Dimensions of maternal behaviour characteristics indomestic and wild�domestic crossbred sows

Marek SÏpinkaa,*, Gudrun Illmanna, Francien de Jongeb,Maria Anderssonc, Teun Schuurmand, Per Jensenc

aGroup of Ethology, Research Institute of Animal Production, CZ-104 01 Prague-UhrÏõÂneÏves, Czech RepublicbDepartment of Ecological Agriculture and Society, Wageningen University and Research Centre,

Haarweg 333, 6709 RZ Wageningen, NetherlandscDepartment of Animal Environment and Health, Faculty of Veterinary Medicine, Section of Ethology,

Swedish University of Agricultural Sciences, POB 234, S-532 23 Skara, SwedendHuman and Animal Physiology Group, Wageningen Institute of Animal Sciences, Wageningen University,

Haarweg 10, 6709 PJ Wageningen, Netherlands

Accepted 29 May 2000

Abstract

We examined the maternal behaviour of seven domestic and seven wild�domestic primiparous

sows during 10 days post partum to investigate two questions: (1) Did maternal behaviour change

during domestication? (2) Can the interindividual variability of maternal behaviour be subsumed

into a few dimensions of maternal temperament? We recorded: (a) willingness to leave the nest for

food on Day 2; (b) reaction to a playback of squeezed piglet distress vocalisation on Day 2; (c)

spontaneous nursing behaviour and spontaneous lying-down behaviour on Day 5 (from an overnight

video recording); (d) reactions to playbacks of various piglet distress vocalisations on Day 6 and (e)

reactions to a human in the `nest' with piglets on Day 9. Moreover, data on baseline cortisol saliva

concentration and its increase during a brief transportation period and novel environment challenge

at the age of 5 months were available. Crossbred sows did not differ from domestic ones in any

aspect of maternal behaviour except for a higher tendency to terminate ®nal massage during

nursings and a higher frequency of changing posture from lying to standing and back during the

night. Factor analysis (based on correlation matrix of 11 behaviour and cortisol variables calculated

for all 14 sows after removing the effect of breed) indicated that 82% of the variability in the data

could be explained by three factors: ®rst, `calmness' on which low night time frequency of major

posture changes, carefulness of lying-down behaviour and high propensity to remain in nursing

position after milk ejection loaded positively while cortisol concentrations during challenge loaded

negatively; second, `protectiveness' with high loadings of the reaction scores to the playbacks of

piglet distress calls and the human presence near the piglets; and third, `nursing activity' which was

Applied Animal Behaviour Science 70 (2000) 99±114

* Corresponding author. Tel.: �420-2-6771-0713; fax: �420-2-6771-0779.

E-mail address: spinka@vuzv.cz (M. SÏpinka).

0168-1591/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.

PII: S 0 1 6 8 - 1 5 9 1 ( 0 0 ) 0 0 1 5 1 - 9

strongly positively associated with nursing frequency, and negatively with the proportion of

nutritive nursings and baseline cortisol values. The results indicate that most aspects of pig maternal

behaviour have not been signi®cantly changed by domestication and that substantial variability in

maternal behaviour exists between sows, perhaps in the form of several behaviour characteristics

which encompass both behaviour and endocrine pro®les of the sows. # 2000 Elsevier Science B.V.

All rights reserved.

Keywords: Maternal behaviour; Pigs; Domestication; Nursing; Cortisol; Vocalisation; Temperament

1. Introduction

Maternal behaviour is important for pig husbandry since it partly determines mortality

and growth of the piglets. About 15% of the live born piglets die before weaning (McKay,

1993; Blackshaw et al., 1994), and among them 70±80% is attributable to the sow

behaviour, directly or indirectly. During the ®rst day of life, piglets get trapped under

the sow while she is lying-down or rolling from her belly to her side (Dyck and Swiestra,

1987; Weary et al., 1996a; Herskin et al., 1998). Individual dams differ in the frequency of

postural changes, and in how often they change posture in a way that is dangerous for the

piglets (i.e. ¯opping straight down to one side versus descending to lying posture in a

vertical plane; lying down when piglets are nearby). Even after a piglet is trapped, the sow's

behaviour is important. Just 20% of piglets trapped under the sow's body are actually

crushed. If a trapped piglet starts screaming, the sow may stand up immediately and release

it. In cases when the sow stood up within 1 min, only 5% of piglets died; in contrast, 67%

of the piglets trapped for longer than 4 min were lost (Weary et al., 1996a). Two aspects of

the sow's behaviour, thus, affect the probability of crushing: ®rst, the sow's general

restlessness and her lying-down behaviour; second, the sow's individual reactivity to a

screaming piglet (Wechsler and Hegglin, 1997).

Another problem in piglet production is low milk intake in some piglets or in whole

litters which results in low or too variable weaning weights. This may even lead to piglet

losses, either directly due to starvation or indirectly due to the weak piglets being crushed

by the sow (Hutson et al., 1993; Weary et al., 1996b). Low frequency of nursings (SÏpinka

et al., 1997; SÏpinka et al., 1999), and a high proportion of non-nutritive nursings (nursings

without milk ejection, CastreÂn et al., 1989; Illmann and Madlafousek, 1995; Illmann et al.,

1999) are probably the aspects of nursing behaviour which can most signi®cantly reduce

milk intake. A tendency to terminate the udder massage by the piglets soon after milk

ejection might also suppress future milk production in the glands, although the evidence is

equivocal (Algers and Jensen, 1991; SÏpinka and Algers, 1995; Jensen et al., 1998; Illmann

et al., 1998).

These undesired aspects in the maternal behaviour of domestic sows may have arisen

during the process of domestication, especially during the intense selective breeding

regimes prevalent in last few decades. Information about whether this was the case can be

useful in judging the likelihood that maternal characteristics of sows could be improved

through selective breeding programmes. The present study was conducted in order to

assess whether sows carrying different proportions of genes from domestic and wild boar

100 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

stock vary in biologically signi®cant aspects of maternal behaviour. In order to answer this

question, we collected data on maternal behaviour in seven domestic and seven crossbred

wild�domestic primiparous sows. During the data analysis, we found that the breeding

effect could explain only a small proportion of the variation in the data. Therefore,

we investigated the interindividual variation through a post-hoc correlational and factor

analysis. We focused on questions of how various aspects of maternal behaviour

relate to each other and whether they could be subsumed into a few `maternal behaviour

traits'.

2. Methods

2.1. Animals

We used 14 primiparous sows which were born to Yorkshire�Dutch Landrace mothers.

Seven of them were sired by various Yorkshire fathers (`domestic' sows) and seven by one

male wild boar (`crossbred' sows). The sows were born and reared in the outdoor housing

system described below. At the age of 20±21 months, all experimental sows were

inseminated with Great Yorkshire sperm. Hence, the offspring interacting with the

experimental mothers in the current study were either 100% domestic or 25% wild boar.

2.2. Housing conditions

The experiment was carried out at the Wageningen Agricultural University, The

Netherlands in April±June 1997. All sows were loose housed during pregnancy. Two

days before the scheduled farrowing induction, the sows were locked in half-covered

concrete-¯oored rectangular outdoor pens (2.2 m�4.5 m). About 5 kg of straw was

provided in the covered half of the farrowing pen, the open half was provided with

drinking nipples for both the sows and the piglets and also served as dunging area. At the

back of the farrowing pens, half-open doors allowed behavioural observations, without

disturbances to sow or litter. At 8 a.m. and 1:30 p.m., sows were allowed to leave the pens in

order to feed in separate feeding crates positioned about 20 m from the pens. Sows which

were to be tested in the same batch (see below) were not housed in adjacent pens in order to

reduce the possibility of habituation to acoustic stimuli in the playbacks tests.

2.3. Procedure

2.3.1. Farrowing and cross-fostering

In order to enable early and balanced cross-fostering between litters, farrowings were

induced on about Day 112 of pregnancy by a prostaglandin analogue (PGF2a) injection.

The induction was scheduled in four batches of four, four, four and two sows. All sows in a

batch farrowed within a time period of 24 h (Day 0). Within 36 h after parturition

(Day 0 or 1), all piglets were weighed, ear tagged and cross-fostered within each batch.

The newly composed litters were balanced with respect to litter size, number of piglets

of each genotype (pure domestic or 25% wild boar), average piglet weight and sex

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 101

ratio (in that order of priority). The resulting litter sizes varied between seven and

nine piglets among which two±four piglets were the sow's own progeny while the rest

originated from the other litters in the batch. Mortality of the piglets was assessed by

noting any losses occurring between cross-fostering and video recording 4 days later (see

below).

2.3.2. Nest-leaving and playback experiment on Day 2

On Day 2 after parturition (Day 2), each sow was tested for her willingness to leave the

nest and the young for food and also for her reactivity to squeezed piglet distress

vocalisation. At the time of the morning feeding, the non-tested sows were ®rst allowed

to feed in a normal way. After they were locked back again in their farrowing pens, the pen

of the ®rst sow to be tested was opened and the latency of her emergence from the

farrowing pen was noted. If she did not leave the pen within 5 min, the gate was closed and

latency was set equal to 300 s. We then waited for the moment when the sow was lying

down in the farrowing pen for the ®rst time. When her body touched the ground, a 30 s

recording of piglet distress vocalisation was played (at about natural volume) from a

loudspeaker through an expanded-metal-covered opening in the outer wall of the farrowing

pen. The recording was the vocalisation of a 2-days-old piglet, unrelated to any animal in

the experiment. During the recording, vocalisations were stimulated by restraining the

piglet on the ground by an experimenter using both hands. The reaction of the sow to the

playback was video recorded and later scored using the following scale: 0 Ð no reaction,

1 Ð head movement, 2 Ð body movement, 3 Ð sits, 4 Ð stands up, 5 Ð makes contact

with the expanded metal covering the loudspeaker, 6 Ð bites or paws at the expanded

metal. Two extra points were added if the sow stood up within 5 s since the beginning of the

playback or one point if she did so within 30 s. After the testing of the ®rst sow was

®nished, we continued with testing of the next sow until all sows from a batch (up to four)

were tested. The same vocal stimulus was used for all sows.

2.3.3. Undisturbed nursing behaviour

During the night between Days 4 and 5 (or between Days 5 and 6) post partum, each sow

was video recorded either between 18:00 and 24:00 or between 00:00 and 06:00 h while

there were no people present near the animal facilities. The videotapes were later analysed

to quantify the following parameters of the undisturbed nursing behaviour: intervals

between nursings, proportion non-nutritive nursings, duration of pre-ejection and post-

ejection udder massage, and proportion of nursings terminated by the sow. Non-nutritive

nursings (nursings without milk ejection) were identi®ed by the absence of synchronised

rapid sucking activity of the litter (Illmann and Madlafousek, 1995). A nursing was

classi®ed as terminated by the sow if she rolled over onto her belly or stood up while some

of the piglets were still active at the udder.

2.3.4. Undisturbed lying-down behaviour

The videotaped six night time hours of undisturbed behaviour were also used to assess

the frequency and carefulness of the sows' lying-down behaviour. Each occurrence of the

sow's descent from a standing to a lying posture was counted and given a carefulness score

as follows: one point for snif®ng the ¯oor before lying-down, one point for rooting, one

102 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

point if the sow descended in a vertical plane (as opposed to ¯opping over to one side),

one point if there were, at that moment, no piglets near her on the side on which she lay

down.

2.3.5. Playback experiment on Day 5 or 6

These tests were conducted on Day 5 or 6 post partum either in the morning or in the

afternoon. The behaviour of the sow was videotaped in her pen while she was subjected to

three different tests of her reactivity to piglet distress vocalisations. The tests were:

Trapped piglet call ± A 35 s playback of a piglet's distress vocalisation which had been

recorded as described in the section on playback experiment on Day 2. However, different

recordings were used from those on Day 2. The playback was started while the sow was

descending to a lying posture at the moment her body touched the ground.

Isolated piglet call ± A 60 s playback of a piglet's grunting and squeaking which had been

recorded when a 7-day old piglet had been isolated from her mother and littermates. The

playback was started at any moment between 20 and 35 min after the last milk ejection

when both the sow and the litter was resting.

Teat fighting calls ± A 50 s playback of loud vocalisations recorded while two piglets had

been fighting for a teat during nursing. The playback was started during a nursing about

15 s after the pre-ejection udder massage by the piglets had begun.

Each of the three tests was repeated twice using vocalisations from different piglets to

those used for the playback stimuli. Hence, each sow was presented with a total of six

playback stimuli during the experimental session. The playbacks were always separated by

at least 15 min. The order of the tests was variable in individual sows as it was dependent on

the spontaneous nursing and resting behaviour of the animals. The behavioural reactions of

the sow were scored using the same scale as in the playback experiment on Day 2. A

preliminary analysis showed that the scores from the three types of tests were highly

intercorrelated and displayed similar results in the domestic-crossbred comparison. There-

fore, the score reactions from the six tests (reactions to trapped-piglet, isolated-piglet and

teat-®ghting calls, each tested twice) were combined into a ®nal score.

2.3.6. The human-in-the-nest test

On Day 8 or 9 post partum, the response of the sow to a human being present near her

piglets was tested. First, the non-tested sows were allowed to feed in a normal way. After

they were locked back again in their farrowing pens, the ®rst sow to be tested was allowed

to leave the pen, enter the feeding crate and was locked there for 10 min while she was

feeding. In the meanwhile, an experimenter entered the pen with her piglets, closed the bar

gate behind him/her and stood quietly about 1 m behind the bars of the pen gate. The sow

was released from the feeding pen and once she approached her pen, her behaviour was

scored. Every 20 s for 5 min, we recorded her location, using instantaneous sampling, as

follows: nose contact with the bar gate Ð 2 points, within 1 m in front of the gate Ð 1

point, other place Ð 0 points. We also scored the average frequency of her calls during each

20 s period, using the following scale: frequent vocalisation (approx. >1 call/s) Ð 2 points,

moderate vocalisation (<1 call/s) Ð 1 point, not vocalising Ð 0 points. Third, a additional

point was given to the sow within the current 20 s interval if she bit the bars of the gate or

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 103

jumped on it with her front legs. The average number of points for a 20 s interval was taken

as a measure of the sow's reaction to the human presence near her young.

2.4. Comparison with daytime nursing behaviour data

In order to see whether the individual characteristics of nursing behaviour were stable

over the diurnal period, we compared the nursing data with the data collected on the same

animals at 4±7 days post partum through direct daytime 4 h observations by Gustafsson

et al. (1999).

2.5. Cortisol data

In the post-hoc analysis (see Section 2.6), the set of behavioural scores was comple-

mented with data on cortisol baseline concentration and cortisol concentrations under a

complex challenge in the same sows. These data were collected about 15 months earlier,

when the animals were 5 months old. The gilts were housed in pairs in small pens and

trained to chew on cottonwool plugs for saliva collection. The ®rst saliva sample was

collected in the gilt's home pen just before the test started (t�0 min). At t�10 min, the gilt

was loaded to a carrier cage and, using a tractor, transported to another building. At the end

of transportation, 20 min after the ®rst sample, a second sample was taken and the gilt was

released to a 3 m�7 m pen with concrete ¯ooring. At 40 min, the third saliva sample was

taken and a gate was opened which connected this pen to a second pen provided richly with

straw. The fourth to sixth samples were collected at 55, 65 and 75 min after the ®rst one. All

tests were performed between 8:00 and 11:30 h in the morning in order to reduce cortisol

concentration differences caused by the circadian rhythm (Ekkel et al., 1996; Ruis et al.,

1997).

The saliva samples were immediately put on ice and centrifuged after the ®nal one was

collected. Plasma was stored atÿ208C until the assay. 0.5 ml of the thawn, undiluted saliva

sample was extracted with 2.5 ml dichloromethane. After evaporation of the solvent the

extract was subjected to a cortisol assay described by Janssens et al. (1994). The cortisol-

antibody used was a rabbit antiserum (K7348) raised against a cortisol-bovine-serum-

albumin-conjugate. Sensitivity of the assay was 0.5 ng/ml. The intra- and inter-assay

variances were 8 and 15%, respectively.

Two variables were extracted from the cortisol data. The ®rst was equalled to the pre-test

sample at time 0 and was labelled `cortisol baseline' concentration. The second was an

average of the second to sixth sample and was named `cortisol under challenge'

concentration.

2.6. Statistical analysis

2.6.1. Domestic�crossbred differences

Data on nest-leaving on Day 2 were treated as a binary variable (sow left the nest within

5 min or did not), and the domestic�crossbred comparison was made using the Fisher

exact probability test. In all other comparisons between the domestic and crossbred sows,

Mann±Whitney nonparametric U-tests were applied.

104 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

2.6.2. Night�daytime nursing behaviour comparison

In order to see whether the nursing behaviour was different over night compared to

daytime, and whether the sows kept their individual nursing style over the diurnal period,

we pooled the data from daytime direct observation (Gustafsson et al., 1999) and night

video recording. The data set was then subjected to an ANOVAwith breed, time (day versus

night), breed�time interaction, and sow nested within breed as the main effects. For the

breed effect, the between-pig MS was used as the error term.

2.6.3. Association of piglet mortality with maternal traits

The dependence of piglet mortality on maternal characteristics was assessed by

computing the Spearman rank correlation between the proportion of piglets dying from

Days 0 to 4 within a given `litter' (the group of piglets composed through cross fostering

and suckled by the given sow) and the scores of their foster mother in the tests and

observations. The breed effect was removed prior to computing the correlations.

2.6.4. Post-hoc correlational and factor analysis

Correlational post-hoc analysis was used to investigate whether the results from the three

tests, nursing and posture change behaviour scores from the night time video recording and

the cortisol data (11 variables in total) related to each other. A factor analysis was carried

out in an attempt to describe the complex results in terms of a few factors. In order to make

the results easier to interpret and present in a ®gure, we inverted some of the variables such

that the higher the level of the variable, the better (presumably) the survival and growth of

the piglets. Instead of proportion of nursings terminated by the sow, we took the proportion

of nursing not terminated by the sow and labelled it `massage allowance'; instead of

nursing interval, the nursing frequency (mean number of nursings per hour) was used; and

instead of proportion of non-nutritive nursings, the proportion of nutritive nursings was

used. The duration of pre-ejection udder massage was changed to `speed of milk ejection'

by converting the number of seconds into negative values. Similarly, the night time number

of standing-up acts was rendered negative and labelled `night time inactivity'. The effect of

breed (domestic versus crossbred) was removed from this modi®ed set of variables by a

GLM procedure with breed as the only main effect. Next, the residual values were used to

construct a matrix of product-moment correlations between the variables. Finally, the

matrix was subjected to factor analysis with the procedure Factor of the SAS software.

Squared multiple correlations were used for the prior communality estimates. Factors with

eigenvalues larger than 1 were retained and subjected to orthogonal varimax prerotation

followed by promax oblique rotation. This procedure was chosen in order to separate out

original variables on the resulting factors as well as possible, and, at the same time, show

whether the factors are correlated to each other or not.

3. Results

3.1. Crossbred�domestic differences

Crossbred sows were more likely to terminate the post-ejection massage during nursings

(57% versus 17% of cases, Mann±Whitney U-test, U�1, p<0.001) which resulted in

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 105

shorter post-ejection udder massage (268 versus 428 s, Mann±Whitney U-test, U�6,

p<0.05). Second, the crossbred sows tended to stand up, on average, three times as often

during the 6 h of night time video recording than the domestic ones (4.7 versus 1.7 times,

Mann±Whitney U-test, U�9, p<0.1). There were no other signi®cant differences between

the two breeds in any of the recorded aspects of maternal reactivity, posture changing

behaviour or nursing behaviour.

No difference was detected in the saliva unbound cortisol baseline concentrations at 5

months of age (Fig. 1: crossbred sows 2.2 ng/ml versus domestic sows 1.8 ng/ml, Mann±

Whitney U-test, U�15, NS). However, the `cortisol under challenge' concentration, i.e. the

average concentration in ®ve samples collected after the transport and during the new

environment challenge (Fig. 1) was more than twice higher in the crossbred sows (5.5

versus 2.7 ng/ml, Mann±Whitney U-test, U�4, p<0.01).

The mortality of piglets during the 4 days following cross-fostering was not different

between those suckled by the crossbred and domestic mothers (10% versus 13%, Mann±

Whitney U-test, U�20, NS; Table 1).

3.2. Comparison with daytime nursing behaviour data

Of the ®ve variables in the pooled day-and-night data set (duration of pre- and post-

ejection udder massage, nursing interval, proportion of non-nutritive nursings, and

proportion of nursings terminated by the sow), the ANOVA model with time, breed,

time�breed interaction and the sow identity as main effects was not signi®cant for the

nursing interval and the proportion of non-nutritive nursings. The proportion of nursings

terminated by the sow was lower (ANOVA, F1, 12�10.67, p<0.01) and post-ejection udder

massage was longer (ANOVA, F1, 12�13.49, p<0.01) in the night than during daytime

(Table 2). Crossbred sows terminated more nursings (ANOVA, F1, 12�86.37, p<0.0001),

and allowed shorter post-ejection massage (ANOVA, F1, 12�7.97, p<0.05). The interaction

between time of day and breed was not signi®cant for any variable. Sow identity in¯uenced

the proportion of sow-terminated nursings (ANOVA, F12, 12�4.55, p<0.01) and the

Fig. 1. Saliva cortisol concentrations of domestic (solid line) and crossbred (hatched line) sows at 5 months of

age during a challenge test. See text for details. Means�S.E. are depicted.

106 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

duration of pre-ejection massage (ANOVA, F12, 12�3.02, p<0.05) but not the nursing

interval, percentage of non-nutritive nursings or the post-ejection massaging duration.

3.3. Piglet losses and maternal traits

Piglet mortality within a given litter between Days 0 and 4 tended to be correlated

negatively with the number of lying-down acts during the 6 h night time recording

(rS�ÿ0.51, p<0.1). Mortality did not correlate signi®cantly with the carefulness of the

sow's lying-down behaviour (rS�ÿ0.41, p�0.15) or with any other of the recorded

maternal behaviours or reaction scores (all rS between ±0.30 and 0.30, p>0.20). Neither

was mortality related to any of the Factors 1±3 (rS�±0.23, 0.36, and ±0.07, respectively).

3.4. Post-hoc correlational and factor analysis of maternal traits

In the correlational matrix of the 11 quantitative behaviour and cortisol variables, there

were 11 signi®cant and three marginally signi®cant correlations, which formed three

interconnected groups of behaviours (Fig. 2). The factor analysis based on the correlation

matrix revealed that ®rst the three factors had eigenvalues of 3.19, 2.98 and 2.06,

explaining 32, 30 and 20% (82% in total) of the variation in the matrix. The fourth factor

had an eigenvalue of only 0.64 and hence, the ®rst three were extracted and subjected to

varimax orthogonal and then promax oblique rotation (Fig. 3). After promax rotation, the

Table 1

Maternal reactivity, posture changing behaviour and nursing behaviour in domestic and crossbred sowsa

Variable Breed U value Significance

Domestic Crossbred

Nest leaving on Day 2 (number of

sows with latency <300 s)

Three out

of seven

Four out

of seven

± NS

Reaction to trapped piglet vocalisations

on Day 2 (arbitrary score)

4.9�1.4 5.9�1.4 22 NS

Reaction to three types of piglet call

on Day 5 (arbitrary score)

27.6�7.2 31.9�7.6 19 NS

Reaction to a human in the piglet

nest on Day 8 (arbitrary score)

22.8�5.0 19.6�4.1 22 NS

Internursing interval (min) 39.7�2.5 35.4�1.6 15 NS

Non-nutritive nursings (%) 13.7�5.4 17.0�3.0 22 NS

Nursings terminated by the sow (%) 16.7�2.7 56.9�19.6 1 ***

Pre-ejection massage duration (s) 96�12 111�10 18 NS

Post-ejection massage duration (s) 428�53 268�32 6 **

Number of standing-ups during six night time hours 1.7�0.6 4.7�1.2 9 *

Carefulness of lying-down (arbitrary score) 3.5�0.3 3.1�0.1 12 NS

Piglet mortality (% of pigs dying between Days 0 and 4) 9.9�2.9 12.9�3.8 20 NS

a With the exception of the variable `nest leaving' on Day 2, means�S.E. are given. Significances as

resulting from the Mann±Whitney U-tests (Fisher exact probability test in the case of nest leaving): NS; P>0.1;

*P< 0.1; **P<0.05; ***P<0.001.

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 107

Table 2

The effects of diurnal period, breed and sow identity on nursing behavioura

Variable Time of day Breed Interaction between time

of day and breed

Sow identify

Day Night Significance Domestic Crossbred Significance Significance Significance

Internursing interval (min) 36.5�1.8 37.6�1.5 NS 37.4�1.7 36.7�1.7 NS NS NS

Non-nutritive nursings (%) 11�3 15�3 NS 12�2 14�3 NS NS NS

Nursings terminated by the sow (%) 51�9 37�7 ** 23�4 65�7 *** NS **

Pre-ejection massage duration (s) 102�11 104�8 NS 95�9 110�10 NS NS *

Post-ejection massage duration (s) 202�22 348�37 ** 331�39 219�27 ** NS NS

a Means�S.E. and significances as resulting from the F-tests within are given: NS; P>0.1; *P<0.05; **P<0.01; ***P<0.001.

10

8M

.S Ïp

inka

eta

l./Ap

plied

An

ima

lB

eha

viour

Scien

ce70

(2000)

99±114

Fig. 2. Clusters of correlated variables from the correlational matrix of 11 behavioural and cortisol variables.

Pearson correlation values are given. Dotted arrows: p<0.1; hatched arrows: p<0.05; solid arrows: p<0.01.

Variable labels: LY, carefulness of lying-down; AM, allow massage, i.e. proportion of nursings which the sows

did not terminate; NI, night inactivity, i.e. negatively taken number of body position changes during 6 h; CC,

cortisol concentration during challenge; RH, reaction to human near piglets; RT, reaction to trapped piglet

screaming on Day 2; RC, reaction of various distress calls by the piglets on Day 6; ES, ejection speed, i.e. the

briefness of pre-ejection massage; NF, nursing frequency; CB, cortisol baseline; %N, proportion of nutritive

nursings.

Fig. 3. Results of the factor analysis: loadings of the 11 behaviour and cortisol variables on the three factors

with eigenvalues >1. Variable labels: %N, proportion of nutritive nursings; AM, allow massage, i.e. proportion

of nursings which the sows did not terminate; CB, cortisol baseline; CC, cortisol concentration during challenge;

ES, ejection speed, i.e. the briefness of pre-ejection massage; LY, carefulness of lying-down; NF, nursing

frequency; NI, night inactivity, i.e. negatively taken number of body position changes during 6 h; RH, reaction

to human near piglets; RT, reaction to trapped piglet screaming on Day 2; RC, reaction of various distress calls

by the piglets on Day 6.

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 109

three factors remained almost orthogonal (uncorrelated) with the correlations between the

factors being at ÿ0.03, 0.01 and ÿ0.16. Three groups of behaviour variables were

suggested by the factor pattern. High propensity to allow post-ejection udder massage,

carefulness of lying-down behaviour and night time inactivity loaded highly on Factor 1.

Reaction scores in the three stimuli tests (to trapped piglet vocalisation on Day 2, to piglet

distress vocalisations on Day 5 or 6, and to human intrusion into the `nest' on Day 8 or 9)

and the speed of milk release loaded highly on Factor 2. Finally, nursing frequency was

strongly positively, and the proportion of nutritive nursings strongly negatively, associated

with Factor 3. Cortisol concentration under challenge loaded highly negatively on Factor 1,

and cortisol baseline negatively on Factor 3. Factor 2 was not associated with cortisol

concentrations at all.

4. Discussion

In this study, we documented striking similarities in maternal behaviour of domestic and

wild boar�domestic pigs. This study complements a previous paper by Gustafsson et al.

(1999) which was based on parallel observations on the same animals and which focused

on prefarrowing nest building behaviour and on daytime nursing and general proximity

behaviours. Since all 14 experimental sows were born to and raised by mothers of the same

breed and kept until testing in the same type of environment, we have controlled for other

than genetic factors. The studies were performed on the assumption that behavioural traits,

especially the quantitative ones like frequency and duration of behaviours, are in¯uenced

by many genes. For such traits, the behavioural phenotype of a hybrid should be about

halfway between that of the two parental genotypes (Price, 1998). A lack of difference

between one of the parental genotypes and the hybrid suggests that the other parent's

genotype is not differing grossly from the hybrid either. This is exactly what was found in

the two studies. The results thus might suggest that domestic sows retain the maternal

behaviour which has evolved under natural selection. This conclusion is in agreement with

previous observations of semi-naturally kept domestic sows (Jensen, 1986, 1988) and

female wild boar (Horrell, 1997). However, our conclusion must be regarded as pre-

liminary for two reasons. First, the sample size was small. Second, all seven crossbred sows

were sired by the same wild boar. The sows within this group were, thus, half-siblings

which may somewhat weaken the genetic comparison.

The present study and that of Gustafsson et al. (1999) found only two behavioural

differences between domestic and crossbred sows. First, the domestic sows had a lower

tendency to terminate nursings early after milk ejection. This may be interpreted as an

enhanced tendency in domestic mothers to invest more into the current litter as opposed to

saving resources for future reproduction (Jensen and Gustafsson, 1997; Gustafsson et al.,

1999). This interpretation is based on the assumption that longer post-ejection massage

increases milk production. Support for such a relationship was found in some studies

(Algers and Jensen, 1991; Jensen et al., 1998) but not in others (SÏpinka and Algers, 1995;

Illmann et al., 1998). An alternative explanation arises from the second behavioural

difference found in the two studies: the domestic sows were generally less active as

manifested in a higher proportion of lying during the day at 3 weeks postpartum

110 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

(Gustafsson et al., 1999) and in less frequent interruptions of the night time rest through

standing-up acts (this study). Lower activity saves energy and might, thus, be a result of the

intense selection for fast growth, ef®cient food conversion, heavier live weight and larger

body size (Robert et al., 1987). As a side effect, the less active domestic sows may remain

in the nursing position longer after milk ejection. A third interpretation is that higher agility

of crossbred sows is due to hybrid vigour, since, presumably, a higher heterozygosity was

achieved in the wild boar�Landrace hybrids than in the Yorkshire�Landrace crosses.

We found a close similarity in the duration of pre-ejection udder massage, nursing

interval, and the proportion of non-nutritive nursings observed by a human observer during

the day and the same variables scored from the night time video recordings. This

demonstrates, ®rst, that these aspects of nursing behaviour were not in¯uenced by daytime

human activity, and second, that the nursing effort was evenly distributed over the diurnal

period. Equal nursing frequency over day and night was also found in domestic sows by

Hartman et al. (1962). During the night, sows terminated fewer nursings and, consequently,

post-ejection udder massage lasted longer than during the day. This was probably

connected to the fact that sows rested in a recumbent posture for most of the night, as

reported previously by Bùe (1991, 1994).

In the ANOVA analysis of the pooled day-and-night nursing data, the sows' identity as a

factor signi®cantly in¯uenced the proportion of nursings terminated by the sow and the pre-

ejection massage duration (Table 2). This documents that these nursing characteristics

were retained in their nursing behaviour throughout the diurnal period. On the other hand,

the internursing interval and the proportion of non-nutritive nursings were not signi®cantly

affected by sow identity, showing that the frequency of nursing and the related frequency of

failures to release milk were not stable over the diurnal period.

Several previous studies found piglet mortality to be in¯uenced by sow reactivity to

piglet distress calls, and to carefulness and frequency of postural changes (McGlone and

Blecha, 1987; Hutson et al., 1993; Wechsler and Hegglin, 1997). In our study, the

frequency of postural changes was moderately related to piglet mortality, but carefulness

of lying-down and the reactivity to vocal stimuli was not. This failure to con®rm the

previously reported relationships may be due to the fact that we followed the mortality only

starting with the moment of cross-fostering, about 12±36 h after the birth of the piglets.

Since more than a half of the live-born piglet mortality occurs during the parturition and

within 24 h after birth (Weary et al., 1996a), we probably missed the period when the

maternal behaviour is most in¯uential on piglets' survival.

The post-hoc correlational analysis indicated three groups of interdependent variables

(Fig. 2). Subsequent factor analysis, although based on low ratio of replicates (14) to

original variables (11), succeeded in visualising this structure by suggesting three factors

which explained more than 80% of the variation and on which the variables separated out

well (Fig. 3). Factor 1 might be tentatively termed `calmness' as it was associated with the

sow remaining in a nursing position for a long time after milk ejection, with long periods of

uninterrupted resting with the piglets in the night, and with careful lying-down behaviours

which preclude trapping piglets under the sow's body. Factor 2 could be perhaps best called

`protectiveness' as it is connected with quick and intense reactions of the sows to stimuli

which signal danger to the piglets. The association of this factor with short duration of pre-

ejection massage might indicate that it represents a general speed and/or intensity of the

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 111

sows' reactions to any piglet-related signals including tactile ones. Factor 3 could be

labelled `nursing activity' as it is associated with high nursing frequency which is, in turn,

known to be connected with high proportion of non-nutritive nursings (this study; Fraser,

1975; CastreÂn et al., 1989; SÏpinka et al., 1997).

The three factors remained almost orthogonal after the oblique promax rotation,

demonstrating that they are independent of each other. This excludes the possibility that

differences in pig maternal style scale along just one dimension of `emotionality' or

`temperament' in the narrow sense (Wilson et al., 1994; Clarke and Boinski, 1995; Gosling

and John, 1999). An important area of future research is how sows' maternal behaviour

characteristics relate to their emotionality in other biologically signi®cant situations, such

as food competition, meeting strangers, novel environment, or behaviour towards humans

(Lawrence et al., 1991; Forkman et al., 1995; Erhard and Mendl, 1999; Erhard et al., 1999;

Thodberg et al., 1999).

The factor analysis also indicated that corticosteroid pro®les of individual sows were

related to their maternal behaviour. Speci®cally, cortisol concentrations under challenge

loaded strongly negatively on Factor 1. This may suggest a common factor underlying

`calm care' behaviour towards the piglets and `calm' physiological reaction to a challenge

(low adrenal cortex reactivity). Factor 3, describing `nursing activity', was negatively

related to cortisol baseline values. However, high or low cortisol pro®les at 5 months were

not predictive for protective reactions to piglets in danger as revealed in Factor 2. The

cortisol data were collected in juvenile animals long before the maternal behaviour was

expressed. However, it is known that cortisol pro®les remain individually stable during

ontogeny for periods of months in pigs (von Borell and Ladewig, 1992; Ladewig et al.,

1993). We, therefore, assume that the correlations between cortisol concentrations and later

maternal behaviour re¯ect a real relationship. To our knowledge, there are no other data on

the relationship between sows' pituitary-adrenocortical (PAC) pro®les and their styles of

maternal behaviour. It would be important to know whether exogenous ACTH or cortisol

will alter maternal behaviour or vice versa, whether maternal experience has an effect on

PAC functioning.

5. Conclusions

Two main conclusions follow from this study.

First, we found no differences in most aspects of maternal behaviour between domestic

and crossbred domestic�wild boar sows. This might indicate that maternal behaviour has

not changed signi®cantly during domestication and/or under modern breeding pro-

grammes.

Second, our post-hoc analysis suggests that pig maternal behaviour may be affected by

three underlying behaviour characteristics: `calmness', scaling from slow and careful

behaviour in the presence of piglets to quick and unrestful; `protectiveness', contrasting

intense and quick versus weak and slow responding to stimuli from the piglets, especially

those signalling distress or danger; and `nursing activity', in which the frequency of nursing

plays a prominent role. These behaviour characteristics or temperament traits may be

correlated with individual differences in pituitary-adrenocortical functioning.

112 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114

Acknowledgements

This study was funded by grants No. 523/98/0789 and 523/99/0985 from the grant

Agency of the Czech Republic and grant No. M02/99/03 from the Czech Ministry of

Agriculture. Suzanne Held kindly helped with the English in two versions of the text.

References

Algers, B., Jensen, P., 1991. Teat stimulation and milk production during early lactation in sows Ð effects of

continuous noise. Can. J. Anim. Sci. 71, 51±60.

Blackshaw, J.K., Blackshaw, A.W., Thomas, F.J., Newman, F.W., 1994. Comparison of behaviour patterns of

sows and litters in a farrowing crate and a farrowing pen. Appl. Anim. Behav. Sci. 39, 281±295.

Bùe, K., 1991. The process of weaning in pigs Ð when the sow decides. Appl. Anim. Behav. Sci. 30, 47±59.

Bùe, K., 1994. Variation in maternal behaviour and production of sows in integrated loose housing systems in

Norway. Appl. Anim. Behav. Sci. 41, 53±62.

CastreÂn, H., Algers, B., Jensen, P., 1989. Occurrence of unsuccessful sucklings in newborn piglets in a semi-

natural environment. Appl. Anim. Behav. Sci. 23, 61±73.

Clarke, A.S., Boinski, S., 1995. Temperament in nonhuman-primates. Am. J. Primatol. 37, 103±125.

Dyck, G.W., Swiestra, E.E., 1987. Causes of piglet death from birth to weaning. Can. J. Anim. Sci. 67, 543±547.

Ekkel, E.D., Dieleman, S.J., Schouten, W.G.P., Portela, A., Cornelissen, G., Tielen, M.J.M., Halberg, F., 1996.

The circadian rhythm of cortisol in the saliva of young pigs. Physiol. Behav. 60, 985±989.

Erhard, H.W., Mendl, M., 1999. Tonic immobility and emergence time in pigs Ð more evidence for behavioural

strategies. Appl. Anim. Behav. Sci. 61, 227±237.

Erhard, H.W., Mendl, M., Christiansen, S.B., 1999. Individual differences in tonic immobility may reflect

behavioural strategies. Appl. Anim. Behav. Sci. 64, 31±46.

Forkman, B., Furuhaug, I.L., Jensen, P., 1995. Personality, coping patterns and aggression in piglets. Appl.

Anim. Behav. Sci. 45, 31±42.

Fraser, D., 1975. The nursing and suckling behaviour of pigs. III. Behaviour when milk ejection is elicited by

manual stimulation of the udder. Brit. Vet. J. 131, 416±426.

Gosling, S.D., John, O.P., 1999. Personality dimensions in nonhuman animals: a cross-species review. Psychol.

Sci. 8, 69±75.

Gustafsson, M., Jensen, P., de Jonge, F.H., Illmann, G., SÏpinka, M., 1999. Maternal behaviour of domestic sows

and crosses between domestic sows and wild boar. Appl. Anim. Behav. Sci. 65, 29±42.

Hartman, D.A., Ludwick, T.M., Wilson, R.F., 1962. Certain aspects of lactation performance in sows. J. Anim.

Sci. 21, 883±886.

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. Appl. Anim. Behav. Sci. 58, 241±254.

Horrell, I., 1997. The characterisation of suckling in wild boar. Appl. Anim. Behav. Sci. 53, 271±277.

Hutson, G.D., Dickenson, L.G., Wilkinson, J.L., Luxford, B.G., 1993. The response of sows to novel visual,

olfactory, auditory and tactile stimuli. Appl. Anim. Behav. Sci. 35, 255±266.

Illmann, G., Madlafousek, J., 1995. Occurrence and characteristics of unsuccessful nursings in minipigs during

the first week of life. Appl. Anim. Behav. Sci. 44, 9±18.

Illmann, G., SÏpinka, M., SÏ teÏtkovaÂ, Z., 1998. Influence of massage during simulated non-nutritive nursings on

piglets' milk intake and weight gain. Appl. Anim. Behav. Sci. 55, 279±289.

Illmann, G., SÏpinka, M., SÏ teÏtkovaÂ, Z., 1999. Predictability of nursings without milk ejection in domestic pigs.

Appl. Anim. Behav. Sci. 61, 303±311.

Janssens, C.J.J.G., Helmond, F.A., Wiegant, V.M., 1994. Increased cortisol response to exogenous

adrenocorticotropic hormone in chronically stressed pigs: influence of housing conditions. J. Anim. Sci.

72, 1771±1777.

Jensen, P., 1986. Observations on the maternal behaviour of free-ranging domestic pigs. Appl. Anim. Behav. Sci.

16, 131±142.

M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114 113

Jensen, P., 1988. Maternal behaviour and mother-young interactions during lactation in free-ranging domestic

pigs. Appl. Anim. Behav. Sci. 20, 297±308.

Jensen, P., Gustafsson, M., 1997. Towards a functional view on domestication. In: Proceedings of the 31st ISAE

Congress, Prague, 13±16 August 1997, Polygrafia SAV, Praha, pp. 43±44.

Jensen, P., Gustafsson, M., Augustsson, H., 1998. Teat massage after milk ingestion in domestic piglets: an

example of honest begging? Anim. Behav. 55, 779±786.

Ladewig, J., de PassilleÂ, A.M., Rushen, J., Schouten, W., Terlouw, E.M.C., von Borell, E., 1993. Stress and the

physiological correlates of stereotypic behaviour. In: Lawrence, A.B., Rushen, J. (Eds.), Stereotypic Animal

Behaviour: Fundamentals and Applications to Welfare, CAB International, Wallingford, pp. 97±118.

Lawrence, A.B., Terlouw, E.M.C., Illius, A.W., 1991. Individual differences in behavioural responses of pigs

exposed to non-social and social challenges. Appl. Anim. Behav. Sci. 30, 73±86.

McGlone, J.J., Blecha, F., 1987. An examination of behavioral, immunological and productive traits in four

management systems for sows and piglets. Appl. Anim. Behav. Sci. 18, 269±289.

McKay, R.M., 1993. Preweaning losses of piglets as a result of index selection for reduced backfat thickness and

increased growth rate. Can. J. Anim. Sci. 73, 437±442.

Price, E.O., 1998. Behavioral genetics and the process of domestication. In: Grandin, T. (Ed.), Genetics and the

Behavior of Domestic Animals, Academic Press, New York, pp. 31±65.

Robert, S., Dancosse, J., Sallaire, A., 1987. Some observations on the role of environment and genetics in

behaviour of wild and domestics forms of Sus scrofa (European wild boars and domestic pigs). Appl. Anim.

Behav. Sci. 17, 253±262.

Ruis, M.A.W., Brake, J.H.A.T., Engel, B., Ekkel, E.D., Buist, W.G., Blokhuis, H.J., Koolhaas, J.M., 1997. The

circadian-rhythm of salivary cortisol in growing pigs Ð effects of age, gender, and stress. Physiol. Behav. 62,

623±630.

SÏpinka, M., Algers, B., 1995. Functional view on udder massage after milk let-down in pigs. Appl. Anim. Behav.

Sci. 43, 197±212.

SÏpinka, M., Illmann, G., Algers, B., SÏ teÏtkovaÂ, Z., 1997. The role of nursing frequency in milk production in

domestic pigs. J. Anim. Sci. 75, 1223±1228.

SÏpinka, M., Illmann, G., SÏ teÏtkovaÂ, Z., KrejcÏõÂ, P., TomaÂnek, M., SedlaÂk, L., LidickyÂ, J., 1999. Prolactin and

insulin levels in lactating sows in relation to nursing frequency. Domestic Anim. Endocrinol. 17, 53±64.

Thodberg, K., Jensen, K.H., Herskin, M.S., 1999. A general reaction pattern across situations in prepubertal

gilts. Appl. Anim. Behav. Sci. 63, 103±119.

von Borell, E., Ladewig, J., 1992. Relationship between behaviour and adrenocortical response pattern in

domestic pigs. Appl. Anim. Behav. Sci. 34, 195±206.

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. Appl. Anim. Behav. Sci. 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? Anim. Behav. 51, 619±624.

Wechsler, B., Hegglin, D., 1997. Individual differences in the behaviour of sows at the nest-site and the crushing

of piglets. Appl. Anim. Behav. Sci. 51, 39±49.

Wilson, D.S., Clark, A.B., Coleman, K., Dearstyne, T., 1994. Shyness and boldness in humans and other

animals. Trends Ecol. Evol. 9, 442±446.

114 M. SÏpinka et al. / Applied Animal Behaviour Science 70 (2000) 99±114