 1987). Ad libitum feeding has been found to prevent aggression between sows
(Barnett et al., 1994). Thus, to prevent aggression, the sows were fed separately (Chapter One
and Two) and were fed ad libitum (Chapter Three and Four). However, ad libitum feeding had
the disadvantage that the condition of the sows could not be controlled as shown by the higher
condition of GH sows in Chapter Four compared to SH sows.
Maternal behaviour
Maternal behaviour has not changed in the progress of domestication (Jensen, 1986;
Stolba and Wood-Gush, 1989). In addition, Špinka et al. (2000) did not detect differences in
maternal behaviour between domestic sows and crossbred sows (wild x domestic). However,
individual differences are found in maternal behaviour between sows (Jensen, 1986;
Špinka et al., 2000; Andersen et al., 2005), although on an individual level the sows were
highly consistent in their maternal performance (Pitts et al., 2002). Several studies have found
genetically correlations between good maternal performance of the sows and certain
behavioural traits such as aggressiveness towards other sows and humans and responsiveness
to their piglets (Grandinson et al., 2003; Hellbrügge et al., 2008b).
In the study of Chapter One, it was investigated whether housing conditions can influence
maternal behaviour. The sows’ reaction to piglet distress calls, separation from, and reunion
with their piglets was tested both in their home pens and in a test arena. An effect of the
99
GH system on maternal behaviour could already be observed in the first litter of the sows.
GH sows showed stronger maternal reactions during testing in their home pens with regard to
piglet distress calls and reunion with their piglets, whereas SH sows reacted more strongly in
the test arena with regard to piglet distress calls. In addition, Wechsler and Hegglin (1997)
discovered that high-responsive sows crushed fewer piglets compared to low-responsive
sows. This is in accordance with the findings of Chapter One that GH sows were highly
responsive in their pens and had fewer piglet losses compared to SH sows. Roehe et al. (2009)
stated that maternal performance should be evaluated under the conditions in which the sows
are kept with their piglets during lactation because enriched environments triggers other
behaviours than barren environments. Furthermore, Baxter et al. (2011) investigated the
influence of the environment and genetics on piglet survival. Sows selected for high-piglet
survival showed lower piglet mortality in an outdoor environment compared to the control
sow line. However, no differences were found in piglet survival between the two lines in an
indoor loose-housing system during lactation. Moreover, the sow line selected for high-piglet
survival was aggressive towards their piglets and performed savaging behaviour in the indoor
environment. This indicates how important it is to evaluate and select the sows regarding their
maternal behaviour in the environment in which the sows are housed during lactation. Thus,
the results in Chapter One of the GH sows in the pens in combination with the lower piglet
mortality can be considered as more meaningful with regard to maternal behaviour compared
to the tests results of the SH sows in the test arena.
The need for good maternal abilities in sows in terms of lower piglet mortality rates due to
crushing becomes more important because the demand for loose-housing systems rises due to
the increased welfare interest of the public. Therefore, maternal tests are a very helpful
instrument in breeding to gain more insights into maternal structures and the behavioural
patterns involved (Grandinson et al., 2003). However, large differences in maternal
performances are found in Chapter One despite testing the same genetics. To obtain more
information on the structure of the maternal behaviour of sows and the significance of these
tests, the observed behavioural parameters of the maternal tests in Chapter One were analysed
with a factor analysis to identify redundancies in behavioural parameters. The interpretation
of the factor loadings revealed four underlying maternal dimensions: communication, care,
contact, and local attachment to the piglets. Communication, thus vocalisation, was extracted
by the factor analysis as the most important maternal factor. This is in accordance with other
studies (Hutson et al., 1991; Weary et al., 1996; Špinka et al., 2000) and with the findings of
100
Chapter One with regard to maternal performances of GH sows, which showed stronger
communication with their piglets.
The piglet scream test which was used in Chapter One loaded on the factor for
Communication. Grandinson et al. (2003) stated a heritability of 6 % and
Hellbrügge et al. (2008b) a heritability of 13 % for the responsiveness of the sows in the
piglet scream test. In their studies, a recorded piglet scream of a crushed piglet was played
when the sows were lying down to observe whether they would stand up to prevent crushing.
A heritability of only 1 % for the responsiveness of the sows was detected
(Grandinson et al., 2003) in a piglet handling test which was implemented in normal routine
treatments of the piglets such as castration.
However, not only maternal tests are able to evaluate maternal performance. Several studies
have found genetic correlations between maternal performance and the aggressiveness of
sows towards humans and other sows. A positive correlation with piglet survival was found in
sows which avoided interactions with humans (Grandinson et al., 2003; Lensink et al., 2009).
Heritabilities of 8 % (Grandinson et al., 2003) during lactation and of 32 % during gestation
(Hellbrügge et al., 2008b) were found for the aggressiveness of sows.
In summary, Chapter One shows that a GH system can promote good maternal performance.
Especially loose-housing systems need reliable maternal abilities to control piglet losses due
to crushing. More consistency in maternal behaviour can be possibly reached by stricter
exclusion of sows with poor maternal performance and higher levels of aggressiveness from
breeding. Moreover, studies have shown that good maternal performance and handling of
sows are compatible. With regard to maternal tests, Chapter Two clarifies how important it is
to use a number of different maternal tests and to interpret different maternal dimensions
correctly. For example, the reunion test in the home pen has been found to be useful to
evaluate the care motivation of the sow to suckle their piglets after reunion, which is essential
for piglet survival. Whereas, the separation test in the test arena investigates the attachment of
the sows to their piglets.
Piglet mortality and pen design
Wild boars have four to five piglets with a range of one to twelve piglets during lactation
(Andersen et al., 2005). Data of farmed wild boars reported 38-40 % piglet mortality with an
incidence of crushed piglets of 17-20 % (Andersen et al., 2005). Piglet mortality due to
crushing in the wild is not documented. Thus, crushing seemed to be not only a genetic
phenomenon of the domestic sows. However, the influence of the housing conditions need to
101
be investigated. Nevertheless, piglet losses due to crushing remain an economical and ethical
issue.
In Chapters One and Three, the sows gave birth on average to 15-17 piglets, which is to be
evaluated as high. The mortality rates of live born piglets were around 18-20 %
(2.9-3.2 piglets/sow) for SH sows. For the GH sows in Chapter One, get-away pens
(sow can leave the pen while piglets stay in the pen) equipped with farrowing crates with a
fixation of the sows until one day post partum were observed to decrease piglet mortality
immensely (up to 11 %; 1.8 piglets/sow). Bohnenkamp et al. (2013c) reported in a previous
study with the same sow genetics in the same housing systems equal piglet losses for
GH sows (2.2 piglets/sows) and SH sows (2.4 piglets/sow). Van Nieuwamerongen et al.
(2015) observed significantly higher total piglet losses and numbers of crushed piglets for
GH sows compared to SH sows. The GH system was provided with free-farrowing pens for
five sows and the sow had the opportunity to leave the nest immediately after birth without
the piglets. Moreover, in Chapter Three, GH sows were investigated enclosed in
free-farrowing pens until six days post partum together with their litters. Here, the GH sows
had significantly higher total piglet losses and numbers of crushed piglets compared to the
SH sows. Piglet mortality rates of 31 % were found for GH sows. This shows that a clear
increase in piglet losses can be determined if the sows are given more freedom to move and at
the same time are housed together for a longer period after birth with their piglets in the pen
without the opportunity to leave the pen. In contrast, well-structured free-farrowing pens such
as the PigSAFE, Werribee, and SCHMID pens with separate nest, dunging and feeding areas
achieved low piglet mortality rates between 11-18 % (Cronin et al., 2000; Weber, 2000;
Baxter et al., 2015). However, in those studies, sows had on average 11-12 live-born piglets,
whereas sows in our studies gave birth to 15-17 piglets on average as mentioned above.
Several studies stated a positive correlation between litter size and piglet mortality
(Roehe and Kalm, 2000; Weber et al., 2009; Pedersen et al., 2011) and a decrease in birth
weights with increasing litter size (Roehe and Kalm, 2000; Wolf et al., 2008). Moreover, an
optimal birth weight is important for piglet survival and vitality (Baxter et al., 2008).
Hellbrügge et al. (2008a) calculated a heritability of live-born piglets and birth and weaning
weight of 10 %. According to the constant numbers of weaned piglets per litter,
Andersen et al. (2011) recommend 10-11 piglets per sows during lactation. In addition,
Wechsler and Hegglin (1997) observed that the quality of maternal behaviour and the
responsiveness of the sows to their piglets was negatively correlated with the number of
live-born piglets.
102
Thus, smaller litter sizes with higher individual birth weights are beneficial with regard to
piglet survival rates during lactation especially in free-farrowing pens. Moreover, functional
areas in free-farrowing pens or the opportunity for the sow to leave the nest one day after birth
decrease the piglet mortality to the extent that the reproductive performances of the sows are
comparable to sows housed in crates (Weber et al., 2007), which is important for
competitiveness.
With regard to pen design, Chapter Three shows that GH sows in smaller free-farrowing pens
(5.2 m2) had higher piglet losses compared to GH sows in larger pens (6.2 m2). These
observations are also in accordance with the fact that smaller pens offer the sows even fewer
opportunities to differentiate between areas. Weber et al. (2009) did not report significant
effects of the pen size of free-farrowing pens on piglet losses. However, they detected a slight
tendency that sows in smaller pens have more total piglet losses and crush more piglets. They
suggested a minimum pen size of 5 m2. The study by Baxter et al. (2015) showed that piglet
mortality rates in pens with the same pen design can shift if the pen size is even larger. Piglet
mortality increased significantly for sows in free-farrowing pens of 9.7 m2 compared to sows
in free-farrowing pens of only 7.9 m2. An explanation for the higher piglet mortality in large
pens (>9.7 m2) was that the sows had more floor space where they could lie down and roll
without coming into contact with supportive structures such as anti-crush rails.
Another key point for piglet crushing especially in loose-housing systems is the postural
changes of sows such as lying down and rolling behaviour related to the pen design. Domestic
sows observed in semi-natural environments stay together with their piglets in the nest with
close contact to facilitate their bonding for the first two days post partum (Jensen, 1986).
These behavioural patterns constitute a major problem in pig husbandry especially for
loose-housing systems due to the higher piglet losses caused by crushing. In Chapters One
and Three, 65 % of crushed piglets were documented in the first days after birth. This is in
line with results from Marchant et al. (2001) and Kilbride et al. (2012), who also found that
piglets are most vulnerable to crushing in the first 24 h post partum and reported 50 % of
crushed piglets in the first two days after birth. To obtain more information on the behavioural
differences of sows which can lead to critical situations regarding crushing, GH sows with
high numbers of crushed piglets and GH sows with low numbers of crushed piglets were
observed via video for 72 hours post partum in the study of Chapter Three. During the
observation, differences between the lying down and rolling movements and the behaviour of
the piglets were documented. The results show that the lying down and rolling behaviour of
the sows represents a danger for piglets to be crushed in equal numbers for sows with high
103
and low numbers of crushed piglets respectively. Several studies have stated a high incidence
of crushed piglets during rolling movements (Weary et al., 1996; Weary et al., 1998;
Danholt et al., 2011), whereas other studies have described a higher number of crushed piglets
during lying down movements (Weary et al., 1998; Marchant et al., 2001). In Chapter Three,
no differences were found in the frequency of lying down movements between sows with low
or high numbers of crushed piglets. Thus, sows with low numbers of crushed piglets seemed
to lie down more carefully, which is associated with lower piglet mortality (Burri et al., 2009).
In addition, sows with low numbers of crushed piglets performed more lying down
movements by using the pen walls, which is the safest way to lie down according to
Marchant et al. (2001). Moreover, piglets of sows with low numbers of crushed piglets were
more active during lying down, which indicates possibly a more carefully performed pre-lying
behaviour of the sows. In addition, the piglet’s activity during lying down was negatively
correlated with the numbers of crushed piglets in the first 72 hours post partum. Pre-lying
behaviour has been found to minimise crushing (Schmid, 1991; Marchant et al., 2001).
Furthermore, sows with high numbers of crushed piglets roll significantly more frequently
compared to sows with low numbers of crushed piglets. In addition, rolling movements from
one side to the other side are highly positively correlated with the number of crushed piglets
in the first 72 hours post partum. Natural precautionary measures to introduce rolling
movements do not exist, as reviewed by Damm et al. (2005). However, in Chapter Three,
piglets of sows with low numbers of crushed piglets were significantly more active during
rolling movements compared to piglets of sows with high numbers of crushed piglets. This
could indicate more interactions between the sows and their piglets during rolling movements.
Furthermore, pen design has also been found to influence lying down and rolling behaviour.
For example, lying down movements were slowed down when the sows used pen walls
(Marchant et al., 2001). Rolling could be reduced on concrete floors (Weary et al., 1998) and
piglet survival was increased on softer floor types such as straw and sand (Herskin et al.,
1998). However, the main problem for crushing especially during rolling movements is that
piglets and also sows have the natural need to stay in close contact during the first days after
birth (Jensen, 1986). Several studies have shown that it is very difficult to motivate piglets
during the first days after birth to rest in the piglet creep areas, which would be the safest
place (Vasdal et al., 2010; Baxter et al., 2015). Vasdal et al. (2010) tested different types of
piglet creep areas to increase the time of piglets in the nest. However, soft bedding material
and improved thermal comfort did not attract the piglets in the nest. Also Baxter et al. (2015)
found no increase in piglet survival by testing different creep area temperatures. Probably,
104
increasing thermal comfort in the creep area is the wrong approach while the stable
temperature is still around 20°C, which is common in commercial stalls. Lower room
temperatures would possibly increase the acceptance of the warm creep areas by the piglets.
The temperature zone of thermal comfort is around 12-22°C for lactating sows
(Black et al., 1993). However, for piglets, thermal comfort is around 34°C after birth
(Berthon et al., 1993). This mismatch could be useful to control the sows’ and piglets’
positions better.
Health status
The health status of the sows and their piglets was investigated during lactation in Chapter
Four. No alarming health problems in GH and SH sows were documented. However, some
significant differences during lactation regarding the health status of the sows and their piglets
were detected. An interesting result was the lower incidence of skin lesions of the udder of
GH sows compared to SH sows. Also, Hultén et al. (1995) observed in a study that GH sows
had fewer frequent teat and udder lesions during lactation compared to SH sows and
explained this with a decrease in sow-piglet interaction. However, Bohnenkamp et al. (2013a)
did not find differences in suckling frequency and duration between GH sows and SH sows,
but detected that gilts had more injured teats during lactation possibly due to more sensitive
udders compared to older sows (Bohnenkamp et al., 2013c). Pajor et al. (2000) concluded
from their study that a constant confinement of sows together with their litters in SH pens is
converse to their natural preference. Thus, permanent exposure to oral manipulative behaviour
of the piglets of SH sows in Chapter Four could explain the higher incidence of skin lesions of
the udder. In addition, GH piglets had significantly fewer skin lesions of the face during the
first week of lactation. Moreover, no differences in the incidence of skin lesions of the face
were detected between GH and SH piglets in the fourth week of lactation. Contrary to our
results, van Nieuwamerongen et al. (2015) found a higher average snout damage score for
piglets from a GH system before weaning compared to piglets from a SH system, probably
due to competition at the udder and cross-suckling. As in the study in Chapter Four,
van Nieuwamerongen et al. (2015) standardised the different litters to 14 piglets per sow and
the piglets were allowed to run together after 6 days post partum. However, in Chapter Four,
collective suckling improved udder health of the sows and did not result in an increased
incidence of skin lesions of the face of the GH piglets. Thus, it was shown that suckling in a
group is not a health risk for sows and piglets.
105
Moreover, more skin lesions on the body were documented for GH sows and GH piglets
during lactation due to social interactions. As mentioned above, good feeding management
(Barnett et al., 1994), escape possibilities for harassed sows (Hemsworth et al., 2013) and
stable groups (Couret et al., 2009) during gestation and lactation are good adjustments to
reduce aggression and thus skin lesions of the sows. A lower aggression level in the group
means more rest for the sows and fewer risky situations for piglets to become trapped,
whereas social interactions of GH piglets during lactation are fundamental to develop social
skills for later life (Pitts et al., 2000; Kutzer et al., 2009; Bohnenkamp et al., 2013b).
However, van Nieuwamerongen et al. (2015) and Kutzer et al. (2009) did not find any
differences regarding the incidence of skin lesions between litters which had had contact with
other litters and unmixed litters during lactation. A reason for more skin lesions of the body of
the GH piglets in Chapter Four could be due to the number of mixed litters. In Chapter Four,
ten litters were mixed, however, in the study by van Nieuwamerongen et al. (2015) only five
litters ran together in the GH system. Possibly, the number of litters which ran together
influenced the index of social interactions.
Conclusion and outlook
This thesis has shown that GH systems of sows and their piglets can be feasible for pig
husbandry with regard to positive effects on maternal performance and the health of sows and
their piglets. However, housing conditions especially in free-farrowing pens shortly before
parturition and the first days after birth need to be improved to reduce piglet mortality by
creating functional areas in the pens for the sows or by giving the sows the opportunity to
leave the pen. Chapter One showed excellent results regarding piglet mortality with a short
fixation of GH sows in crates with the possibility to leave the pen one day after birth.
Moreover, consistent and reliable maternal performance is mandatory. It was found that
maternal tests evaluate different maternal dimensions. Thus, it is important to use a number of
different maternal tests and to interpret the sows’ performance correctly. In order to reduce
piglet mortality in further studies, sows’ responses in maternal tests should be correlated to
reproductive performances and behavioural performances such as lying down and rolling
movements.
Another factor to reduce the crushing of piglets could be a more natural farrowing
environment. At the moment, pig husbandry tries to motivate piglets to leave the sow and to
rest in a creep area. However, this behaviour is unnatural for the sows and the piglets. Still, a
safe place for the piglets is essential in the first days after birth to decrease the risk of
106
crushing. Thus, for further studies, it would be interesting to investigate the role of
nest-building behaviour for the sows and their piglets. If the sows have the possibility to
perform their natural behaviour and to build functional nests for their piglets, will it decrease
crushing? The fact that the sows themselves determine an appropriate place for their nest in
the pen may influence the acceptance of the piglets to stay in the nest and may promote more
carefulness in the sows. Andersen et al. (2005) detected that lower numbers of crushed piglets
are associated with the increased nest-building behaviour of the sows.
It is to be concluded that a GH system meets the natural needs of the sows and their piglets
and is feasible for pig husbandry. Further research is needed on free-farrowing systems before
implementation to decrease pre-weaning mortality.
107
REFERENCES
Andersen, I.L., Berg, S., Bøe, K.E., 2005. Crushing of piglets by the mother sow (Sus scrofa)
- purely accidental or a poor mother? Applied Animal Behaviour Science 93, 229-243.
Andersen, I.L., Nævdal, E., Bøe, K.E., 2011. Maternal investment, sibling competition, and
offspring survival with increasing litter size and parity in pigs (Sus scrofa). Behavioral
ecology and sociobiology 65, 1159-1167.
Arellano, P.E., Pijoan, C., Jacobson, L.D., Algers, B., 1992. Stereotyped behaviour, social
interactions and suckling pattern of pigs housed in groups or in single crates. Applied
Animal Behaviour Science 35, 157-166.
Barnett, J.L., Cronin, G.M., McCallum, T.H., Newman, E.A., 1994. Effects of food and time
of day on aggression when grouping unfamiliar adult pigs. Applied Animal Behaviour
Science 39, 339-347.
Baxter, E.M., Adeleye, O.O., Jack, M.C., Farish, M., Ison, S.H., Edwards, S.A., 2015.
Achieving optimum performance in a loose-housed farrowing system for sows: the
effects of space and temperature. Applied Animal Behaviour Science 169, 9-16.
Baxter, E.M., Jarvis, S., D’eath, R.B., Ross, D.W., Robson, S.K., Farish, M., Nevison, I.M.,
Lawrence, A.B., Edwards, S.A., 2008. Investigating the behavioural and physiological
indicators of neonatal survival in pigs. Theriogenology 69, 773-783.
Baxter, E.M., Jarvis, S., Sherwood, L., Farish, M., Roehe, R., Lawrence, A.B., Edwards, S.A.,
2011. Genetic and environmental effects on piglet survival and maternal behaviour of
the farrowing sow. Applied Animal Behaviour Science 130, 28-41.
Berthon, D., Herpin, P., Duchamp, C., Dauncey, M.J., Le Dividich, J., 1993. Modification of
thermogenic capacity in neonatal pigs by changes in thyroid status during late
gestation. Journal of developmental physiology 19, 253-261.
Black, J.L., Mullan, B.P., Lorschy, M.L., Giles, L.R., 1993. Lactation in the sow during heat
stress. Livestock Production Science 35, 153-170.
Bohnenkamp, A.-L., Meyer, C., Müller, K., Krieter, J., 2013a. Group housing with
electronically controlled crates for lactating sows. Effect on farrowing, suckling and
activity behavior of sows and piglets. Applied Animal Behaviour Science 145, 37-43.
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013b. Comparison of
growth performance and agonistic interaction in weaned piglets of different weight
classes from farrowing systems with group or single housing. animal 7, 309-315.
108
Bohnenkamp, A.L., Traulsen, I., Meyer, C., Müller, K., Krieter, J., 2013c. Group housing for
lactating sows with electronically controlled crates: 1. Reproductive traits, body
condition, and feed intake. Journal of animal science 91, 3413-3419.
Burri, M., Wechsler, B., Gygax, L., Weber, R., 2009. Influence of straw length, sow
behaviour and room temperature on the incidence of dangerous situations for piglets in
a loose farrowing system. Applied Animal Behaviour Science 117, 181-189.
Couret, D., Otten, W., Puppe, B., Prunier, A., Merlot, E., 2009. Behavioural, endocrine and
immune responses to repeated social stress in pregnant gilts. animal 3, 118-127.
Cronin, G.M., Lefébure, B., McClintock, S., 2000. A comparison of piglet production and
survival in the Werribee Farrowing Pen and conventional farrowing crates at a
commercial farm. Australian Journal of Experimental Agriculture 40, 17-23.
Csermely, D., Wood-Gush, D.G.M., 1987. Different methods of grouping and their effects on
the agonistic behaviour of sows. Applied Animal Behaviour Science 18, 389.
Damm, B.I., Forkman, B., Pedersen, L.J., 2005. Lying down and rolling behaviour in sows in
relation to piglet crushing. Applied Animal Behaviour Science 90, 3-20.
Danholt, L., Moustsen, V.A., Nielsen, M.B.F., Kristensen, A.R., 2011. Rolling behaviour of
sows in relation to piglet crushing on sloped versus level floor pens. Livestock Science
141, 59-68.
Dardaillon, M., 1988. Wild boar social groupings and their seasonal changes in the Camargue,
southern France. Zeitschrift für Säugetierkunde 53, 22-30.
Edwards, S.A., Mauchline, S., Stewart, A.H., 1993. Designing pens to minimise aggression
when sows are mixed. Farm Building Progress (United Kingdom).
Geverink, N.A., Schouten, W.G.P., Gort, G., Wiegant, V.M., 2003. Individual differences in
behaviour, physiology and pathology in breeding gilts housed in groups or stalls.
Applied Animal Behaviour Science 81, 29-41.
Grandinson, K., Rydhmer, L., Strandberg, E., Thodberg, K., 2003. Genetic analysis of on-
farm tests of maternal behaviour in sows. Livestock Production Science 83, 141-151.
Hellbrügge, B., Tölle, K.H., Bennewitz, J., Henze, C., Presuhn, U., Krieter, J., 2008a. Genetic
aspects regarding piglet losses and the maternal behaviour of sows. Part 1. Genetic
analysis of piglet mortality and fertility traits in pigs.
Hellbrügge, B., Tölle, K.H., Bennewitz, J., Henze, C., Presuhn, U., Krieter, J., 2008b. Genetic
aspects regarding piglet losses and the maternal behaviour of sows. Part 2. Genetic
relationship between maternal behaviour in sows and piglet mortality. animal 2, 1281-
1288.
109
Hemsworth, P.H., Rice, M., Nash, J., Giri, K., Butler, K.L., Tilbrook, A.J., Morrison, R.S.,
2013. Effects of group size and floor space allowance on grouped sows: aggression,
stress, skin injuries, and reproductive performance. Journal of animal science 91,
4953-4964.
Herskin, M.S., Jensen, K.H., Thodberg, K., 1998. Influence of environmental stimuli on
maternal behaviour related to bonding, reactivity and crushing of piglets in domestic
sows. Applied Animal Behaviour Science 58, 241-254.
Hultén, F., Lundeheim, N., Dalin, A.M., Einarsson, S., 1995. A field study on group housing
of lactating sows with special reference to sow health at weaning. Acta Veterinaria
Scandinavica 36, 201-212.
Hutson, G.D., Wilkinson, J.L., Luxford, B.G., 1991. The response of lactating sows to tactile,
visual and auditory stimuli associated with a model piglet. Applied Animal Behaviour
Science 32, 129-137.
Jarvis, S., Calvert, S.K., Stevenson, J., VanLeeuwen, N., Lawrence, A.B., 2002. Pituitary-
adrenal activation in pre-parturient pigs (Sus scrofa) is associated with behavioural
restriction due to lack of space rather than nesting substrate. Animal Welfare 11, 371-
384.
Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs.
Applied Animal Behaviour Science 16, 131-142.
Karlen, G.A.M., Hemsworth, P.H., Gonyou, H.W., Fabrega, E., Strom, A.D., Smits, R.J.,
2007. The welfare of gestating sows in conventional stalls and large groups on deep
litter. Applied Animal Behaviour Science 105, 87-101.
Kilbride, A.L., Mendl, M., Statham, P., Held, S., Harris, M., Cooper, S., Green, L.E., 2012. A
cohort study of preweaning piglet mortality and farrowing accommodation on 112
commercial pig farms in England. Preventive Veterinary Medicine 104, 281-291.
Kutzer, T., Bünger, B., Kjaer, J.B., Schrader, L., 2009. Effects of early contact between non-
littermate piglets and of the complexity of farrowing conditions on social behaviour
and weight gain. Applied Animal Behaviour Science 121, 16-24.
Lawrence, A.B., Petherick, J.C., McLean, K.A., Deans, L.A., Chirnside, J., Gaughan, A.,
Clutton, E., Terlouw, E.M.C., 1994. The effect of environment on behaviour, plasma
cortisol and prolactin in parturient sows. Applied Animal Behaviour Science 39, 313-
330.
110
Lensink, B.J., Leruste, H., Le Roux, T., Bizeray-Filoche, D., 2009. Relationship between the
behaviour of sows at 6 months old and the behaviour and performance at farrowing.
animal 3, 128-134.
Marchant, J.N., Broom, D.M., Corning, S., 2001. The influence of sow behaviour on piglet
mortality due to crushing in an open farrowing system. Animal science 72, 19-28.
Oliviero, C., Heinonen, M., Valros, A., Hälli, O., Peltoniemi, O.A.T., 2008. Effect of the
environment on the physiology of the sow during late pregnancy, farrowing and early
lactation. Animal reproduction science 105, 365-377.
Pajor, E.A., Kramer, D.L., Fraser, D., 2000. Regulation of contact with offspring by domestic
sows: temporal patterns and individual variation. Ethology 106, 37-51.
Pedersen, L.J., Berg, P., Jørgensen, G., Andersen, I.L., 2011. Neonatal piglet traits of
importance for survival in crates and indoor pens. J. Anim. Sci. 89, 1207-1218.
Pitts, A.D., Weary, D.M., Fraser, D., Pajor, E.A., Kramer, D.L., 2002. Alternative housing for
sows and litters.: Part 5. Individual differences in the maternal behaviour of sows.
Applied Animal Behaviour Science 76, 291-306.
Pitts, A.D., Weary, D.M., Pajor, E.A., Fraser, D., 2000. Mixing at young ages reduces
fighting in unacquainted domestic pigs. Applied Animal Behaviour Science 68, 191-
197.
Podgórski, T., Lusseau, D., Scandura, M., Sönnichsen, L., Jędrzejewska, B., 2014. Long-
lasting, kin-directed female interactions in a spatially structured wild boar social
network. PloS one 9, e99875.
Roehe, R., Kalm, E., 2000. Estimation of genetic and environmental risk factors associated
with pre-weaning mortality in piglets using generalized linear mixed models. Animal
science 70, 227-240.
Roehe, R., Shrestha, N.P., Mekkawy, W., Baxter, E.M., Knap, P.W., Smurthwaite, K.M.,
Jarvis, S., Lawrence, A.B., Edwards, S.A., 2009. Genetic analyses of piglet survival
and individual birth weight on first generation data of a selection experiment for piglet
survival under outdoor conditions. Livestock Science 121, 173-181.
Schmid, H., 1991. Natürliche Verhaltenssicherungen der Hausschweine (Sus scrofa) gegen
das Erdrücken der Ferkel durch die Muttersau und die Auswirkungen
haltungsbedingter Störungen (Doctoral dissertation).
Špinka, M., Illmann, G., de Jonge, F., Andersson, M., Schuurman, T., Jensen, P., 2000.
Dimensions of maternal behaviour characteristics in domestic and wildxdomestic
crossbred sows. Applied Animal Behaviour Science 70, 99-114.
111
Stolba, A., Wood-Gush, D.G.M., 1989. The behaviour of pigs in a semi-natural environment.
Animal Production 48, 419-425.
van Nieuwamerongen, S.E., Soede, N.M., van der Peet-Schwering, C.M.C., Kemp, B.,
Bolhuis, J.E., 2015. Development of piglets raised in a new multi-litter housing
system vs. conventional single-litter housing until 9 weeks of age. Journal of animal
science 93, 5442-5454.
Vasdal, G., Glærum, M., Melišová, M., Bøe, K.E., Broom, D.M., Andersen, I.L., 2010.
Increasing the piglet's use of the creep area-A battle against biology? Applied Animal
Behaviour Science 125, 96-102.
Waran, N.K., Broom, D.M., 1993. The influence of a barrier on the behaviour and growth of
early-weaned piglets. Animal science 56, 115-119.
Weary, D.M., Pajor, E.A., Fraser, D., Honkanen, A.-M., 1996. Sow body movements that
crush piglets: a comparison between two types of farrowing accommodation. Applied
Animal Behaviour Science 49, 149-158.
Weary, D.M., Phillips, P.A., Pajor, E.A., Fraser, D., Thompson, B.K., 1998. Crushing of
piglets by sows: effects of litter features, pen features and sow behaviour. Applied
Animal Behaviour Science 61, 103-111.
Weber, R., 2000. Alternative housing systems for farrowing and lactating sows.
PUBLICATION-EUROPEAN ASSOCIATION FOR ANIMAL PRODUCTION 102,
109-116.
Weber, R., Keil, N.M., Fehr, M., Horat, R., 2007. Piglet mortality on farms using farrowing
systems with or without crates. Animal Welfare 16, 277-279.
Weber, R., Keil, N.M., Fehr, M., Horat, R., 2009. Factors affecting piglet mortality in loose
farrowing systems on commercial farms. Livestock Science 124, 216-222.
Wechsler, B., Hegglin, D., 1997. Individual differences in the behaviour of sows at the nest-
site and the crushing of piglets. Applied Animal Behaviour Science 51, 39-49.
Wolf, J., Žáková, E., Groeneveld, E., 2008. Within-litter variation of birth weight in
hyperprolific Czech Large White sows and its relation to litter size traits, stillborn
piglets and losses until weaning. Livestock Science 115, 195-205.
112
113
GENERAL SUMMARY
The aim of this thesis was to investigate the performances of lactating sows and their piglets
in group-housing systems with different housing designs with the focus especially on
maternal behaviour, health indicators and reproductive traits. Different GH systems were
evaluated with regard to their feasibility for pig husbandry.
In Chapter One, a group-housing system was tested in which six sows were fixed in crates
three days ante partum until one day post partum. For the remaining amount of time they were
able to choose between their home pen and a shared running area. Five days post partum, the
running area was opened for the piglets. The group-housed sows (n=23) were compared with
single-housed (n=24) sows which were housed individually with permanent fixation in crates.
In four batches the reproductive traits of the sows were documented. Group-housed sows had
significantly fewer total piglet losses and crushed fewer piglets compared to single-housed
sows. Furthermore, maternal behaviour of sows was observed in six successive tests in the
second week and fourth week of lactation. The sows’ reaction to piglet distress calls, and
separation from and reunion with their piglets was tested both in their home pens and in a test
arena. Group-housed sows showed stronger maternal reactions to piglet distress calls and
reunion with their piglets when tested in their home pens, whereas single-housed sows reacted
to piglet distress calls more strongly in the test arena. To conclude, the housing system had an
influence on maternal behaviour. Further research is needed to obtain more information on
whether the significantly fewer piglet losses of group-housed sows are related to the stronger
maternal reactions.
Based on the findings in Chapter One on the differences between sows regarding their
maternal behaviour, Chapter Two examined the behavioural parameters observed in the
maternal tests by means of a factor analysis to identify redundancies in behavioural
parameters. Five factors were extracted, explaining together 89 % of the total variance in the
data. The interpretation of the factor loadings revealed four underlying maternal factors:
communication, care, contact and local attachment to the piglets. Communication, thus
vocalisation, was extracted by the factor analysis as the most important maternal factor and is
used in sows to call their piglets in threatening situations, to contact their piglets using
nose-to-nose contacts, to call the piglets for nursing and to synchronise nursing with other
sows. Summarising the findings, it can be said that the experimental environment has an
influence on the significance of the maternal tests. Moreover, the results showed that maternal
behaviour is multidimensional, thus several tests are needed to cover all maternal
114
characteristics. For example, the reunion test in the home pen has been found to be useful to
evaluate the care motivation of the sow to suckle their piglets after reunion, which is essential
for piglet survival. Whereas, the separation test in the test arena investigates the attachment of
the sows to their piglets.
In Chapter Three, two different sizes of group-housing systems were investigated. Ten sows
were housed with their litters in free-farrowing pens from three days ante partum until six
days post partum. The sows and their piglets then had the opportunity to leave the pens and to
mix with other sows and piglets. The group-housed sows were compared with single-housed
sows housed individually by permanent fixation in crates. The reproductive traits of four
batches of all sows (group-housing systembig: n=40; group-housing systemsmall: n=40;
single-housing system: n=63) were documented during lactation. Sows from both
group-housing systems had significantly higher total piglet losses and crushed more piglets
compared to single-housed sows. Moreover, significantly more live-born piglets died in
smaller free-farrowing systems compared to larger free-farrowing systems. To gain more
insights into critical situations for piglets crushed in free-farrowing pens, the lying down and
rolling behaviour of group-housed sows with few crushed piglets (n=10) were compared with
group-housed sows with many crushed piglets (n=10) in the first 72 hours after giving birth.
Equal numbers of piglets were crushed during the lying down and rolling movements within
the two groups. However, the group-housed sows with many crushed piglets performed
significantly more lying down movements without using the pen walls, which is more
dangerous for piglets to be crushed and rolled significantly more frequently. Especially rolling
movements from one side to the other were highly positively correlated with the number of
crushed piglets during the first 72 hours post partum. In conclusion, the safety of the piglets
was reduced in the free-farrowing pens of the group-housing systems related to the higher
pre-weaning mortality. In addition, more piglets died in smaller free-farrowing pens. The
detailed observation of the group-housed sows detected a high variation in the maternal
behaviour. Sows with many crushed piglets performed more lying down movements without
using the pen walls and rolled more frequently.
The sows in the experiment in Chapter Four were housed in the group-housing systems as
described in Chapter Three. Four batches of data were collected from all sows and their
piglets with regard to different health indicators. After four weeks of lactation, group-housed
sows had significantly fewer skin lesions of the udder and the tail, however, were dirtier and
had more skin lesions of the body compared to single-housed sows. Moreover, group-housed
115
piglets had significantly fewer skin lesions of the face in the first week of lactation and did not
differ with regard to their incidence of skin lesions of the face in the fourth week of lactation
compared to single-housed piglets. Over the four batches, more skin lesions of the body were
documented for group-housed piglets. To conclude, no alarming health problems were
documented for the sows or their piglets. Multi-suckling resulted in reduced incidence of skin
lesions of the udder of the sows and in no increase in skin lesions of the face of the piglets.
More skin lesions o the body of the sows and piglets of the group-housing system were
caused by social interactions. Moreover, saliva cortisol samples were collected from all sows
two weeks ante partum before they moved to the farrowing housing systems and in the second
and fourth week of lactation. During lactation, group-housed sows had significantly higher
saliva cortisol levels compared to single-housed sows. Furthermore, higher cortisol levels
during lactation were detected in the group-housed sows. A possible explanation for this
could be social stress and more physical activity. Thus, the sows in the group-housing system
were subject to more stress compared to the sows in the single-housing system.
116
117
ZUSAMMENFASSUNG
In der vorliegenden Thesis wurden verschiedene Gruppenhaltungssysteme für die Haltung
ferkelführender Sauen untersucht. Ein besonderer Fokus lag dabei auf dem mütterlichen
Verhalten, auf Gesundheitsindikatoren und der Leistung der Sauen.
Im ersten Kapitel wurde eine Gruppenhaltung für laktierende Sauen (n=23) getestet, in der
sechs Sauen zusammen gehalten wurden. Diese Sauen wurden mit einzelgehaltenen Sauen
(n=24) mit permanenter Fixierung im Ferkelschutzkorb verglichen. Die Einzelbuchten der
gruppengehaltenen Sauen waren mit einem Ferkelschutzkorb ausgestattet, in dem die Sauen
jeweils drei Tage ante partum bis einen Tag post partum fixiert wurden. Die restliche Zeit
konnten die Sauen zwischen ihrer Bucht oder einem gemeinsamen Freilaufbereich frei
wählen. Am fünften Tag post partum wurde der Freilaufbereich ebenfalls für die Ferkel
geöffnet. In vier Durchgängen wurden alle Sauen hinsichtlich ihrer Leistung und ihrem
mütterlichen Verhalten untersucht. Signifikant weniger Ferkelverluste wurden für
gruppengehaltene Sauen dokumentiert, verglichen mit den Sauen im Ferkelschutzkorb. Des
Weiteren wurden alle Sauen in der zweiten und vierten Woche der Laktation in sechs
aufeinanderfolgenden Verhaltenstests zur Mütterlichkeit getestet, die in der jeweiligen Bucht
der Sau oder in einer Testarena durchgeführt wurden. Die Reaktionen der Sauen auf die
Schreie der Ferkel wurden genauso beurteilt, wie die Separation von und die
Wiedervereinigung mit ihren Ferkeln. Für die Sauen der Gruppenhaltung wurden stärkere
mütterliche Reaktionen im Ferkelschreitest und im Wiedervereinigungstest in der Bucht
beobachtet, hingegen reagierten Sauen im Ferkelschutzkorb stärker im Ferkelschreitest in der
Testarena. Die Studie zeigte, dass die Haltungsbedingungen die mütterlichen Reaktionen der
Sauen veränderten. Die signifikant geringeren Ferkelverluste in Kombination mit den
stärkeren Reaktionen der gruppengehaltenen Sauen könnten auf verbesserte
Muttereigenschaften hindeuten.
Aufgrund der unterschiedlichen mütterlichen Reaktionen der Sauen und ihren individuellen
Unterschieden innerhalb der Testreihen im ersten Kapitel, wurden die Verhaltensdaten im
zweiten Kapitel mit einer Faktoranalyse analysiert, welche fünf Faktoren extrahierte, die
89 % der gesamten Datenvarianz erklärten. Die Interpretation der Faktoren ergab vier
Dimensionen mütterlichen Verhaltens: Kommunikation, Kontaktaufnahme, Fürsorge und
lokale Verbundenheit mit den Ferkeln. Kommunikation ist das wichtigste Instrument für
Sauen, um mit ihren Ferkeln in bedrohlichen Situationen in Verbindung zu bleiben. Zudem
stellt die Kommunikation ein wichtiges Element für die Sau bei der Kontaktaufnahme mit
118
ihren Ferkeln und beim Säugen als Rufsignal der Ferkel und als Signal zur Synchronisation
mit anderen Sauen dar. Die Anzahl der Faktoren zeigte, dass Mütterlichkeit von Sauen
multidimensionale Eigenschaften besitzt und von der Testumgebung beeinflusst wird. Eine
Reihe von Tests und die richtige Auswahl an Testkombinationen je nach gewünschter
Aussage sind zur validen Evaluierung maternaler Eigenschaften notwendig. Zum Beispiel
eignet sich der Wiedervereinigungstest in der Bucht um die Motivation der Sau zu testen ihre
Ferkel zu säugen, was für die Ferkel überlebenswichtig sein kann. Hingegen überprüft der
Separationstest in der Testarena die Verbundenheit der Sau mit ihren Ferkeln.
Im dritten Kapitel wurden weitere Varianten der Gruppenhaltung untersucht, die sich in der
Größe der freien Abferkelbuchten und dem gemeinsamen Freilaufbereich unterschieden. Die
gruppengehaltenen Sauen (Gruppenhaltunggroß: n=40; Gruppenhaltungklein: n=40) wurden mit
im Ferkelschutzkorb gehaltenen Sauen (n=63) verglichen. In den Gruppenhaltungen wurden
die Sauen drei Tage ante partum bis sechs Tage post partum in freien Abferkelbuchten
gehalten. Davor und danach hatten die Sauen und Ferkel die Möglichkeit, einen gemeinsamen
Freilaufbereich zu nutzen. In vier Durchgängen wurden die Leistungsdaten der Sauen
dokumentiert. Sauen beider Gruppenhaltungen hatten signifikant höhere Gesamtferkelverluste
und eine signifikant höhere Anzahl an erdrückten Ferkeln, verglichen mit den im
Ferkelschutzkorb gehaltenen Sauen. Des Weiteren waren die Gesamtferkelverluste signifikant
höher für die Sauen in der kleineren Gruppenhaltung verglichen mit den Sauen in der
größeren Gruppenhaltung. Darüber hinaus wurden gruppengehaltene Sauen mit wenig
erdrückten Ferkeln (n=10) mit gruppengehaltenen Sauen mit vielen erdrückten Ferkel (n=10)
in den ersten 72 Stunden post partum beobachtet, hinsichtlich gefährlicher Situationen für
Ferkel erdrückt zu werden. Das Abliege- und Rollverhalten der Sauen sowie das Verhalten
der Ferkel wurde untersucht. Die Beobachtungen ergaben, dass die Ferkel innerhalb der
jeweiligen Gruppe das gleiche Risiko hatten entweder beim Abliegen oder beim Rollen der
Sau erdrückt zu werden. Dennoch nutzten Sauen mit vielen erdrückten Ferkeln signifikant
weniger die Buchtenwände beim Abliegen und rollten signifikant häufiger, verglichen mit
Sauen mit wenig erdrückten Ferkeln. Das Rollen von einer Seite auf die andere Seite
korrelierte stark mit der Anzahl der erdrücken Ferkel in den ersten 72 Stunden post partum.
Diese Studie verdeutlicht, dass die Bewegungsfreiheit der Sauen in den freien
Abferkelbuchten ein höheres Risiko für die Ferkel darstellt, erdrückt zu werden. Zudem
wurde beobachtet, dass die kleinen, freien Abferkelbuchten die Ferkelverluste erhöhten. Des
Weiteren nutzen Sauen mit vielen erdrückten Ferkeln weniger die Buchtenwände beim
Abliegen und rollten häufiger.
119
Die Sauen und Ferkel im vierten Kapitel beziehen sich auf die Datenerhebung, welche im
dritten Kapitel beschrieben wurde. Verschiedene Gesundheitsindikatoren wurden für die
Sauen und ihre Ferkel dokumentiert. Nach vierwöchiger Laktation wiesen gruppengehaltene
Sauen signifikant weniger Hautläsionen am Gesäuge und am Schwanz auf, hingegen waren
diese Sauen signifikant verschmutzter und hatten mehr Hautläsionen am Körper, verglichen
mit den Sauen im Ferkelschutzkorb. Die Ferkel, die in der Gruppenhaltung aufwuchsen,
hatten signifikant weniger Hautläsionen am Maul in der ersten Laktationswoche und gleich
viele Hautläsionen am Maul in der vierten Laktationswoche verglichen mit Ferkeln, die in
Einzelbuchten mit fixierter Sau aufwuchsen. Zudem wurden signifikant mehr Hautläsionen
am Körper der gruppengehaltenen Ferkel vor dem Absetzen beobachtet. Daraus lässt sich
schließen, dass keine alarmierenden Gesundheitsprobleme für Sauen und Ferkel beider
Haltungssysteme festgestellt werden konnten. Zudem resultierte das gemeinsame Säugen in
der Gruppenhaltung in verringerten Hautläsionen am Gesäuge der Sauen und in keinem
Anstieg der Hautläsionen am Maul der Ferkel. Die Hautläsionen am Körper der
gruppengehaltenen Sauen und Ferkel sind mit Interaktionen zwischen den Tieren zu erklären.
Darüber hinaus wurden zwei Wochen ante partum und zwei und vier Wochen post partum
von allen Sauen Speichelproben zur Bestimmung des Kortisolgehaltes genommen. Während
der Laktation wiesen Sauen der Gruppenhaltung signifikant höhere Kortisolgehalte auf, als
die im Ferkelschutzkorb gehaltenen Sauen. Während der Laktation wurden höhere
Kortisolgehalte bei den gruppengehaltenen Sauen festgestellt, welche durch sozialen Stress
und erhöhte physische Aktivität bedingt sein könnten. Demnach waren die Sauen in der
Gruppenhaltung gestresster während der Laktation, verglichen mit den Sauen im
Ferkelschutzkorb.
DANKSAGUNG
An dieser Stelle möchte ich mich bei all denjenigen bedanken, die zur Umsetzung und zum
Gelingen dieser Arbeit beigetragen haben.
Mein aufrichtiger Dank gilt meinem Doktorvater Herrn Prof. Dr. Joachim Krieter für die
Überlassung des interessanten Themas, die gute wissenschaftliche Betreuung und die
gewährten Freiräume während der Erstellung der Arbeit. Zudem möchte ich mich für die
Möglichkeiten bedanken meine Ergebnisse auf Tagungen im In- und Ausland präsentieren zu
können.
Bei Frau Prof. Dr. Imke Traulsen möchte ich mich ganz herzlich für die Übernahme des
Koreferates bedanken.
Mein besonderer Dank gilt Frau Dr. Kathrin Büttner und Frau Dr. Irena Czycholl für die
motivierende Betreuung und die ständige Bereitschaft zum Korrekturlesen.
Die Förderung erfolgte dankenswerter Weise aus Mitteln des Zweckvermögens des Bundes
bei der Landwirtschaftlichen Rentenbank, bei denen ich mich ganz herzlich bedanken möchte.
Dem Lehr- und Versuchszentrum Futterkamp danke ich für die kreative Zusammenarbeit und
die Hilfestellungen während der Projektarbeit. Mein Dank gilt hier im Besonderen Herrn
Harm Kruse, Herrn Christian Meyer und Herrn Dr. Onno Burfeind.
Für die schöne Zeit am Institut mit freundschaftlichem Arbeitsklima, Ausflügen,
Unterstützung, Hilfsbereitschaft und den vielen lieben und lustigen Gesprächen möchte ich
mich bei allen Kolleginnen und Kollegen bedanken, besonders bei den jetzigen und
ehemaligen ‚Containerbewohnern‘.
Ein großes Dankeschön geht auch an meine lieben Freunde außerhalb des Institutes für die
laufende moralische Unterstützung während der gesamten Zeit.
Mein größter Dank gilt meinem Ehemann, meinen Eltern und meiner Schwester. Ihr habt
mich immer uneingeschränkt unterstützt und an mich geglaubt. Ohne euch wäre ich heute
nicht dort wo ich bin.
LEBENSLAUF
Name Charlotte Gertud Elisabeth Grimberg-Henrici (geb. Grimberg)
Geburtsdatum 2. August 1986
Geburtsort Köln
Staatangehörigkeit Deutsch
Familienstand Verheiratet
Schulische und akademische Bildung
___________________________________________________________________________
2011-2013 M.Sc. Tierwissensschaften (Animal Sciences)
Spezialisierung: Gesundheit und Verhalten (Nutztier)
Wageningen University, Niederlande
2007-2011 B.Sc. Tiermanagement
Spezialisierung: Pferd und Management
Internationale Hochschule Van Hall Larenstein, Niederlande
1997-2006 Abitur
Liebfrauenschule-Staatlich genehmigtes Gymnasium des Erzbistum,
Köln
Praktika
___________________________________________________________________________
02/2010-01/2011 Delft University of Technology, Niederlande
‚Dog Cognition‘ Department Man-Machine Interaction Group
05/2009-07/2009 Practical Horsemanship, England
01/2009-04/2009 Pferdeausbildung und -zucht, Deutschland
Berufliche Tätigkeit
__________________________________________________________________________
Seit März 2014 Wissenschaftliche Mitarbeiterin am Institut für Tierzucht und
Tierhaltung der Christian-Albrechts-Universität zu Kiel bei
Herrn Prof. Dr. Krieter